[go: up one dir, main page]
More Web Proxy on the site http://driver.im/

CN112883648A - Training method and device for automobile fuel consumption prediction model and computer equipment - Google Patents

Training method and device for automobile fuel consumption prediction model and computer equipment Download PDF

Info

Publication number
CN112883648A
CN112883648A CN202110210292.1A CN202110210292A CN112883648A CN 112883648 A CN112883648 A CN 112883648A CN 202110210292 A CN202110210292 A CN 202110210292A CN 112883648 A CN112883648 A CN 112883648A
Authority
CN
China
Prior art keywords
automobile
fuel
oil consumption
target
parameter
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110210292.1A
Other languages
Chinese (zh)
Other versions
CN112883648B (en
Inventor
曾泽泉
吴启昌
白志刚
袁晶
郭亚昌
徐忠宇
李凌
李鹏
岳涛
杨佳亮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
FAW Jiefang Automotive Co Ltd
Original Assignee
FAW Jiefang Automotive Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by FAW Jiefang Automotive Co Ltd filed Critical FAW Jiefang Automotive Co Ltd
Priority to CN202110210292.1A priority Critical patent/CN112883648B/en
Publication of CN112883648A publication Critical patent/CN112883648A/en
Application granted granted Critical
Publication of CN112883648B publication Critical patent/CN112883648B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Human Resources & Organizations (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • General Health & Medical Sciences (AREA)
  • Tourism & Hospitality (AREA)
  • Development Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Game Theory and Decision Science (AREA)
  • General Business, Economics & Management (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Combined Controls Of Internal Combustion Engines (AREA)

Abstract

The application relates to a training method and device for an automobile fuel consumption prediction model, computer equipment and a storage medium. The method comprises the following steps: acquiring driving parameters, oil consumption parameters and automobile parameters of a plurality of fuel automobiles corresponding to the category of a target automobile; determining a first oil consumption value of a standard route of each fuel automobile according to the running parameters of each fuel automobile; screening out a second oil consumption value of the standard route from the first oil consumption values of the plurality of standard routes according to a first normal distribution state of the first oil consumption values of the plurality of standard routes; determining a target fuel automobile corresponding to the second fuel consumption value of each standard route; obtaining a corresponding oil consumption parameter distribution class diagram based on target automobile parameters of each target fuel automobile, wherein each oil consumption parameter distribution class diagram and a corresponding second oil consumption value of the standard route form a training sample; and performing model training on the automobile oil consumption prediction model based on all training samples. By adopting the method, the fuel consumption of the fuel automobile to be predicted can be accurately predicted.

Description

Training method and device for automobile fuel consumption prediction model and computer equipment
Technical Field
The application relates to the technical field of automobile oil consumption prediction, in particular to a training method and device of an automobile oil consumption prediction model, computer equipment and a storage medium.
Background
Commercial vehicles are important pillars and supporting industries for national economic development, more than 400 thousands of commercial vehicles are produced and sold in 2019 years, with the development of commercial vehicle markets, the consumption speed of fossil resources caused by oil consumption is increased, and meanwhile, the emission of pollutants such as CO, NOx, HC and the like is increased. A great number of policies are also issued in succession by the country to promote energy conservation and emission reduction (such as improving fuel cleanliness, reducing fuel consumption rate, reducing pollutant discharge and the like), and various new energy vehicle types (hybrid power and electric vehicles) and new oil-saving technologies are also successively released by all vehicle manufacturers. From the aspect of vehicle type development and use, the oil consumption of each type of vehicle is greatly different due to different requirements of transportation environment, driving users, goods loading and local regulations.
In order to meet the requirements of different domestic transportation markets, whole car manufacturers often develop multiple types of cars (power matching optimization, fuel economy, etc.). In order to realize the monitoring of the oil consumption, part of finished automobile manufacturers transmit the acquired instantaneous data of the engine ECU back to a remote server through a T-Box, and the oil consumption condition is monitored through a basic statistical method (summation and average). However, in this way, the oil consumption data monitoring result lacks deep analysis and application, and the use oil consumption of the new vehicle cannot be effectively and accurately predicted.
Disclosure of Invention
Therefore, in order to solve the technical problems, it is necessary to provide a training method, an apparatus, a computer device and a storage medium for a fuel consumption prediction model of an automobile, which can accurately predict the fuel consumption of a fuel automobile to be predicted.
A training method of a fuel consumption prediction model of an automobile comprises the following steps:
acquiring automobile driving state parameters of a plurality of fuel automobiles corresponding to the target automobile category, wherein the automobile driving state parameters comprise driving parameters, oil consumption parameters and automobile parameters;
determining a first oil consumption value of a standard route corresponding to each fuel automobile according to the driving parameters of each fuel automobile;
according to a first normal distribution state of standard route first oil consumption values corresponding to a plurality of fuel automobiles respectively, screening a standard route second oil consumption value corresponding to the target automobile type from the standard route first oil consumption values based on the first normal distribution state;
determining a target fuel automobile corresponding to the second fuel consumption value of each standard journey, and acquiring a target automobile driving state parameter corresponding to the target fuel automobile; the target automobile running state parameters comprise target running parameters, target oil consumption parameters and target automobile parameters;
counting the target fuel consumption parameters based on the parameter intervals corresponding to the target vehicle parameters of each target fuel vehicle respectively to obtain a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle;
converting each sample oil consumption parameter distribution table into a corresponding oil consumption parameter distribution class diagram respectively, and forming the oil consumption parameter distribution class diagram corresponding to each target fuel automobile and a second oil consumption value of a standard route into one of training samples of the target automobile category;
and for all training samples in the target automobile category, taking the oil consumption parameter distribution class diagram in each training sample as the input of the automobile oil consumption prediction model to be trained, taking the corresponding second oil consumption value of the standard distance as the training label of the automobile oil consumption prediction model to be trained, and carrying out model training on the automobile oil consumption prediction model corresponding to the target automobile category.
In one embodiment, the determining, according to the driving parameters of each fuel automobile, a first fuel consumption value of a standard route corresponding to each fuel automobile includes:
for each fuel oil automobile, dividing the corresponding route information into a plurality of single routes according to the load information; wherein, the load information is changed once, namely, the load information is marked as the starting point and/or the end point of a single journey;
determining an oil consumption parameter corresponding to each single journey, and calculating a standard journey initial oil consumption value of each single journey according to the plurality of single journeys and the corresponding oil consumption parameters;
and determining the first oil consumption value of the standard route corresponding to each fuel automobile according to the second normal distribution state of the initial oil consumption value of the standard route corresponding to each fuel automobile.
In one embodiment, the second normal distribution state includes a second normal mean and a second standard deviation; the step of determining the first fuel consumption value of the standard route corresponding to each fuel automobile according to the second normal distribution state of the initial fuel consumption value of the standard route corresponding to each fuel automobile comprises the following steps:
when the first oil consumption value of the standard route is the average oil consumption value of the standard route, the first oil consumption value of the standard route is a second normal mean value;
and when the first oil consumption value of the standard route is the optimal oil consumption value of the standard route, the first oil consumption value of the standard route is the difference value between the second normal mean value and the second standard deviation.
In one embodiment, the first normal distribution state includes a first normal mean and a first standard deviation, the first normal distribution state according to the standard routes first fuel consumption values respectively corresponding to the plurality of fuel automobiles screens out a standard route second fuel consumption value corresponding to the target automobile category from the plurality of standard route first fuel consumption values based on the first normal distribution state, including,
when the standard route first oil consumption value is the standard route average oil consumption value, screening out the standard route first oil consumption value with the deviation not larger than the first standard deviation from the plurality of standard route first oil consumption values as a standard route second oil consumption value;
and when the first oil consumption value of the standard route is the optimal oil consumption value of the standard route, screening the first oil consumption value of the standard route which is not more than the first normal mean value from the first oil consumption values of the standard route to be used as a second oil consumption value of the standard route.
In one embodiment, the vehicle parameters include an engine rotation speed parameter and an engine torque parameter, and the obtaining a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle by performing statistics on the target fuel consumption parameters based on parameter intervals corresponding to the target vehicle parameters of each target fuel vehicle includes:
determining at least one first parameter interval corresponding to the engine speed parameter and at least one second parameter interval corresponding to the engine torque parameter;
acquiring an initial parameter distribution table, wherein the initial parameter distribution table comprises at least one parameter distribution domain, and each parameter distribution domain corresponds to a first parameter interval and a second parameter interval;
and counting the target fuel consumption parameters corresponding to each target fuel vehicle based on the obtained initial parameter distribution table, and obtaining a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle.
In one embodiment, the converting each sample fuel consumption parameter distribution table into a corresponding fuel consumption parameter distribution class diagram includes:
respectively carrying out normalization processing on each sample oil consumption parameter distribution table to obtain a plurality of initial fuel oil automobile oil consumption distribution class diagrams;
and determining training pixels of the automobile oil consumption prediction model, and amplifying each initial fuel oil automobile oil consumption distribution class diagram according to the training pixels to obtain a plurality of fuel oil automobile oil consumption distribution class diagrams.
In one embodiment, the method further comprises:
determining a fuel automobile to be predicted belonging to the target automobile category, and acquiring a trained target automobile fuel consumption prediction model;
acquiring a to-be-processed fuel consumption parameter distribution table corresponding to the to-be-predicted vehicle;
obtaining a fuel consumption distribution class diagram of the fuel vehicle to be processed corresponding to the vehicle to be predicted according to the fuel consumption parameter distribution table to be processed;
and inputting the fuel automobile fuel consumption distribution diagram to be processed into a trained target automobile fuel consumption prediction model to obtain a prediction standard journey fuel consumption value of the current vehicle to be predicted.
A training device for a fuel consumption prediction model of an automobile, the device comprising:
the system comprises a parameter acquisition module, a parameter processing module and a parameter processing module, wherein the parameter acquisition module is used for acquiring automobile driving state parameters of a plurality of fuel automobiles corresponding to a target automobile type, and the automobile driving state parameters comprise driving parameters, oil consumption parameters and automobile parameters;
the first calculation module is used for determining a first fuel consumption value of a standard route corresponding to each fuel automobile according to the running parameters of each fuel automobile;
the second calculation module is used for screening out a standard route second oil consumption value corresponding to the target automobile type from the standard route first oil consumption values based on the first normal distribution state according to the first normal distribution state of the standard route first oil consumption values corresponding to the plurality of fuel automobiles respectively;
the screening module is used for determining a target fuel automobile corresponding to the second fuel consumption value of each standard journey and acquiring a target automobile driving state parameter corresponding to the target fuel automobile; the target automobile running state parameters comprise target running parameters, target oil consumption parameters and target automobile parameters;
the statistical module is used for carrying out statistics on the target fuel consumption parameters based on the parameter intervals corresponding to the target automobile parameters of each target fuel automobile respectively to obtain a sample fuel consumption parameter distribution table corresponding to each target fuel automobile;
the conversion module is used for converting each sample oil consumption parameter distribution table into a corresponding oil consumption parameter distribution class diagram respectively, and forming the oil consumption parameter distribution class diagram corresponding to each target fuel automobile and a second oil consumption value of a standard route into one of the training samples of the target automobile class;
and the training module is used for performing model training on the automobile oil consumption prediction model corresponding to the target automobile category by taking the oil consumption parameter distribution class diagram in each training sample as the input of the automobile oil consumption prediction model to be trained and taking the corresponding second oil consumption value of the standard distance as a training label of the automobile oil consumption prediction model to be trained for all the training samples in the target automobile category.
A computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
acquiring automobile driving state parameters of a plurality of fuel automobiles corresponding to the target automobile category, wherein the automobile driving state parameters comprise driving parameters, oil consumption parameters and automobile parameters;
determining a first oil consumption value of a standard route corresponding to each fuel automobile according to the driving parameters of each fuel automobile;
according to a first normal distribution state of standard route first oil consumption values corresponding to a plurality of fuel automobiles respectively, screening a standard route second oil consumption value corresponding to the target automobile type from the standard route first oil consumption values based on the first normal distribution state;
determining a target fuel automobile corresponding to the second fuel consumption value of each standard journey, and acquiring a target automobile driving state parameter corresponding to the target fuel automobile; the target automobile running state parameters comprise target running parameters, target oil consumption parameters and target automobile parameters;
counting the target fuel consumption parameters based on the parameter intervals corresponding to the target vehicle parameters of each target fuel vehicle respectively to obtain a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle;
converting each sample oil consumption parameter distribution table into a corresponding oil consumption parameter distribution class diagram respectively, and forming the oil consumption parameter distribution class diagram corresponding to each target fuel automobile and a second oil consumption value of a standard route into one of training samples of the target automobile category;
and for all training samples in the target automobile category, taking the oil consumption parameter distribution class diagram in each training sample as the input of the automobile oil consumption prediction model to be trained, taking the corresponding second oil consumption value of the standard distance as the training label of the automobile oil consumption prediction model to be trained, and carrying out model training on the automobile oil consumption prediction model corresponding to the target automobile category.
A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:
acquiring automobile driving state parameters of a plurality of fuel automobiles corresponding to the target automobile category, wherein the automobile driving state parameters comprise driving parameters, oil consumption parameters and automobile parameters;
determining a first oil consumption value of a standard route corresponding to each fuel automobile according to the driving parameters of each fuel automobile;
according to a first normal distribution state of standard route first oil consumption values corresponding to a plurality of fuel automobiles respectively, screening a standard route second oil consumption value corresponding to the target automobile type from the standard route first oil consumption values based on the first normal distribution state;
determining a target fuel automobile corresponding to the second fuel consumption value of each standard journey, and acquiring a target automobile driving state parameter corresponding to the target fuel automobile; the target automobile running state parameters comprise target running parameters, target oil consumption parameters and target automobile parameters;
counting the target fuel consumption parameters based on the parameter intervals corresponding to the target vehicle parameters of each target fuel vehicle respectively to obtain a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle;
converting each sample oil consumption parameter distribution table into a corresponding oil consumption parameter distribution class diagram respectively, and forming the oil consumption parameter distribution class diagram corresponding to each target fuel automobile and a second oil consumption value of a standard route into one of training samples of the target automobile category;
and for all training samples in the target automobile category, taking the oil consumption parameter distribution class diagram in each training sample as the input of the automobile oil consumption prediction model to be trained, taking the corresponding second oil consumption value of the standard distance as the training label of the automobile oil consumption prediction model to be trained, and carrying out model training on the automobile oil consumption prediction model corresponding to the target automobile category.
The training method, the device, the computer equipment and the storage medium of the automobile fuel consumption prediction model firstly obtain automobile driving state parameters of a plurality of fuel automobiles corresponding to specific target automobile types, then firstly determine a first fuel consumption value of a standard route of each automobile according to the automobile driving state parameters of each automobile, then screen out a second fuel consumption value of the standard route corresponding to the target automobile type based on the normal distribution state of the first fuel consumption values of the plurality of standard routes corresponding to the plurality of automobiles in the target automobile types, can determine a target fuel automobile corresponding to the target automobile type through the second fuel consumption value of the standard route, and can obtain a fuel consumption sample parameter distribution table corresponding to each target fuel automobile by specifically processing the target automobile driving state parameters of the target fuel automobiles in the target automobile types, the corresponding oil consumption parameter distribution class diagram can be obtained by further processing the sample oil consumption parameter distribution table, each oil consumption parameter distribution class diagram and the corresponding second oil consumption value of the standard journey form a training sample corresponding to the target automobile category, and after all the training samples of the target automobile category are obtained, the training samples can be used for training the automobile oil consumption prediction model of the target automobile category. By the method, the required target automobile running state parameters can be effectively screened out from the automobile running state parameters of the plurality of fuel oil automobiles corresponding to the target automobile type, the parameters are subjected to statistics and graph-like processing to become objects which are easy to identify by an automobile fuel consumption prediction model, and meanwhile, the result output by the automobile fuel consumption prediction model is guided by the corresponding training labels, so that the obtained trained automobile fuel consumption prediction model can accurately predict the fuel consumption of the fuel oil automobile to be predicted.
Drawings
FIG. 1 is an application environment diagram of a training method of a fuel consumption prediction model of a vehicle according to an embodiment;
FIG. 2 is a schematic flow chart illustrating a method for training a fuel consumption prediction model of a vehicle according to an embodiment;
FIG. 3 is a schematic flowchart illustrating the training procedure of the fuel consumption prediction model of the vehicle according to an embodiment;
FIG. 4 is a block diagram of a training apparatus for a fuel consumption prediction model of a vehicle according to an embodiment;
FIG. 5 is a diagram illustrating an internal structure of a computer device according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
As the demand for fuel economy of the whole vehicle is becoming stronger and stronger, reducing fuel consumption will still be an important technical direction in the coming years. The fuel economy is improved through the new method and the new technology, and the method is one of the core capabilities of future competitiveness of all vehicle manufacturers. At present, all manufacturers still adopt the traditional whole vehicle calibration method to develop vehicle types, the oil consumption of the whole vehicle cannot be accurately predicted in the early stage of product development, and the fuel consumption of the market can only be explored through small-batch product release in the early stage of batch market investment. The training method of the automobile oil consumption prediction model in the embodiment is a commercial automobile oil consumption monitoring and prediction method based on big data and a neural network, monitoring, analysis and modeling are carried out on a large number of oil consumption samples of the automobile type sold in the past, and the established optimal oil consumption model can predict oil consumption besides knowing the oil consumption performance of the automobile type in the market in real time, so that the product development cost and the product development period are greatly improved, the use cost of market users can be reduced, and the method has great advantages in the aspect of product competitiveness. The invention aims to solve the problems that the oil consumption monitoring of the whole vehicle products in the market is realized, the regional optimal oil consumption model is established through analysis modeling, and the oil consumption of new products is predicted.
The training method of the automobile fuel consumption prediction model can be applied to the application environment shown in fig. 1. Wherein the terminal 102 may communicate with the server 104 over a network. Specifically, the terminal 102 collects and obtains vehicle driving state parameters of a plurality of fuel vehicles and sends the vehicle driving state parameters to the server 104, and the server 104 processes the received vehicle driving state parameters of the plurality of fuel vehicles and completes training of the vehicle fuel consumption prediction model according to a processing result. The terminal 102 may be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices, the manner of obtaining the vehicle driving state parameters of a plurality of fuel vehicles may be implemented by various structures inherent in the fuel vehicles, such as devices with a data acquisition function, such as a vehicle CAN bus, a GPS module, and a vehicle data recorder, and the server 104 may be implemented by an independent server or a server cluster formed by a plurality of servers, and may be used to complete data storage (database), calculation, and the like, and train a fuel consumption prediction model.
In a specific embodiment, the vehicle identification vin, the time, the vehicle speed, the mileage per unit time, the engine speed, the engine torque, the accelerator opening, the oil consumption per unit time and the engine power are determined by the vehicle identification vin on the vehicle CAN bus, the GPS module, the driving recorder and other devicesAnd the latitude, longitude, elevation, the state of an ignition switch, the gear and load signals are periodically acquired, and then the acquired data is periodically transmitted back to a remote server for data storage through the T-Box to obtain a multi-vehicle data set. The multi-vehicle data set Z is first stored in a specific vehicle database, and vehicle driving state parameters corresponding to a plurality of vehicle types and a plurality of fuel vehicles are stored in the vehicle database. When oil consumption prediction needs to be carried out on a specific vehicle type, a target vehicle type is selected according to the specific vehicle type, then vehicle running state parameters corresponding to fuel vehicles corresponding to the target vehicle type are found out from the vehicle database, the vehicle running state parameters corresponding to each fuel vehicle take a whole vehicle identifier vin as a set identifier, and the vin corresponding to each fuel vehicle is unique. For example, for a target automobile category, single-automobile data can be screened from the set Z through the whole automobile identifier vin to obtain a single-automobile data set ZvinAnd forming all fuel automobile data sets W corresponding to the target automobile type by a plurality of single automobile data sets.
Before describing the training method of the fuel consumption prediction model of the vehicle in the present application, the following explanations are first made for some terms involved in the embodiments of the present application:
the automobile category: the automobile attribution categories with similar characteristics can be distinguished through functions, the classification of the automobile categories has various standards, the classification is not specifically distinguished in the embodiment, and the automobile attribution categories can be an automobile with special purposes, such as an automobile for transporting a certain special substance; or the automobiles are distinguished by load limitation, namely the automobile with 10 tons of load limitation; or an automobile differentiated by application environments, for example, an automobile suitable for an extremely cold environment.
Vehicle driving state parameters: some specific parameters that may be involved in the driving process of the vehicle include driving parameters, external parameters of the vehicle during the driving process, such as specific load information, distance information, and the like; the fuel consumption parameters are related to fuel consumption of the automobile in the driving process, such as accelerator opening, fuel consumption per unit time, engine power and the like; vehicle parameters, which are intrinsic to the vehicle based on its own structure point, such as engine speed, engine torque, etc.; and other vehicle driving related parameters such as ignition switch state, gear, vehicle identification vin, time, vehicle speed, etc.
First fuel consumption value of standard route: after each fuel automobile is analyzed, the obtained specific fuel parameter value of the current fuel automobile in a standard distance (such as hundreds of kilometers and kilokilometers) can be the fuel consumption value of the current fuel automobile in a historical distance, which shows the best performance, or the average fuel consumption value of the current fuel automobile in the standard distance based on the whole distance.
And (3) standard distance second oil consumption value: and after a plurality of fuel automobiles are analyzed, screening out a target value from the first fuel consumption values in a plurality of standard routes.
Sample oil consumption parameter distribution table: and counting the target oil consumption parameters based on the target automobile parameters to obtain a table result.
Oil consumption parameter distribution class diagram: the oil consumption parameter distribution table similar to the image obtained by further processing the sample oil consumption parameter distribution table can be used for training an automobile oil consumption prediction model.
For the automobile industry, multiple automobile categories exist, and for each automobile category in the multiple automobile categories, the corresponding automobile oil consumption prediction model can be trained by adopting the training method of the automobile oil consumption prediction model in the application, so that the automobile oil consumption prediction model capable of correspondingly predicting each automobile category is obtained.
In an embodiment, as shown in fig. 2, a method for training a fuel consumption prediction model of a vehicle is provided, which is described by taking the method as an example applied to the server in fig. 1, and includes the following steps:
step S202, obtaining automobile driving state parameters of a plurality of fuel automobiles corresponding to the target automobile category, wherein the automobile driving state parameters comprise driving parameters, oil consumption parameters and automobile parameters.
Specifically, for a specific target vehicle type of fuel-powered vehicle, the server first obtains vehicle driving state parameters corresponding to each fuel-powered vehicle in the target vehicle type, such as driving parameters: load information and route information; oil consumption parameters: accelerator opening, oil consumption per unit time and engine power; automobile parameters: engine speed, engine torque, etc.
In a specific embodiment, the data set Z corresponding to the data CAN be obtained by periodically acquiring vehicle identification vin, time, vehicle speed, mileage per unit time, engine speed, engine torque, accelerator opening, oil consumption per unit time, engine power, latitude, longitude, elevation, ignition switch state, gear and load signals on a vehicle CAN bus, a GPS module, a vehicle data recorder and other devices. The data set Z can be stored in a server, the whole vehicle identification vin serves as a data label, and one data label corresponds to all vehicle running state parameters of one fuel vehicle.
And step S204, determining a first fuel consumption value of the standard route corresponding to each fuel automobile according to the running parameters of each fuel automobile.
Specifically, according to the driving parameters of each fuel automobile, the server can determine a first fuel consumption value of the current fuel automobile in the driving process in the standard route, wherein the first fuel consumption value of the standard route is a reference fuel consumption standard of each fuel automobile in the driving process.
In a specific embodiment, after obtaining an automobile driving state parameter corresponding to one fuel automobile according to the entire automobile identifier vin, the information of the current fuel automobile driving route is divided, that is, a plurality of routes corresponding to the current fuel automobile in the entire route information are determined, and then a first fuel consumption value of the current fuel automobile in a standard route corresponding to the entire route information is determined based on the fuel consumption parameter of each route. In a specific embodiment, the first fuel consumption value of the standard route is a fuel consumption corresponding to a one hundred kilometers standard route of a current fuel automobile, and according to an actual calculation requirement, other mileage may also be used as the standard route for calculation, which is not specifically limited in this embodiment.
And S206, screening out a standard route second oil consumption value corresponding to the target automobile type from the plurality of standard route first oil consumption values based on the first normal distribution state according to the first normal distribution state of the standard route first oil consumption values corresponding to the plurality of fuel automobiles respectively.
Specifically, after the server obtains the first fuel consumption values of the standard routes corresponding to the plurality of fuel automobiles, the fuel automobiles need to be further processed. In this embodiment, the server screens the obtained first fuel consumption values of the plurality of standard routes in a first normal distribution state corresponding to the obtained first fuel consumption values of the plurality of standard routes based on the first normal distribution state, and further screens a second fuel consumption value of the standard route corresponding to the target automobile category from the first fuel consumption value of the plurality of standard routes.
And the second fuel consumption value of the standard route obtained at the moment corresponds to a plurality of fuel automobiles with the best performance in each fuel automobile in the current target automobile category. Namely, through the second fuel consumption value of each standard journey, a plurality of fuel automobiles with the best performance in the current target automobile category can be determined.
Step S208, determining a target fuel automobile corresponding to the second fuel consumption value of each standard journey, and acquiring a target automobile driving state parameter corresponding to the target fuel automobile; the target automobile driving state parameters comprise target driving parameters, target oil consumption parameters and target automobile parameters.
Specifically, through the second fuel consumption value of each standard trip, the server can determine the corresponding target fuel-oil vehicle, and further can acquire the target vehicle running state parameters corresponding to each target fuel-oil vehicle. As mentioned above, the target vehicle driving state parameters include a target driving parameter, a target fuel consumption parameter, and a target vehicle parameter. That is to say, the second fuel consumption value of the standard route is determined by screening the first fuel consumption value of the standard route, and according to the second fuel consumption value of the standard route, a specific target fuel vehicle is found out from a large number of fuel vehicles corresponding to the target vehicle category, and the vehicle running state parameter corresponding to the specific target fuel vehicle is determined.
Step S210, counting the target fuel consumption parameters based on the parameter intervals corresponding to the target vehicle parameters of each target fuel vehicle respectively to obtain a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle.
Specifically, for the determined target fuel automobiles, the server firstly determines corresponding parameter intervals for the target automobile parameters of each target fuel automobile, and counts the target fuel consumption parameters of each parameter interval to obtain a sample fuel consumption parameter distribution table corresponding to each target fuel automobile. For example, for a specific vehicle parameter, there are 3 parameter intervals, for a first fuel consumption interval a, for a second fuel consumption interval B, the fuel consumption parameter value is B, and for a third fuel consumption interval C, the fuel consumption parameter value is C, so that a sample fuel consumption parameter distribution table as shown in table 1 below can be obtained by plotting.
TABLE 1 sample Fuel consumption parameter distribution Table
Target vehicle parameter interval Oil consumption parameter
a A
b B
c C
The sample fuel consumption parameter distribution table is only used as a specific example, and is not to be considered as a specific limitation to the present application. In fact, there may be a plurality of target vehicle parameter intervals, and a plurality of target vehicle parameters may also be selected, and the sample fuel consumption parameter distribution table obtained by statistics according to the above method is also within the scope of the present application.
Step S212, each sample oil consumption parameter distribution table is converted into a corresponding oil consumption parameter distribution class diagram, and the oil consumption parameter distribution class diagram corresponding to each target fuel automobile and the second oil consumption value of the standard journey form one of the training samples of the target automobile category.
Specifically, in order to enable the sample fuel consumption parameter distribution table corresponding to the target fuel automobile to be better identified by the automobile fuel consumption prediction model, the server further processes the sample fuel consumption parameter distribution table, that is, the sample fuel consumption parameter distribution table corresponding to the target fuel automobile is respectively and correspondingly converted into the fuel consumption parameter distribution class diagram. Specifically, in this embodiment, the table units in the sample fuel consumption parameter distribution table are selected and processed, so that the table units in the numerical meaning are converted into the table units in the image meaning, and thus the vehicle fuel consumption prediction model can better identify the table units. For each target fuel automobile, the corresponding fuel consumption parameter distribution class diagram and the second fuel consumption value of the standard journey jointly form one of the training samples of the target automobile category. For all training samples, each training sample corresponds to a target fuel car.
Step S214, regarding all training samples in the target automobile category, taking the oil consumption parameter distribution class diagram in each training sample as the input of the automobile oil consumption prediction model to be trained, taking the corresponding second oil consumption value of the standard distance as the training label of the automobile oil consumption prediction model to be trained, and performing model training on the automobile oil consumption prediction model corresponding to the target automobile category.
Specifically, for all training samples in the target automobile category, the server takes the oil consumption parameter distribution class diagram in each training sample as the input of the automobile oil consumption prediction model to be trained, the automobile oil consumption prediction model identifies the input oil consumption parameter distribution diagram and provides a corresponding predicted value, then the corresponding second oil consumption value of the standard route is used as a training label to train the automobile oil consumption prediction model corresponding to the target automobile category, and through the training, the output value of the automobile oil consumption prediction model corresponding to the target automobile category can gradually approach the second oil consumption value of the standard route corresponding to the input oil consumption parameter distribution diagram.
The training method of the automobile fuel consumption prediction model comprises the steps of firstly obtaining automobile running state parameters of a plurality of fuel automobiles corresponding to specific target automobile types, then firstly determining a first fuel consumption value of a standard route of each automobile according to the automobile running state parameters of each automobile, then screening out a second fuel consumption value of the standard route corresponding to the target automobile type based on the normal distribution state of the first fuel consumption values of the plurality of standard routes corresponding to the plurality of automobiles in the target automobile types, determining the corresponding target fuel automobiles in the target automobile types through the second fuel consumption values of the standard routes, specifically processing the target automobile running state parameters of the target fuel automobiles in the target automobile types to obtain a sample fuel consumption parameter distribution table corresponding to each target fuel automobile, further processing the sample fuel consumption parameter distribution table to obtain a corresponding fuel consumption parameter distribution class diagram, each fuel consumption parameter distribution class diagram and the corresponding second fuel consumption value of the standard journey form a training sample corresponding to the target automobile category, and after all the training samples of the target automobile category are obtained, the training samples can be used for training the automobile fuel consumption prediction model of the target automobile category. By the method, the required target automobile running state parameters can be effectively screened out from the automobile running state parameters of the plurality of fuel oil automobiles corresponding to the target automobile type, the parameters are subjected to statistics and graph-like processing to become objects which are easy to identify by an automobile fuel consumption prediction model, and meanwhile, the result output by the automobile fuel consumption prediction model is guided by the corresponding training labels, so that the obtained trained automobile fuel consumption prediction model can accurately predict the fuel consumption of the fuel oil automobile to be predicted.
In one embodiment, the driving parameters include load information and route information, and the determining the first fuel consumption value of the standard route corresponding to each fuel vehicle according to the driving parameters of each fuel vehicle includes: for each fuel oil automobile, dividing corresponding route information into a plurality of single routes according to the load information; wherein, the load information is changed once, namely, the load information is marked as the starting point and/or the end point of a single journey; determining an oil consumption parameter corresponding to each single journey, and calculating a standard journey initial oil consumption value of each single journey according to the plurality of single journeys and the corresponding oil consumption parameters; and determining the first oil consumption value of the standard route corresponding to each fuel automobile according to the second normal distribution state of the initial oil consumption value of the standard route corresponding to each fuel automobile.
Specifically, for each fuel automobile in the target automobile category, the first fuel consumption value of the standard route corresponding to each fuel automobile needs to be determined according to the fuel consumption information corresponding to the route information of each fuel automobile which is already driven. This specific process is as follows: considering one loading and transporting process as a single journey, namely one loading as the starting point of the single journey, and the corresponding unloading as the end point of the single journey, the journey information of each fuel automobile which has already run and completed may correspond to a plurality of single journeys. Because the driving load of each single journey may change, the corresponding driving states are different, the fuel consumption parameters of each single journey can be determined by independent calculation, and the accuracy of overall prediction is improved.
Further, for each single journey, respectively acquiring the corresponding oil consumption parameter in the process of completing the single journey, and then determining the initial oil consumption value of the standard journey corresponding to the fuel automobile in the single journey based on the driving distance of each single journey and the specific oil consumption parameter. At the moment, for a fuel automobile with multiple sections of single routes, multiple standard route initial fuel consumption values are corresponding to the fuel automobile, a second normal distribution state of the multiple standard route initial fuel consumption values is obtained, and the standard route first fuel consumption value corresponding to each fuel automobile is determined according to the second normal distribution state.
For example, with the aforementioned bicycle data ZvinFor example, assuming that the standard distance is one hundred kilometers, the solution process of the first fuel consumption value of the standard distance is as follows:
step a, splitting the distance information of the fuel automobile into a single-way running interval data set Z according to the load change conditionvin-wWhen the load of the fuel automobile changes once, the change can be recorded as a single automobile one-way (single journey).
B, collecting data Z of single-vehicle one-way operation intervalvin-wEach of z ink vin-wThe two-dimensional statistics can obtain the single-vehicle one-way hundred kilometers average oil consumption set Yvin-w. Specifically, single-trip information and single-fuel consumption information corresponding to each single trip are obtained first, and the single-trip information and the single-fuel consumption information are converted into single-fuel-consumption-per-hundred-kilometer values. According to the single-trip set corresponding to each vehicle, the corresponding single-trip hundred-kilometer average oil consumption set can be obtained, and then a first oil consumption value of the hundred-kilometer standard trip corresponding to each fuel oil vehicle is obtained.
C, according to the vehicle identification vin of the same type of vehicle sold in the market in a certain area, repeatedly executing the processes of the step a and the step b to obtain a hundred-kilometer average fuel consumption set Y of multiple vehiclesvin-w-stdOr obtaining the optimal oil consumption set Y of multiple vehicles in one hundred kilometersvin-w-minThe first fuel consumption value is a set formed by the first fuel consumption values of the standard routes respectively corresponding to each fuel automobile in the above embodiment.
In the embodiment, the first fuel consumption value of the standard route corresponding to each fuel automobile can be accurately obtained through the initial fuel consumption values of the standard routes corresponding to the multiple single routes of each fuel automobile based on the corresponding normal distribution state, and the accuracy of the training sample of the automobile fuel consumption prediction model is improved.
In one embodiment, the second normal distribution state includes a second normal mean and a second standard deviation; determining the first oil consumption values of the standard routes respectively corresponding to each fuel automobile according to the second normal distribution state of the initial oil consumption values of the standard routes corresponding to each fuel automobile, and the method comprises the following steps: when the first oil consumption value of the standard route is the average oil consumption value of the standard route, the first oil consumption value of the standard route is a second normal mean value; and when the first oil consumption value of the standard route is the optimal oil consumption value of the standard route, the first oil consumption value of the standard route is the difference value between the second normal mean value and the second standard deviation.
Specifically, the first fuel consumption value of the standard route may correspond to an average fuel consumption value of the standard route or an optimal fuel consumption value of the standard route according to a specific target of the predicted value. The standard route average oil consumption value is the standard route average oil consumption of each fuel automobile in the route which is completed in history and in the average driving state, and the standard route optimal oil consumption value is the standard route optimal oil consumption corresponding to the best driving state which is shown in the route which is completed in history of each fuel automobile. In this embodiment, according to the second normal mean and the second standard deviation of the second normal distribution state, the standard-distance average oil consumption is the second normal mean, and the standard-distance optimal oil consumption is a difference between the second normal mean and the second standard deviation.
The set Y of the average oil consumption per hundred kilometers per pass of the single vehiclevin-wFor example, the average fuel consumption per hundred kilometers per trip of a single vehicle is set Yvin-wAnd calculating a normal mean value and a standard deviation, and calculating a first oil consumption value of one hundred kilometers corresponding to each fuel oil automobile by using the normal mean value and the standard deviation. At this time, according to different specific calculation purposes, if the first fuel consumption value of the standard route to be calculated is the average fuel consumption value of the standard route, the normal average value is taken as the average fuel consumption y of the single vehicle per hundred kilometersvin-w-stdIf the first fuel consumption value of the standard route to be calculated is the optimal fuel consumption value of the standard route, subtracting a standard deviation from the normal mean value to be used as the optimal fuel consumption y of the single vehicle per hundred kilometersvin-w-min。
In the embodiment, the normal distribution processing is performed on the initial oil consumption value of the standard route of the single route, and the corresponding first oil consumption value of the standard route is obtained according to the normal distribution processing result, so that the result is more accurate, the error is smaller, the accuracy of the training sample of the automobile oil consumption prediction model is further improved, and the accuracy of the automobile oil consumption prediction model is further improved.
In one embodiment, the first normal distribution state comprises a first normal mean value and a first standard deviation, and the standard distance second fuel consumption value corresponding to the target automobile category is screened from the plurality of standard distance first fuel consumption values based on the first normal distribution state according to the first normal distribution state of the standard distance first fuel consumption values corresponding to the plurality of fuel automobiles respectively; and when the first oil consumption value of the standard route is the optimal oil consumption value of the standard route, screening the first oil consumption value of the standard route which is not more than the first normal mean value from the first oil consumption values of the plurality of standard routes as a second oil consumption value of the standard route.
Specifically, for a plurality of fuel-powered automobiles in the category of target automobiles, after the first fuel consumption value of the standard route corresponding to each fuel-powered automobile is determined, the fuel-powered automobiles further need to be screened, and a plurality of target fuel-powered automobiles with the best driving performance are selected from the fuel-powered automobiles. In this embodiment, the standard route second oil consumption value screening is based on a corresponding standard route first oil consumption value, and corresponds to the standard route first oil consumption value, when the standard route first oil consumption value is a standard route average oil consumption value, the standard route first oil consumption value, which is compared with a first normal mean value and has a deviation not greater than a first standard deviation, needs to be screened from a plurality of standard route first oil consumption values as the standard route second oil consumption value; when the first fuel consumption value of the standard route is the optimal fuel consumption value of the standard route, the first fuel consumption value of the standard route which is not more than the first normal mean value is screened out from the first fuel consumption values of the plurality of standard routes and is used as the second fuel consumption value of the standard route.
The average fuel consumption set Y of multiple vehicles in hundred kilometers in the previous embodimentvin-w-stdOr obtaining the optimal oil consumption set Y of multiple vehicles in one hundred kilometersvin-w-minFor example, set Y may be further pairedvin-w-stdAnd set Yvin-w-minAnd respectively calculating a normal mean value and a standard deviation. Specifically, when the first fuel consumption value of the standard route is the optimal fuel consumption value of the standard route, the optimal fuel consumption set Y of multiple vehicles in one hundred kilometers is screenedvin-w-minIs not greater than normal mean value (
Figure BDA0002948988490000141
) The vehicle one-way operation interval data form an optimal fuel consumption vehicle data set WbestWhen the first fuel consumption value of the standard route is the average fuel consumption value of the standard route, the set Y is screenedvin-w-stdThe vehicle single-pass operation interval data in the middle normal mean +/-one standard deviation interval form an average fuel consumption vehicle data set Wstd. The optimal fuel consumption vehicle data set WbestOr average fuel consumption vehicle data set WstdEach set element Z ink vin-w-bestThe target vehicle running state parameters are corresponding to the target fuel vehicles corresponding to the second fuel consumption value of the standard journey.
In the embodiment, the first fuel consumption values of the standard routes are further screened, so that the screening work of 'best from best' in the target automobile categories is realized, the second fuel consumption values of the screened standard routes correspond to a plurality of target fuel automobiles with the best driving performance, and the plurality of target fuel automobiles with the best driving performance are used as the construction basis of the training samples, so that more accurate training samples and automobile fuel consumption prediction models can be obtained.
In one embodiment, the vehicle parameters comprise an engine rotating speed parameter and an engine torque parameter, and the fuel consumption parameters are counted based on parameter intervals corresponding to target vehicle parameters of each target fuel vehicle respectively to obtain a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle; the method comprises the following steps: determining at least one first parameter interval corresponding to the engine speed parameter and at least one second parameter interval corresponding to the engine torque parameter; acquiring an initial parameter distribution table, wherein the initial parameter distribution table comprises at least one parameter distribution domain, and each parameter distribution domain corresponds to a first parameter interval and a second parameter interval; and counting the target fuel consumption parameters corresponding to each target fuel vehicle based on the obtained initial parameter distribution table, and obtaining a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle.
Specifically, in this embodiment, an engine speed parameter and an engine torque parameter of the vehicle parameters are further selected as specific vehicle parameters to perform statistics on the fuel consumption parameter. Therefore, in order to facilitate statistics, an initial parameter distribution table is set up in this embodiment, the initial parameter distribution table includes at least one parameter distribution domain, the number of the parameter distribution domains is related to the number of parameter intervals (first parameter intervals) of the engine speed parameter and the number of parameter intervals (second parameter intervals) of the engine torque parameter, and each parameter distribution domain corresponds to one first parameter interval and one second parameter interval. For example, if there are M engine speed parameter intervals and N engine torque parameter intervals, the number of parameter distribution domains is M × N, in this embodiment, the M × N parameter distribution domains are arranged correspondingly according to the engine speed parameter and the engine torque parameter intervals, so as to form an M × N matrix form, that is, an initial parameter distribution table. And counting the target fuel consumption parameters of each target fuel automobile based on the initial parameter distribution table.
The vehicle data set W with the optimal fuel consumptionbestThis embodiment will be further explained as an example. For example, the vehicle data set W for the above-mentioned optimum fuel consumptionbestSample data Z in (1)k vin-w-bestAnd respectively carrying out oil consumption parameter distribution statistics based on the rotation speed parameters and the torque parameters of the segmented engine, wherein the oil consumption parameters can be average throttle opening statistics, average oil consumption distribution statistics, average engine power and the like, and further respectively obtaining distribution table sets KD, KY and KG. The following table 2 is a specific statistical example, in which the fuel consumption parameter is the average throttle opening.
TABLE 2 sample Fuel consumption parameter distribution Table (KD)
Figure BDA0002948988490000151
Figure BDA0002948988490000161
Similarly, the average fuel consumption distribution statistics, the average engine power corresponding sample fuel consumption parameter distribution table (KY) and the sample fuel consumption parameter distribution table (KG) may be obtained continuously.
In the embodiment, the vehicle parameters are further determined, the parameter distribution domain is divided to carry out statistics on the fuel consumption parameters, so that the statistical result is more accurate, the fuel consumption parameter distribution table for manufacturing the training sample is more accurate, the fuel consumption performance of the used vehicle type in the market can be effectively analyzed, and the training effect of the vehicle fuel consumption prediction model is better.
In one embodiment, converting each sample fuel consumption parameter distribution table into a corresponding fuel consumption parameter distribution class diagram respectively includes: respectively carrying out normalization processing on each sample oil consumption parameter distribution table to obtain a plurality of initial fuel oil automobile oil consumption distribution class diagrams; and determining training pixels of the automobile oil consumption prediction model, and amplifying each initial fuel oil automobile oil consumption distribution class diagram according to the training pixels to obtain a plurality of fuel oil automobile oil consumption distribution class diagrams.
Specifically, for the sample fuel consumption parameter distribution table corresponding to each target fuel automobile, the sample fuel consumption parameter distribution table needs to be further converted into a fuel consumption parameter distribution class diagram which is convenient for an automobile fuel consumption prediction model to identify easily. In this embodiment, the sample fuel consumption parameter distribution table is normalized, and then the normalized sample fuel consumption parameter distribution table is scaled according to the training pixel accuracy of the automobile fuel consumption prediction model, so as to obtain a fuel automobile fuel consumption distribution class diagram meeting the training pixel accuracy requirement of the automobile fuel consumption prediction model.
The present embodiment will be described by taking the sample fuel consumption parameter distribution table (KD), the sample fuel consumption parameter distribution table (KY), and the sample fuel consumption parameter distribution table (KG) as examples. After the elements in the 3 distribution tables KD, KY, and KG are normalized respectively, the elements in the distribution tables are amplified by a times (a is 255 or other positive integer) respectively to correspondingly obtain 3 picture-like data sets Pkd、Pky、Pkg. In this embodiment, the sample fuel consumption parameter may be compared with the specific training pixel of the automobile fuel consumption prediction modelThe distribution table is scaled and is not specifically limited herein.
Assuming that the predicted value to be output by the automobile fuel consumption prediction model is the standard optimal fuel consumption value of the journey,
for the above-mentioned 3 classes picture data set Pkd、Pky、PkgIn this embodiment, a neural network is further utilized to respectively and correspondingly establish 3 image recognition models Mkd、Mky、MkgUsing a set of picture-like data Pkd、Pky、PkgRespectively as image recognition model Mkd、Mky、MkgCorresponding input, i.e. PkdInput Mkd、PkyInput Mky、PkgInput Mkg. At the same time, the same as the above-mentioned class picture data set Pkd、Pky、PkgCorresponding Yvin-w-minAnd as a data label, correcting the output of the automobile fuel consumption prediction model.
In the specific training process, P is firstly putkdInput MkdWill PkyInput MkyWill PkgInput MkgImage recognition model Mkd、Mky、MkgCorresponding output prediction result set Rkd、Rky、Rkg. Output result based on the image recognition model, and the image data set Pkd、Pky、PkgCorresponding data tag yvin-w-minObtaining the set X ═ Rkd，Rky，Rkg，yvin-w-min}。
Big data analysis classification model M established by using neural networkresultPredicting the optimal one hundred kilometers oil consumption of a single vehicle in one way, wherein the input is Rkd、Rky、RkgThe output is Yvin-w-min. That is, the model M is matched with a plurality of data sets X corresponding to a plurality of target fuel-powered vehiclesresultTraining to obtain a final prediction model Mresult。
In other words, the fuel consumption prediction model of the automobile comprises an image recognition model and a neural network model. For the specific training process of the automobile fuel consumption prediction model, the image recognition model is used for training a fuel consumption parameter distribution class diagram, the number of the fuel consumption parameter distribution class diagrams can be multiple, each fuel consumption parameter distribution class diagram corresponds to one fuel consumption parameter, and each fuel consumption parameter corresponds to one image recognition model. When there are a plurality of image recognition models, the neural network model is further required to process each image recognition result of each image recognition model, that is, each image recognition result is input into the neural network model, and the neural network is controlled to output a corresponding standard route second oil consumption value (or the standard route second oil consumption value is used as a data tag to correct a result correspondingly output by the neural network).
And for each training sample corresponding to the target automobile category, respectively training the automobile oil consumption prediction model according to the process.
In the embodiment, the adaptation degree between the fuel automobile fuel consumption distribution class diagram and the automobile fuel consumption prediction model can be effectively improved by means of normalization and scaling on the premise of ensuring the precision of the training sample, so that the automobile fuel consumption prediction model can well identify the information characteristics of the fuel automobile fuel consumption distribution class diagram and give accurate prediction based on the information characteristics.
In one embodiment, the method further comprises: determining a fuel automobile to be predicted belonging to the category of a target automobile, and acquiring a trained fuel consumption prediction model of the target automobile; acquiring a to-be-processed fuel consumption parameter distribution table corresponding to a to-be-predicted vehicle; obtaining a fuel consumption distribution class diagram of the fuel automobile to be processed corresponding to the vehicle to be predicted according to the fuel consumption parameter distribution table to be processed; and inputting the fuel automobile fuel consumption distribution diagram to be processed into the trained target automobile fuel consumption prediction model to obtain the predicted standard journey fuel consumption value of the current vehicle to be predicted.
Specifically, for a specific target automobile category, after the training samples corresponding to the target automobile category are determined and the training of the automobile fuel consumption prediction model is completed based on the training samples, the trained automobile fuel consumption prediction model can be used for fuel consumption prediction of the vehicle with the temporarily incomplete automobile driving state parameters. And (3) directly obtaining a corresponding fuel oil automobile fuel consumption distribution class diagram to be processed based on the existing parameter information by adopting the same processing mode as the training sample, and inputting the fuel oil automobile fuel consumption distribution class diagram to be processed into the automobile fuel consumption prediction model, so that the predicted standard distance fuel consumption value of the current vehicle to be predicted can be obtained.
Further, for a specific fuel-powered automobile to be predicted, when the fuel-powered automobile is a new automobile type in development, the target automobile fuel consumption prediction model for predicting the fuel-powered automobile may be an automobile fuel consumption prediction model corresponding to an automobile type with the closest vehicle configuration. For example, if a fuel-powered vehicle of a new vehicle type has no sufficient sample vehicle driving state parameters to realize prediction of a corresponding standard fuel consumption value of a distance, at this time, a vehicle class closest to the new vehicle type may be selected as a target vehicle class, preferably a vehicle class closest to vehicle configuration, and prediction of the standard fuel consumption value of the distance of the new vehicle type is realized according to a vehicle fuel consumption prediction model corresponding to the closest vehicle class.
For example, when the fuel consumption of a new vehicle needs to be predicted, a statistical table kd of average fuel consumption based on the segmented engine speed and the segmented torque needs to be generated in advance according to the existing whole vehicle calibration dataxAverage oil consumption distribution statistical table kyxAverage engine power distribution statistical table kgxThen, after normalization, amplifying by a times to obtain image-like data pkdx、pkyx、pkgx(ii) a Then p is addedkdx、pkyx、pkgxSeparately input model Mkd、Mky、MkgObtaining the result rkd、rky、rkgThen r iskd、rky、rkgInput model MresultAnd obtaining the predicted standard journey oil consumption value of the vehicle to be predicted.
In the above embodiments of the present application, W is preferably usedbestAnd Yvin-w-minTo the method of the present applicationThe following description is not intended to limit the present invention, and the average fuel consumption vehicle data set W is usedstdInstead of WbestUsing Yvin-w-stdIn place of Yvin-w-minIf the predicted standard distance oil consumption value in the final prediction result is the average hundred kilometers of oil consumption of a single vehicle per pass, other calculation modes with the same effect are used for replacing the WbestAnd Yvin-w-minAlso, the present invention falls within the technical scope of the claims.
In the embodiment, the used automobile oil consumption prediction model is obtained by analyzing and modeling a large number of collected oil consumption samples of users in the existing market based on big data analysis, is the optimal oil consumption model corresponding to the target automobile category, can effectively analyze the oil consumption performance of the target automobile category in the operation process, is beneficial to the optimization of the whole automobile oil consumption calibration strategy, and can timely and accurately predict the market adaptability of similar new automobile types.
It should be understood that, although the steps in the flowchart of fig. 2 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps in fig. 2 may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, which are not necessarily performed in sequence, but may be performed in turn or alternately with other steps or at least a portion of the other steps or stages.
As shown in fig. 3, it is a complete training process for predicting fuel consumption of a vehicle. The specific process is as follows:
step (1), acquiring a data set z by periodically (delta t is 1s) acquiring finished automobile identification vin, time t, automobile speed v, unit time mileage s, engine rotating speed n, engine torque te, accelerator opening k, unit time oil consumption y, engine power p, ignition switch state a, gear d and load w signals on equipment such as a finished automobile CAN bus, a GPS module, a driving recorder and the like,
z＝{vin,t,v,s,n,te,k,y,p,x,y,z,a,d,w}，
and the s is v, delta t, and y is the accumulated oil consumption in delta t time.
And (2) regularly transmitting the collected data Z (with the period being delta T x 10) back to a remote server for data storage through the T-Box, so as to obtain a multi-vehicle data set Z (Z [) Z [1,z2,z3,z4,z5,z6,......zn]；
And (3) screening the single vehicle data from the set Z through the vehicle identification vin to obtain a single vehicle data set Zvin＝[z1 vin,z2 vin,z3 vin,z4 vin,z5 vin,z6 vin,......,zn vin]All vehicle data sets are denoted as W ═ Z1 vin,Z2 vin,Z3 vin,......,Zn vin]；
Step (4), the bicycle data set ZvinSplitting a single-vehicle one-way operation interval data set Z according to the change condition of the load wvin-w＝[z1 vin-w,z2 vin-w,z3 vin-w,z4 vin-w,z5 vin-w,z6 vin-w,......,zn vin-w]. The method for splitting the operation interval by using the load w comprises the following steps: zvinTwo adjacent data z in the setk vin，zk+1 vinLoad weight component difference wk+1-wk>F (F is a threshold value, is a positive value, has a unit of kg, and can be set according to the actual vehicle type), and is judged as a loading event; if wk+1-wk<F, judging as a discharging event. And (4) loading and unloading goods in two adjacent times, and judging as an operation interval if the front part is loading goods and the rear part is unloading goods.
Step (5), collecting data Z of single-vehicle one-way operation intervalvin-wEach of z ink vin-wPerforming two-dimensional statistics to obtain a sheetAverage fuel consumption per hundred kilometers per pass of vehicle set Yvin-w＝[y1 vin-w,y2 vin-w,y3 vin-w,......,yn vin-w]，yk vin-wThe calculation method is as follows:
Figure BDA0002948988490000191
i is more than or equal to 1 and less than or equal to n, n is a set zk vin-wNumber of inner samples.
Step (6), collecting the average oil consumption per hundred kilometers per trip of the single vehicle Yvin-wCalculating normal mean value and standard deviation, and taking the normal mean value as average fuel consumption y per hundred kilometers of a single vehiclevin-w-stdAnd subtracting one standard deviation from the normal mean value to be used as the optimal oil consumption y of the single vehicle per hundred kilometersvin-w-min。
And (7) repeatedly executing the processes of the step (3), the step (4), the step (5) and the step (6) according to the vehicle identification vin of the same type of vehicle sold in a certain area market to obtain a multi-vehicle hundred-kilometer average oil consumption set Yvin-w-std＝[y1 vin-w-std,y2 vin-w-std,y3 vin-w-std,......,yn vin-w-std]And simultaneously obtaining the optimal oil consumption set Y of a plurality of vehicles in one hundred kilometersvin-w-min＝[y1 vin-w-min,y2 vin-w-min,y3 vin-w-min,......,yn vin-w-min,]。
Step (8) of comparing the set Y in step (7)vin-w-stdAnd set Yvin-w-minRespectively calculating normal mean value and standard deviation, and taking the normal mean value as the average oil consumption per hundred kilometers of the vehicle type in the regional market
Figure BDA0002948988490000192
Subtracting one standard deviation from the normal mean value to be used as the optimal oil consumption of one hundred kilometers of the model in the regional market
Figure BDA0002948988490000193
Figure BDA0002948988490000194
Vehicle one-way running interval data Zk vin-wForm an optimal fuel consumption vehicle data set Wbest＝[Z1 vin-w-best,Z2 vin-w-best,Z3 vin-w-best,......,Zn vin-w-best](ii) a Screening the vehicle single-pass operation interval data in the normal mean value plus or minus one standard deviation interval to form an average oil consumption vehicle data set Wstd。
Step (9), collecting the vehicle data W with optimal oil consumptionbestSample data Z in (1)k vin-w-bestSequentially carrying out average throttle opening statistics based on the rotation speed and the sectional torque of the sectional engine to obtain an optimal oil consumption vehicle throttle opening distribution table set KD ═ KD1,kd2,kd3,...,kdn]Wherein kdkAs shown in the following sample examples:
Figure BDA0002948988490000201
step (10), for KD in the result set KD obtained in step (9)kAfter normalization, amplifying by a times (a is 255 or other positive integer), and obtaining a picture-like data set Pkd＝[p1 kd，p2 kd，p3 kd，......,pn kd]Corresponding to the oil consumption set Y of one hundred kilometers in one-way running interval of a single vehiclevin-w-min＝[y1 vin-w-min,y2 vin-w-min,y3 vin-w-min,......,yn vin-w-min]Wherein y isk vin-w-minA positive integer value of four or five is required.
Step (11), establishing an accelerator opening image recognition model M by utilizing a neural networkkdUsing a set of picture-like data PkdAs input, Yvin-w-minTraining model M as outputkdModel late stage identification accuracy and PkdAnd Yvin-w-minThe number of samples in the set is positively correlated.
Step (1)2) referring to the processes of the step (9), the step (10) and the step (11), carrying out average fuel consumption distribution statistics based on the rotation speed and the sectional torque of the sectional engine to obtain an optimal fuel consumption vehicle fuel consumption distribution table set KY ═ KY1,ky2,ky3,...,kyn]Further obtain a class picture data set Pky＝[p1 ky，p2 ky，p3 ky，......,pn ky]And establishing a fuel consumption identification model M on the same principlekyIn combination with PkyAs input, Yvin-w-minTraining model M as outputky。
And (13) referring to the processes of the step (9), the step (10) and the step (11), carrying out average engine power distribution statistics based on the segmented engine rotating speed and the segmented torque, and obtaining an optimal oil consumption vehicle engine power distribution table set KG ═ KG1,kg2,kg3,...,kgn]Further obtain a class picture data set Pkg＝[p1 kg，p2 kg，p3 kg，......,pn kg]Establishing an engine power identification model M based on the same principlekgIn combination with PkgAs input, Yvin-w-minTraining model M as outputkg。
Step (14) of adding Pkd、Pky、PkgAs inputs, model M is used separatelykd、Mky、MkgPredicting to obtain a prediction result set Rkd、Rky、RkgFurther, the set X ═ { R ═ R is obtainedkd，Rky，Rkg，Yvin-w-min}。
Step (15), establishing a big data analysis classification model M by using a neural networkresultPredicting the optimal one hundred kilometers oil consumption of a single vehicle in one way, wherein the input is Rkd、Rky、RkgThe output is Yvin-w-minTraining the model by using the data set X; obtaining a final prediction model Mresult。
When the fuel consumption of a new vehicle is required to be predicted, the fuel consumption needs to be predicted according to the existing situationThe method comprises the steps that an average throttle opening statistical table kd based on the rotation speed and the sectional torque of a sectional engine is generated in advance through finished automobile calibration dataxAverage oil consumption distribution statistical table kyxAverage engine power distribution statistical table kgxThen, after normalization and normalization, amplifying by a times to obtain picture-like data pkdx、pkyx、pkgx(ii) a Then p is addedkdx、pkyx、pkgxSeparately input model Mkd、Mky、MkgObtaining the result rkd、rky、rkgThen r iskd、rky、rkgInput model MresultAnd obtaining the single-pass optimal hundred kilometer oil consumption of the single vehicle.
Further, if the vehicle data set W with average fuel consumption is used from step (9) to step (16)stdInstead of WbestUsing Yvin-w-stdIn place of Yvin-w-minThe final prediction result will be the average hundred kilometers per pass oil consumption of the fuel-oil automobile to be predicted.
In one embodiment, as shown in fig. 4, a training device for a fuel consumption prediction model of a vehicle is provided, which includes: a parameter obtaining module 402, a first calculating module 404, a second calculating module 406, a screening module 408, a statistic module 410, a converting module 412, and a training module 414, wherein:
the parameter obtaining module 402 is configured to obtain vehicle driving state parameters of a plurality of fuel-powered vehicles corresponding to the target vehicle category, where the vehicle driving state parameters include a driving parameter, a fuel consumption parameter, and a vehicle parameter.
The first calculating module 404 is configured to determine a first fuel consumption value of the standard route corresponding to each fuel vehicle according to the driving parameter of each fuel vehicle.
The second calculating module 406 is configured to screen a standard route second oil consumption value corresponding to the category of the target vehicle from the plurality of standard route first oil consumption values based on the first normal distribution state according to the first normal distribution state of the standard route first oil consumption values corresponding to the plurality of fuel vehicles, respectively.
The screening module 408 is used for determining a target fuel automobile corresponding to the second fuel consumption value of each standard journey and acquiring a target automobile driving state parameter corresponding to the target fuel automobile; the target automobile driving state parameters comprise target driving parameters, target oil consumption parameters and target automobile parameters.
The statistical module 410 is configured to perform statistics on the target fuel consumption parameters based on the parameter intervals corresponding to the target vehicle parameters of each target fuel vehicle, so as to obtain a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle.
The conversion module 412 is configured to convert each sample fuel consumption parameter distribution table into a corresponding fuel consumption parameter distribution class diagram, and form the fuel consumption parameter distribution class diagram corresponding to each target fuel vehicle and the second fuel consumption value of the standard route into one of the training samples of the target vehicle category.
The training module 414 is configured to, for all training samples in the target automobile category, use the oil consumption parameter distribution class diagram in each training sample as an input of the automobile oil consumption prediction model to be trained, use the corresponding second oil consumption value of the standard route as a training label of the automobile oil consumption prediction model to be trained, and perform model training on the automobile oil consumption prediction model corresponding to the target automobile category.
The training device of the automobile fuel consumption prediction model firstly obtains automobile running state parameters of a plurality of fuel automobiles corresponding to specific target automobile types, then according to the automobile running state parameters of each automobile, firstly determines a first fuel consumption value of a standard route of each automobile, then based on the normal distribution state of the first fuel consumption values of the plurality of standard routes corresponding to the plurality of automobiles in the target automobile types, screens out a second fuel consumption value of the standard route corresponding to the target automobile type, can determine the corresponding target fuel automobile in the target automobile type through the second fuel consumption value of the standard route, can obtain a sample fuel consumption parameter distribution table corresponding to each target fuel automobile by specifically processing the target automobile running state parameters of the target fuel automobile in the target automobile types, and can obtain a corresponding fuel consumption parameter distribution class diagram by further processing the sample fuel consumption parameter distribution table, each fuel consumption parameter distribution class diagram and the corresponding second fuel consumption value of the standard journey form a training sample corresponding to the target automobile category, and after all the training samples of the target automobile category are obtained, the training samples can be used for training the automobile fuel consumption prediction model of the target automobile category. By the method, the required target automobile running state parameters can be effectively screened out from the automobile running state parameters of the plurality of fuel oil automobiles corresponding to the target automobile type, the parameters are subjected to statistics and graph-like processing to become objects which are easy to identify by an automobile fuel consumption prediction model, and meanwhile, the result output by the automobile fuel consumption prediction model is guided by the corresponding training labels, so that the obtained trained automobile fuel consumption prediction model can accurately predict the fuel consumption of the fuel oil automobile to be predicted.
In one embodiment, the first computing module is further configured to: for each fuel oil automobile, dividing the corresponding route information into a plurality of single routes according to the load information; wherein, the load information is changed once, namely, the load information is marked as the starting point and/or the end point of a single journey; determining an oil consumption parameter corresponding to each single journey, and calculating a standard journey initial oil consumption value of each single journey according to the plurality of single journeys and the corresponding oil consumption parameters; and determining the first oil consumption value of the standard route corresponding to each fuel automobile according to the second normal distribution state of the initial oil consumption value of the standard route corresponding to each fuel automobile.
In the embodiment, the first fuel consumption value of the standard route corresponding to each fuel automobile can be accurately obtained through the initial fuel consumption values of the standard routes corresponding to the multiple single routes of each fuel automobile based on the corresponding normal distribution state, and the accuracy of the training sample of the automobile fuel consumption prediction model is improved.
In one embodiment, the first computing module is further configured to: when the first oil consumption value of the standard route is the average oil consumption value of the standard route, the first oil consumption value of the standard route is a second normal mean value; and when the first oil consumption value of the standard route is the optimal oil consumption value of the standard route, the first oil consumption value of the standard route is the difference value between the second normal mean value and the second standard deviation.
In the embodiment, the normal distribution processing is performed on the initial oil consumption value of the standard route of the single route, and the corresponding first oil consumption value of the standard route is obtained according to the normal distribution processing result, so that the result is more accurate, the error is smaller, the accuracy of the training sample of the automobile oil consumption prediction model is further improved, and the accuracy of the automobile oil consumption prediction model is further improved.
In one embodiment, the second calculation module is further configured to: when the standard route first oil consumption value is the standard route average oil consumption value, screening out the standard route first oil consumption value with the deviation not larger than the first standard deviation from the plurality of standard route first oil consumption values as a standard route second oil consumption value; and when the first oil consumption value of the standard route is the optimal oil consumption value of the standard route, screening the first oil consumption value of the standard route which is not more than the first normal mean value from the first oil consumption values of the standard route to be used as a second oil consumption value of the standard route.
In the embodiment, the first fuel consumption values of the standard routes are further screened, so that the screening work of 'best from best' in the target automobile categories is realized, the second fuel consumption values of the screened standard routes correspond to a plurality of target fuel automobiles with the best driving performance, and the plurality of target fuel automobiles with the best driving performance are used as the construction basis of the training samples, so that more accurate training samples and automobile fuel consumption prediction models can be obtained.
In one embodiment, the statistics module is further configured to: determining at least one first parameter interval corresponding to the engine speed parameter and at least one second parameter interval corresponding to the engine torque parameter; acquiring an initial parameter distribution table, wherein the initial parameter distribution table comprises at least one parameter distribution domain, and each parameter distribution domain corresponds to a first parameter interval and a second parameter interval; and counting the target fuel consumption parameters corresponding to each target fuel vehicle based on the obtained initial parameter distribution table, and obtaining a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle.
In the embodiment, the vehicle parameters are further determined, the parameter distribution domain is divided to carry out statistics on the fuel consumption parameters, so that the statistical result is more accurate, the fuel consumption parameter distribution table for manufacturing the training sample is more accurate, the fuel consumption performance of the used vehicle type in the market can be effectively analyzed, and the training effect of the vehicle fuel consumption prediction model is better.
In one embodiment, the conversion module is further configured to: respectively carrying out normalization processing on each sample oil consumption parameter distribution table to obtain a plurality of initial fuel oil automobile oil consumption distribution class diagrams; and determining training pixels of the automobile oil consumption prediction model, and amplifying each initial fuel oil automobile oil consumption distribution class diagram according to the training pixels to obtain a plurality of fuel oil automobile oil consumption distribution class diagrams.
In the embodiment, the adaptation degree between the fuel automobile fuel consumption distribution class diagram and the automobile fuel consumption prediction model can be effectively improved by means of normalization and scaling on the premise of ensuring the precision of the training sample, so that the automobile fuel consumption prediction model can well identify the information characteristics of the fuel automobile fuel consumption distribution class diagram and give accurate prediction based on the information characteristics.
In one embodiment, the apparatus further comprises a prediction module to: determining a fuel automobile to be predicted belonging to the target automobile category, and acquiring a trained target automobile fuel consumption prediction model; acquiring a to-be-processed fuel consumption parameter distribution table corresponding to the to-be-predicted vehicle; obtaining a fuel consumption distribution class diagram of the fuel vehicle to be processed corresponding to the vehicle to be predicted according to the fuel consumption parameter distribution table to be processed; and inputting the fuel automobile fuel consumption distribution diagram to be processed into a trained target automobile fuel consumption prediction model to obtain a prediction standard journey fuel consumption value of the current vehicle to be predicted.
In the embodiment, the used automobile oil consumption prediction model is obtained by analyzing and modeling a large number of collected oil consumption samples of users in the existing market based on big data analysis, is the optimal oil consumption model corresponding to the target automobile category, can effectively analyze the oil consumption performance of the target automobile category in the operation process, is beneficial to the optimization of the whole automobile oil consumption calibration strategy, and can timely and accurately predict the market adaptability of similar new automobile models so as to determine whether to adjust the technical parameters of the new automobile models.
For specific limitations of the training device for the fuel consumption prediction model of the vehicle, reference may be made to the above limitations on the training method for the fuel consumption prediction model of the vehicle, and details are not described herein again. All or part of each module in the training device of the automobile fuel consumption prediction model can be realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a server, the internal structure of which may be as shown in fig. 5. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer equipment is used for storing the training data of the automobile oil consumption prediction model. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to realize a training method of the automobile fuel consumption prediction model.
Those skilled in the art will appreciate that the architecture shown in fig. 5 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is further provided, which includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the above method embodiments when executing the computer program.
In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored which, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above examples only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims (10)

1. A training method of an automobile fuel consumption prediction model is characterized by comprising the following steps:
acquiring automobile driving state parameters of a plurality of fuel automobiles corresponding to the target automobile category, wherein the automobile driving state parameters comprise driving parameters, oil consumption parameters and automobile parameters;
determining a first oil consumption value of a standard route corresponding to each fuel automobile according to the driving parameters of each fuel automobile;
according to a first normal distribution state of standard route first oil consumption values corresponding to a plurality of fuel automobiles respectively, screening a standard route second oil consumption value corresponding to the target automobile type from the standard route first oil consumption values based on the first normal distribution state;
determining a target fuel automobile corresponding to the second fuel consumption value of each standard journey, and acquiring a target automobile driving state parameter corresponding to the target fuel automobile; the target automobile running state parameters comprise target running parameters, target oil consumption parameters and target automobile parameters;
counting the target fuel consumption parameters based on the parameter intervals corresponding to the target vehicle parameters of each target fuel vehicle respectively to obtain a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle;
converting each sample oil consumption parameter distribution table into a corresponding oil consumption parameter distribution class diagram respectively, and forming the oil consumption parameter distribution class diagram corresponding to each target fuel automobile and a second oil consumption value of a standard route into one of training samples of the target automobile category;
and for all training samples in the target automobile category, taking the oil consumption parameter distribution class diagram in each training sample as the input of the automobile oil consumption prediction model to be trained, taking the corresponding second oil consumption value of the standard distance as the training label of the automobile oil consumption prediction model to be trained, and carrying out model training on the automobile oil consumption prediction model corresponding to the target automobile category.
2. The method according to claim 1, wherein the driving parameters comprise load information and route information, and the determining the standard route first oil consumption value respectively corresponding to each fuel automobile according to the driving parameters of each fuel automobile comprises:
for each fuel oil automobile, dividing the corresponding route information into a plurality of single routes according to the load information; wherein, the load information is changed once, namely, the load information is marked as the starting point and/or the end point of a single journey;
determining an oil consumption parameter corresponding to each single journey, and calculating a standard journey initial oil consumption value of each single journey according to the plurality of single journeys and the corresponding oil consumption parameters;
and determining the first oil consumption value of the standard route corresponding to each fuel automobile according to the second normal distribution state of the initial oil consumption value of the standard route corresponding to each fuel automobile.
3. The method according to claim 2, wherein the second normal distribution state comprises a second normal mean and a second standard deviation; the step of determining the first fuel consumption value of the standard route corresponding to each fuel automobile according to the second normal distribution state of the initial fuel consumption value of the standard route corresponding to each fuel automobile comprises the following steps:
when the first oil consumption value of the standard route is the average oil consumption value of the standard route, the first oil consumption value of the standard route is a second normal mean value;
and when the first oil consumption value of the standard route is the optimal oil consumption value of the standard route, the first oil consumption value of the standard route is the difference value between the second normal mean value and the second standard deviation.
4. The method according to claim 3, wherein the first normal distribution state includes a first normal mean value and a first standard deviation, the first normal distribution state according to the standard routes first fuel consumption values respectively corresponding to the plurality of fuel automobiles screens out a standard route second fuel consumption value corresponding to the target automobile category from the plurality of standard route first fuel consumption values based on the first normal distribution state, and the method includes,
when the standard route first oil consumption value is the standard route average oil consumption value, screening out the standard route first oil consumption value with the deviation not larger than the first standard deviation from the plurality of standard route first oil consumption values as a standard route second oil consumption value;
and when the first oil consumption value of the standard route is the optimal oil consumption value of the standard route, screening the first oil consumption value of the standard route which is not more than the first normal mean value from the first oil consumption values of the standard route to be used as a second oil consumption value of the standard route.
5. The method according to claim 1, wherein the vehicle parameters include an engine speed parameter and an engine torque parameter, and the step of obtaining a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle by performing statistics on the target fuel consumption parameter based on a parameter interval corresponding to each target vehicle parameter of each target fuel vehicle comprises:
determining at least one first parameter interval corresponding to the engine speed parameter and at least one second parameter interval corresponding to the engine torque parameter;
acquiring an initial parameter distribution table, wherein the initial parameter distribution table comprises at least one parameter distribution domain, and each parameter distribution domain corresponds to a first parameter interval and a second parameter interval;
and counting the target fuel consumption parameters corresponding to each target fuel vehicle based on the obtained initial parameter distribution table, and obtaining a sample fuel consumption parameter distribution table corresponding to each target fuel vehicle.
6. The method according to claim 1, wherein the converting each of the sample fuel consumption parameter distribution tables into a corresponding fuel consumption parameter distribution class diagram comprises:
respectively carrying out normalization processing on each sample oil consumption parameter distribution table to obtain a plurality of initial fuel oil automobile oil consumption distribution class diagrams;
and determining training pixels of the automobile oil consumption prediction model, and amplifying each initial fuel oil automobile oil consumption distribution class diagram according to the training pixels to obtain a plurality of fuel oil automobile oil consumption distribution class diagrams.
7. The method according to any one of claims 1-6, further comprising:
determining a fuel automobile to be predicted belonging to the target automobile category, and acquiring a trained target automobile fuel consumption prediction model;
acquiring a to-be-processed fuel consumption parameter distribution table corresponding to the to-be-predicted vehicle;
obtaining a fuel consumption distribution class diagram of the fuel vehicle to be processed corresponding to the vehicle to be predicted according to the fuel consumption parameter distribution table to be processed;
and inputting the fuel automobile fuel consumption distribution diagram to be processed into a trained target automobile fuel consumption prediction model to obtain a prediction standard journey fuel consumption value of the current vehicle to be predicted.
8. A training device for a fuel consumption prediction model of an automobile is characterized by comprising:
the system comprises a parameter acquisition module, a parameter processing module and a parameter processing module, wherein the parameter acquisition module is used for acquiring automobile driving state parameters of a plurality of fuel automobiles corresponding to a target automobile type, and the automobile driving state parameters comprise driving parameters, oil consumption parameters and automobile parameters;
the first calculation module is used for determining a first fuel consumption value of a standard route corresponding to each fuel automobile according to the running parameters of each fuel automobile;
the second calculation module is used for screening out a standard route second oil consumption value corresponding to the target automobile type from the standard route first oil consumption values based on the first normal distribution state according to the first normal distribution state of the standard route first oil consumption values corresponding to the plurality of fuel automobiles respectively;
the screening module is used for determining a target fuel automobile corresponding to the second fuel consumption value of each standard journey and acquiring a target automobile driving state parameter corresponding to the target fuel automobile; the target automobile running state parameters comprise target running parameters, target oil consumption parameters and target automobile parameters;
the statistical module is used for carrying out statistics on the target fuel consumption parameters based on the parameter intervals corresponding to the target automobile parameters of each target fuel automobile respectively to obtain a sample fuel consumption parameter distribution table corresponding to each target fuel automobile;
the conversion module is used for converting each sample oil consumption parameter distribution table into a corresponding oil consumption parameter distribution class diagram respectively, and forming the oil consumption parameter distribution class diagram corresponding to each target fuel automobile and a second oil consumption value of a standard route into one of the training samples of the target automobile class;
and the training module is used for performing model training on the automobile oil consumption prediction model corresponding to the target automobile category by taking the oil consumption parameter distribution class diagram in each training sample as the input of the automobile oil consumption prediction model to be trained and taking the corresponding second oil consumption value of the standard distance as a training label of the automobile oil consumption prediction model to be trained for all the training samples in the target automobile category.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 7.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.
CN202110210292.1A 2021-02-23 2021-02-23 Training method and device for automobile fuel consumption prediction model and computer equipment Active CN112883648B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110210292.1A CN112883648B (en) 2021-02-23 2021-02-23 Training method and device for automobile fuel consumption prediction model and computer equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110210292.1A CN112883648B (en) 2021-02-23 2021-02-23 Training method and device for automobile fuel consumption prediction model and computer equipment

Publications (2)

Publication Number Publication Date
CN112883648A true CN112883648A (en) 2021-06-01
CN112883648B CN112883648B (en) 2022-06-17

Family

ID=76054464

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110210292.1A Active CN112883648B (en) 2021-02-23 2021-02-23 Training method and device for automobile fuel consumption prediction model and computer equipment

Country Status (1)

Country Link
CN (1) CN112883648B (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113250836A (en) * 2021-06-07 2021-08-13 一汽解放汽车有限公司 Engine control method, engine control device, computer equipment and storage medium
CN113624291A (en) * 2021-07-27 2021-11-09 三一专用汽车有限责任公司 Oil consumption monitoring method, oil consumption monitoring device and engineering vehicle
CN114426025A (en) * 2022-03-17 2022-05-03 一汽解放汽车有限公司 Driving assistance method, driving assistance device, computer equipment and storage medium
CN114781245A (en) * 2022-03-28 2022-07-22 北京理工大学 Transient automobile oil consumption estimation method and system based on high oil consumption related parameter correction
CN114834454A (en) * 2022-06-02 2022-08-02 中国第一汽车股份有限公司 Method and device for determining gasoline grade of vehicle and computer readable storage medium
CN116644865A (en) * 2023-07-27 2023-08-25 中汽信息科技(天津)有限公司 Commercial vehicle fuel consumption prediction method, electronic equipment and storage medium
CN116737845A (en) * 2023-05-24 2023-09-12 瑞修得信息科技(无锡)有限公司 Economic vehicle speed analysis method and system

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1956203A1 (en) * 2007-02-09 2008-08-13 Peugeot Citroën Automobiles S.A. Device and method for estimating the oil consumption of a vehicle engine
CN110147820A (en) * 2019-04-11 2019-08-20 北京远航通信息技术有限公司 Recommended method, device, equipment and the storage medium of the additional oil mass of flight
CN110509922A (en) * 2019-08-20 2019-11-29 一汽解放汽车有限公司 A kind of vehicle prediction cruise control method based on high-precision map
CN110516813A (en) * 2019-08-09 2019-11-29 常伟 A method of batteries of electric automobile RDR prediction is carried out based on big data machine learning
CN111476432A (en) * 2020-04-24 2020-07-31 北方工业大学 Vehicle oil consumption estimation method and system based on real-time road condition prediction
CN111523701A (en) * 2020-03-27 2020-08-11 广州亚美信息科技有限公司 Method and device for evaluating vehicle running condition, computer equipment and storage medium
CN111860317A (en) * 2020-07-20 2020-10-30 青岛特利尔环保集团股份有限公司 Boiler operation data acquisition method, system, equipment and computer medium

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1956203A1 (en) * 2007-02-09 2008-08-13 Peugeot Citroën Automobiles S.A. Device and method for estimating the oil consumption of a vehicle engine
CN110147820A (en) * 2019-04-11 2019-08-20 北京远航通信息技术有限公司 Recommended method, device, equipment and the storage medium of the additional oil mass of flight
CN110516813A (en) * 2019-08-09 2019-11-29 常伟 A method of batteries of electric automobile RDR prediction is carried out based on big data machine learning
CN110509922A (en) * 2019-08-20 2019-11-29 一汽解放汽车有限公司 A kind of vehicle prediction cruise control method based on high-precision map
CN111523701A (en) * 2020-03-27 2020-08-11 广州亚美信息科技有限公司 Method and device for evaluating vehicle running condition, computer equipment and storage medium
CN111476432A (en) * 2020-04-24 2020-07-31 北方工业大学 Vehicle oil consumption estimation method and system based on real-time road condition prediction
CN111860317A (en) * 2020-07-20 2020-10-30 青岛特利尔环保集团股份有限公司 Boiler operation data acquisition method, system, equipment and computer medium

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
D MOSER等: "Flexible spacing adaptive cruise control using stochastic model predictive control", 《IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY》, 31 December 2017 (2017-12-31), pages 1 - 14 *
冯红晶: "基于VSP的山区高速公路商用车油耗计算模型研究", 《中国优秀硕士学位论文全文数据库 (工程科技Ⅱ辑)》, no. 2, 15 February 2017 (2017-02-15), pages 035 - 799 *
刘扬等: "基于神经认知计算模型的高分辨率遥感图像场景分类", 《系统工程与电子技术》, no. 11, 15 November 2015 (2015-11-15), pages 205 - 215 *
王运运等: "基于AIOHMM模型的驾驶行为预测", 《信息通信》, no. 03, 15 March 2019 (2019-03-15), pages 125 - 128 *
黄登高: "用于公交车辆行驶油耗优化的工况预测模型及其自学习策略", 《中国博士学位论文全文数据库 (工程科技Ⅱ辑)》, no. 8, 15 August 2018 (2018-08-15), pages 035 - 11 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113250836A (en) * 2021-06-07 2021-08-13 一汽解放汽车有限公司 Engine control method, engine control device, computer equipment and storage medium
CN113250836B (en) * 2021-06-07 2022-10-14 一汽解放汽车有限公司 Engine control method, engine control device, computer equipment and storage medium
CN113624291A (en) * 2021-07-27 2021-11-09 三一专用汽车有限责任公司 Oil consumption monitoring method, oil consumption monitoring device and engineering vehicle
CN114426025A (en) * 2022-03-17 2022-05-03 一汽解放汽车有限公司 Driving assistance method, driving assistance device, computer equipment and storage medium
CN114426025B (en) * 2022-03-17 2023-11-14 一汽解放汽车有限公司 Driving assistance method, driving assistance device, computer device, and storage medium
CN114781245A (en) * 2022-03-28 2022-07-22 北京理工大学 Transient automobile oil consumption estimation method and system based on high oil consumption related parameter correction
CN114781245B (en) * 2022-03-28 2024-05-28 北京理工大学 Transient automobile fuel consumption estimation method and system based on fuel consumption strong correlation parameter correction
CN114834454A (en) * 2022-06-02 2022-08-02 中国第一汽车股份有限公司 Method and device for determining gasoline grade of vehicle and computer readable storage medium
CN116737845A (en) * 2023-05-24 2023-09-12 瑞修得信息科技(无锡)有限公司 Economic vehicle speed analysis method and system
CN116737845B (en) * 2023-05-24 2024-02-02 瑞修得信息科技(无锡)有限公司 Economic vehicle speed analysis method and system
CN116644865A (en) * 2023-07-27 2023-08-25 中汽信息科技(天津)有限公司 Commercial vehicle fuel consumption prediction method, electronic equipment and storage medium
CN116644865B (en) * 2023-07-27 2023-10-20 中汽信息科技(天津)有限公司 Commercial vehicle fuel consumption prediction method, electronic equipment and storage medium

Also Published As

Publication number Publication date
CN112883648B (en) 2022-06-17

Similar Documents

Publication Publication Date Title
CN112883648B (en) Training method and device for automobile fuel consumption prediction model and computer equipment
CN111452619B (en) Online energy consumption prediction method and system for electric vehicle
CN109697854B (en) Multi-dimensional urban road traffic state evaluation method
US10928277B1 (en) Intelligent telematics system for providing vehicle vocation
WO2022174679A1 (en) Method and apparatus for predicting voltage inconsistency fault of battery cells, and server
CN116206451A (en) Intelligent traffic flow data analysis method
CN116558834A (en) PEMS emission analysis method, PEMS emission analysis device, PEMS emission analysis computer equipment and PEMS emission analysis storage medium
CN106157657A (en) A kind of moving state identification system and method for mobile subscriber
Barbado et al. Vehicle fuel optimization under real-world driving conditions: An explainable artificial intelligence approach
CN112836967B (en) New energy automobile battery safety risk assessment system
Shahariar et al. Real-driving CO2, NOx and fuel consumption estimation using machine learning approaches
CN111754261B (en) Method and device for evaluating taxi willingness and terminal equipment
CN114419894A (en) Method and system for setting and monitoring use of parking space in road
CN111815250B (en) Cargo state identification method and device for logistics and classification modeling method
Makarova et al. Improving the reliability of autonomous vehicles in a branded service system using big data
EP3819840A1 (en) Method for classifying trade industry of a vehicle
Khoshkangini et al. Vehicle usage extraction using unsupervised ensemble approach
Makarova et al. Changing the Maintenance and Repair System While Expanding the Connected Vehicles Fleet.
Gosain et al. Customer usage space classification and representative duty cycle development using K-means clustering
Mortabit et al. Instantaneous Vehicle Fuel Consumption Estimation Using Neural Networks
CN117093818B (en) Method and system for measuring and calculating carbon emission of urban distribution logistics vehicle
CN113515829B (en) Situation awareness method for transmission line hardware defects under extremely cold disasters
CN117315943B (en) Monitoring analysis and early warning method and system for overrun transportation violations
CN113064933B (en) Vehicle weight prediction method and system
Ponce et al. A novel approach in constructing virtual real driving emission trips through genetic algorithm optimization

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant