CN115907104B - Hydraulic pump fault prediction method based on big data of Internet of things and machine learning - Google Patents
Hydraulic pump fault prediction method based on big data of Internet of things and machine learning Download PDFInfo
- Publication number
- CN115907104B CN115907104B CN202211371106.3A CN202211371106A CN115907104B CN 115907104 B CN115907104 B CN 115907104B CN 202211371106 A CN202211371106 A CN 202211371106A CN 115907104 B CN115907104 B CN 115907104B
- Authority
- CN
- China
- Prior art keywords
- data
- fault
- hydraulic pump
- parameter
- value
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000010801 machine learning Methods 0.000 title claims abstract description 14
- 238000012545 processing Methods 0.000 claims abstract description 42
- 238000012549 training Methods 0.000 claims abstract description 39
- 238000011156 evaluation Methods 0.000 claims description 40
- 230000002159 abnormal effect Effects 0.000 claims description 9
- 238000005086 pumping Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 5
- 238000011109 contamination Methods 0.000 claims description 4
- 238000012217 deletion Methods 0.000 claims description 4
- 230000037430 deletion Effects 0.000 claims description 4
- 230000001105 regulatory effect Effects 0.000 claims description 4
- 238000011425 standardization method Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 238000010219 correlation analysis Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 230000006798 recombination Effects 0.000 claims description 3
- 238000005215 recombination Methods 0.000 claims description 3
- 238000012423 maintenance Methods 0.000 abstract description 10
- 239000003921 oil Substances 0.000 description 7
- 230000008569 process Effects 0.000 description 4
- 239000010720 hydraulic oil Substances 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000001174 ascending effect Effects 0.000 description 2
- 230000003111 delayed effect Effects 0.000 description 2
- 239000007788 liquid Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Fluid-Pressure Circuits (AREA)
Abstract
The invention relates to the technical field of hydraulic pumps, in particular to a hydraulic pump fault prediction method based on big data of the Internet of things and machine learning, which comprises the following steps: s1, selecting characteristic parameters for carrying out fault prediction and a corresponding prediction model according to the type of the hydraulic pump fault to be predicted; the characteristic parameter includes a pump pressure; the prediction model is used for predicting pump pressure data of a preset time in the future according to the actual parameter data; s2, installing corresponding sensors at corresponding positions of the equipment according to the selected characteristic parameters, and acquiring corresponding data of the selected characteristic parameters; s3, carrying out quality processing on the data acquired by the sensor according to the quality requirement of the parameter data to obtain operation data; s4, performing preset pretreatment on the operation data to obtain clean data meeting training requirements. The hydraulic pump fault prediction method and the hydraulic pump fault prediction device can accurately predict the faults of the hydraulic pump and improve the pertinence of hydraulic pump maintenance.
Description
Technical Field
The invention relates to the technical field of hydraulic pumps, in particular to a hydraulic pump fault prediction method based on big data of the Internet of things and machine learning.
Background
The hydraulic pump is a hydraulic element for providing pressurized liquid for hydraulic transmission, is a pump, is widely applied to various engineering and industrial mechanical equipment, and has the function of converting mechanical energy into pressure energy of liquid. The hydraulic pump is one of the elements with more faults in the industrial equipment, and once the hydraulic pump fails, the normal operation of the hydraulic system of the equipment is immediately affected, and even the hydraulic system cannot work. The importance of hydraulic pumps to industrial equipment is self-evident, but at present, the fault treatment of hydraulic pumps is basically a method for periodically maintaining and then repairing after the fault. The processing mode is not strong in pertinence, and the mode of repairing after failure has a great influence on related work.
Therefore, how to accurately predict the failure of the hydraulic pump and improve the pertinence of the maintenance of the hydraulic pump becomes a problem to be solved at present.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides the hydraulic pump fault prediction method based on big data of the Internet of things and machine learning, which can accurately predict the hydraulic pump fault and improve the pertinence of hydraulic pump maintenance.
In order to solve the technical problems, the invention adopts the following technical scheme:
a hydraulic pump fault prediction method based on big data of the Internet of things and machine learning comprises the following steps:
s1, selecting characteristic parameters for carrying out fault prediction and a corresponding prediction model according to the type of the hydraulic pump fault to be predicted; the characteristic parameter includes a pump pressure; the prediction model is used for predicting pump pressure data of a preset time in the future according to the actual parameter data;
s2, installing corresponding sensors at corresponding positions of the equipment according to the selected characteristic parameters, and acquiring corresponding data of the selected characteristic parameters;
s3, carrying out quality processing on the data acquired by the sensor according to the quality requirement of the parameter data to obtain operation data;
s4, carrying out preset pretreatment on the operation data to obtain clean data meeting training requirements;
s5, performing feature processing on clean data meeting training requirements according to the features of the data required by the selected prediction model to obtain training data meeting the model requirements;
s6, training and optimizing the prediction model through training data to obtain a parameter prediction model; the pump pressure data output by the parameter prediction model is brought into a preset fault evaluation algorithm to perform fault prediction, and a set value in the fault evaluation algorithm is adjusted by combining a fault prediction result of the fault evaluation algorithm and an actual fault condition, so that the fault prediction accuracy of the fault evaluation algorithm meets a preset accuracy requirement;
s7, inputting the collected real-time parameters into a parameter prediction model, and then, taking the output pumping pressure data of the parameter prediction model into a fault evaluation algorithm with a well-regulated set value to predict pumping pressure faults of the equipment.
Preferably, in S6, the preset fault assessment algorithm is:
if it isDetermining that the hydraulic pump has a fault; wherein N is the predicted data amount of the pump pressure on the same day; y is i Pump pressure data of the ith time of the day predicted by the parameter prediction model; y is r Is a preset pump pressure standard value; w is a set deviation value; threshold is set; the set values include a set deviation value and a set threshold value.
Preferably, in S2, the corresponding positions of the apparatus include node positions of the steering, radial clearance, axial clearance, oil contamination and/or oil delivery of the hydraulic pump.
Preferably, the parameter prediction model and the fault evaluation algorithm are both arranged on the platform of the internet of things; and S2, after the corresponding sensor is installed at the corresponding position of the equipment, the sensor is also connected with the Internet of things platform in a long way.
Preferably, S3 comprises:
s301, processing discrete variable parameters to enable the discrete variable parameters to be changed into dummy variable parameters;
s302, maintaining and processing the dummy variable parameters and the continuity variable parameters according to preset quality requirements to obtain operation data; the quality requirements include integrity, normalization, consistency, accuracy, and relevance.
Preferably, S4 comprises:
s401, checking operation data, and converting checked parameters into a data format meeting preset requirements to obtain clean data;
s402, performing exception processing on clean data, marking the data exceeding the actual value range of the field as an abnormal value, and deleting the data;
s403, performing deletion processing on the clean data;
s404, performing protocol processing on continuous variable data in the clean data by adopting a Zscore standardization method, and performing protocol processing on discrete variable data in the clean data by adopting single-heat coding to obtain the clean data meeting training requirements.
Preferably, in S401, the checking includes: check whether the data type of each field accords with logic and whether the value range accords with reality.
Preferably, S403 includes:
s4031, deleting columns with a null value of > 70%;
s4032, deleting columns with only one value except the null value;
s4033, deleting rows with a null of > 70%;
s403, filling the missing value by using an upward filling method.
Preferably, S5 comprises:
s501, screening clean data of equipment in a working state;
s502, performing characteristic transformation processing on the category type variable through single-heat coding;
s503, combining a working model of the equipment, and performing variable recombination operation on the clean data to obtain new parameter data;
s503, carrying out correlation analysis on each parameter in the clean data and the newly obtained parameter, and if the parameters with correlation exceeding 85% exist, only preserving one parameter; training data is obtained.
Preferably, in S7, if the pump failure prediction result of the device is that the pump failure will occur, an early warning is sent.
Compared with the prior art, the invention has the following beneficial effects:
1. according to the invention, a technician can select specific characteristic parameters according to the specific type of the hydraulic pump fault to be detected, and obtain training data for training the prediction model by collecting corresponding characteristic parameter data and performing quality processing, data processing and characteristic processing. After the training data is used for training the prediction model, the training model can predict pump pressure data of a preset time (such as 24 hours) in the future through real-time parameter data. And then, carrying pump pressure data output by the parameter prediction model into a preset fault evaluation algorithm to perform fault prediction, and adjusting a set value in the fault evaluation algorithm by combining a fault prediction result of the fault evaluation algorithm and an actual fault condition to ensure that the fault prediction accuracy of the fault evaluation algorithm meets a preset accuracy requirement. Stated another way, by the failure assessment algorithm, it is possible to know whether the hydraulic pump is failing within a predetermined time in the future. And then, the collected real-time parameters can be input into a parameter prediction model, and the pump pressure fault prediction is carried out on the equipment by combining a fault evaluation algorithm with the well-adjusted set value.
In summary, the hydraulic pump fault prediction method can accurately predict the faults of the hydraulic pump and improve the pertinence of hydraulic pump maintenance.
2. The invention skillfully uses the prediction model and the fault evaluation algorithm, and various fault types of various devices are processed according to the steps of the method, corresponding acquired data are selected and processed, and the set value in the fault evaluation algorithm is adjusted according to the actual situation. The method has the advantages of good prediction accuracy, wide application range and good suitability.
3. The invention provides a specific fault evaluation algorithm, and the core idea of the algorithm is that the difference between a predicted pump pressure value and a pump pressure standard value is calculated as a deviation value, and the ratio of the deviation value to the standard value can be regarded as the deviation degree of the pump pressure. Further, by analyzing the duty ratio at which the degree of deviation is larger than the set value, if the degree of deviation is larger than the set threshold value, it is definitely known that the pump pressure of the hydraulic pump is in an abnormal state in a time exceeding the threshold time duty ratio, and therefore, it is possible to determine that the predicted state of the hydraulic pump is a failure state. The method is strict and accurate, and is very simple to adjust, and the set deviation value and the set threshold value are only required to be adjusted by combining the actual state of the hydraulic pump.
4. The hydraulic pump can also send out early warning when the hydraulic pump is predicted to fail. The technical staff can maintain the hydraulic pump even if the hydraulic pump fails, so that the situation that a large amount of work is delayed due to maintenance after the hydraulic pump fails can be prevented, and the economic loss caused by unplanned shutdown can be effectively reduced. In addition, because the fault prediction uses the selected parameters as related parameters, technicians can have pertinence during maintenance, and the maintenance efficiency can be improved.
5. The invention provides a specific method for selecting and processing parameter data, which can ensure the validity of the data during the training of the prediction model.
Drawings
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings, in which:
FIG. 1 is a flow chart of a first embodiment;
fig. 2 is a schematic diagram illustrating descriptions of main variables of a data set in the second embodiment.
Detailed Description
The following is a further detailed description of the embodiments:
example 1
As shown in fig. 1, the embodiment discloses a hydraulic pump fault prediction method based on big data of internet of things and machine learning, which comprises the following steps:
s1, selecting characteristic parameters for carrying out fault prediction and a corresponding prediction model according to the type of the hydraulic pump fault to be predicted; the characteristic parameter includes a pump pressure; the prediction model is used for predicting pump pressure data of a preset time in the future according to the actual parameter data;
s2, installing corresponding sensors at corresponding positions of the equipment according to the selected characteristic parameters, acquiring corresponding data of the selected characteristic parameters, and establishing long connection between the sensors and the Internet of things platform. Wherein the corresponding positions of the apparatus include node positions of the steering, radial clearance, axial clearance, oil contamination and/or oil delivery of the hydraulic pump.
And S3, carrying out quality processing on the data acquired by the sensor according to the quality requirement of the parameter data to obtain operation data. In specific implementation, S3 includes:
s301, processing discrete variable parameters to enable the discrete variable parameters to be changed into dummy variable parameters;
s302, maintaining and processing the dummy variable parameters and the continuity variable parameters according to preset quality requirements to obtain operation data; the quality requirements include integrity, normalization, consistency, accuracy, and relevance.
S4, performing preset pretreatment on the operation data to obtain clean data meeting training requirements.
In specific implementation, S4 includes:
s401, checking operation data, and converting checked parameters into a data format meeting preset requirements to obtain clean data; wherein the verifying comprises: check whether the data type of each field accords with logic and whether the value range accords with reality.
S402, performing exception processing on clean data, marking the data exceeding the actual value range of the field as an abnormal value, and deleting the data;
s403, performing deletion processing on the clean data;
specifically, S403 includes:
s4031, deleting columns with a null value of > 70%;
s4032, deleting columns with only one value except the null value;
s4033, deleting rows with a null of > 70%;
s403, filling the missing value by using an upward filling method.
S404, performing protocol processing on continuous variable data in the clean data by adopting a Zscore standardization method, and performing protocol processing on discrete variable data in the clean data by adopting single-heat coding to obtain the clean data meeting training requirements.
S5, performing feature processing on the clean data meeting the training requirements according to the features of the data required by the selected prediction model to obtain the training data meeting the model requirements.
In specific implementation, S5 includes:
s501, screening clean data of equipment in a working state;
s502, performing characteristic transformation processing on the category type variable through single-heat coding;
s503, combining a working model of the equipment, and performing variable recombination operation on the clean data to obtain new parameter data;
s503, carrying out correlation analysis on each parameter in the clean data and the newly obtained parameter, and if the parameters with correlation exceeding 85% exist, only preserving one parameter; training data is obtained.
S6, training and optimizing the prediction model through training data to obtain a parameter prediction model; and the pump pressure data output by the parameter prediction model is brought into a preset fault evaluation algorithm to perform fault prediction, and the set value in the fault evaluation algorithm is adjusted by combining the fault prediction result of the fault evaluation algorithm and the actual fault condition, so that the fault prediction accuracy of the fault evaluation algorithm meets the preset accuracy requirement.
In specific implementation, the preset fault evaluation algorithm is as follows:
if it isDetermining that the hydraulic pump has a fault; wherein N is the predicted data amount of the pump pressure on the same day; y is i Pump pressure data of the ith time of the day predicted by the parameter prediction model; y is r Is a preset pump pressure standard value; w is a set deviation value; threshold is set; the set values include a set deviation value and a set threshold value.
The core idea of the algorithm is that the difference between the predicted pump pressure value and the pump pressure standard value is calculated as a deviation value, and then the ratio of the deviation value to the standard value can be regarded as the deviation degree of the pump pressure. Further, by analyzing the duty ratio at which the degree of deviation is larger than the set value, if the degree of deviation is larger than the set threshold value, it is definitely known that the pump pressure of the hydraulic pump is in an abnormal state in a time exceeding the threshold time duty ratio, and therefore, it is possible to determine that the predicted state of the hydraulic pump is a failure state. The method is strict and accurate, and is very simple to adjust, and the set deviation value and the set threshold value are only required to be adjusted by combining the actual state of the hydraulic pump.
S7, inputting the collected real-time parameters into a parameter prediction model, and then, taking the output pumping pressure data of the parameter prediction model into a fault evaluation algorithm with a well-regulated set value to predict pumping pressure faults of the equipment. And if the pump pressure fault prediction result of the equipment is that the pump pressure fault will occur, sending out early warning.
It should be noted that, the parameter prediction model and the fault evaluation algorithm are both set on the platform of the internet of things.
According to the invention, a technician can select specific characteristic parameters according to the specific type of the hydraulic pump fault to be detected, and obtain training data for training the prediction model by collecting corresponding characteristic parameter data and performing quality processing, data processing and characteristic processing. After the training data is used for training the prediction model, the training model can predict pump pressure data of a preset time (such as 24 hours) in the future through real-time parameter data. And then, carrying pump pressure data output by the parameter prediction model into a preset fault evaluation algorithm to perform fault prediction, and adjusting a set value in the fault evaluation algorithm by combining a fault prediction result of the fault evaluation algorithm and an actual fault condition to ensure that the fault prediction accuracy of the fault evaluation algorithm meets a preset accuracy requirement. Stated another way, by the failure assessment algorithm, it is possible to know whether the hydraulic pump is failing within a predetermined time in the future. And then, the collected real-time parameters can be input into a parameter prediction model, and the pump pressure fault prediction is carried out on the equipment by combining a fault evaluation algorithm with the well-adjusted set value.
In addition, the invention skillfully uses the prediction model and the fault evaluation algorithm, and various fault types of various devices are processed according to the steps of the method, corresponding acquired data are selected and processed, and the set value in the fault evaluation algorithm is adjusted according to the actual situation. The method has the advantages of good prediction accuracy, wide application range and good suitability. In addition, the hydraulic pump can also send out early warning when the hydraulic pump is predicted to be in fault. The technical staff can maintain the hydraulic pump even if the hydraulic pump fails, so that the situation that a large amount of work is delayed due to maintenance after the hydraulic pump fails can be prevented, and the economic loss caused by unplanned shutdown can be effectively reduced. In addition, because the fault prediction uses the selected parameters as related parameters, technicians can have pertinence during maintenance, and the maintenance efficiency can be improved.
Example two
In order to facilitate a better understanding of the technical content of the first embodiment, a technical solution in the first embodiment will be described below by way of an example. In this example, the equipment that carries the hydraulic pump is an excavator, and the faults to be predicted are hydraulic wear and hydraulic failure due to oil contamination. In this case, the manifestation of the failure is typically the following:
1. the hydraulic pump is seriously worn and unbalanced in rotation, and abnormal sound is generated;
2. the abrasion of the hydraulic pump increases the internal discharge, the oil output of the hydraulic pump becomes smaller, and the pressure is low;
3. the amount of casing leakage is large, resulting in a high hydraulic oil temperature.
In addition, the power of the hydraulic pump is derived from the engine, so that the characteristic selection aspect needs to select the parameters of the engine and the related parameters of oil. If the hydraulic pump fails, the data of the current sensor are combined according to three manifestations of the failure, and the analysis can be performed by adopting a mode of predicting the pump pressure. From the above analysis, the parameters to be acquired can be known.
And then, installing corresponding sensors at key positions of the excavator and the hydraulic pump thereof to acquire corresponding parameter data, and transmitting the parameter data to the platform of the Internet of things. It should be noted that if the system has corresponding parameter data, the parameter data may also be directly imported into the internet of things system, which is not described herein.
In this example, the parameter data source is real-time data of each minute collected by an excavator from 2021, 6, 1, to 2022, 10, 7, and the total of 101359 lines of data includes 47 columns of parameter values, and the collected data is divided into discrete variables and continuous variables according to the data value range, as shown in fig. 2. The values of the continuous variables represent numerical meanings, but the values of the discrete variables are not numerical meanings, although they may be numerical, and they need to be processed for use, and a common processing method is to convert them into dummy variables.
And then, carrying out predictive preprocessing on the operation data to obtain clean data meeting the training requirements.
The process mainly comprises four steps: data preprocessing, exception handling, miss handling, and reduction handling.
1. Data preprocessing
The stage mainly converts the collected original data into clean and complete data which can be used for modeling after verification and processing. It is first necessary to check whether the data type of each field is logical. And then, whether the value range accords with the reality is checked, and the original data is converted into the data required by people through a series of processing. Descriptive statistics of the parameter data are shown in table 1:
descriptive statistics of the data of Table 1
| Metn | Std | Min | 25％ | 50％ | 75％ | Max | |
| Rotational speed | 1697.89 | 476.75 | 113.50 | 1002.6 | 1997.4 | 2042.0 | 2144.80 |
| Percent torque | 44.15 | 35.61 | 0.00 | 4.50 | 44.50 | 78.50 | 100.00 |
| Water temperature | 77.56 | 13.59 | 8.08 | 68.58 | 77.18 | 89.84 | 103.21 |
| Torque moment | 345.76 | 308.72 | 0.00 | 38.97 | 289.54 | 620.58 | 1038.84 |
| Output power | 2.73 | 3.12 | 0.00 | 0.02 | 1.35 | 5.06 | 10.82 |
| Action | 0.22 | 0.41 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 |
| Action type | 96.31 | 308.55 | 0.00 | 0.00 | 70.00 | 89.00 | 655356.0 |
| Mode of operation | 22.00 | 0.07 | 22.00 | 22.00 | 22.00 | 22.00 | 34.00 |
| Pump pressure | 131.87 | 89.58 | 10.00 | 34.00 | 127.00 | 208.00 | 378.00 |
Before abnormal value and missing value processing, the data are sorted according to the ascending order of the acquisition time by machine number.
Before abnormal value and missing value processing, the data are sorted according to the ascending order of the acquisition time by machine number.
2. Exception handling
The abnormal value determination method comprises the following steps: data that is outside the actual value range of each field is considered an outlier.
The outlier processing method comprises the following steps: and deleting.
3. Deletion handling
1) Deleting columns with a null of > 70%;
2) Deleting columns with only one value, except for the null value;
3) Deleting rows with a null of > 70%;
4) The continuous variable and the discrete variable are filled with missing values by using an upward filling method.
4. Protocol processing
Continuous variable: adopting a Zscore standardization method;
discrete variable: and (5) single-heat coding.
Then, data modeling and training are carried out, and the process is mainly completed through feature engineering and XGBOOST algorithm.
1. Characteristic engineering:
1) Data preliminary screening
The parameters associated with the hydraulic system are selected and data is in the operating state (i.e., the engine speed >0 and torque percentage >0, etc. are met) such that the selected data is valid and reflects the operating state of the excavator. After processing, only 98777 lines of data remain.
2) Feature transformation
Since the partial model only supports numerical data as input, we need to pre-process the class type variables in the dataset by single-hot encoding in advance. One-Hot encoding, the One-Hot encoding, also known as "One-bit efficient encoding", changes a feature with m possible values into m binary features, and only One of the m binary features is activated at a time.
3) Feature combination
After the single-heat coding, the variables are recombined by combining the information of the related modes (P, E, etc.) of the excavator. For continuous variables, there are some relations in the continuous variables, and new important parameters such as torque, output work and the like can be obtained through some operations.
4) Feature screening
Firstly, according to professional knowledge, parameters with higher association degree with the pump pressure are hydraulic oil temperature, hydraulic pump flow, hydraulic pump internal leakage, engine rotation speed, torque percentage and the like (wherein the data of the hydraulic oil temperature, the hydraulic pump flow, the hydraulic pump internal leakage and the like cannot be obtained at present), and interference parameters irrelevant to study variables are removed. And according to the correlation coefficient of the characteristic, only one variable with the correlation coefficient exceeding 85% is reserved.
Through the above process, training data for training the predictive model can be obtained.
Then, training and optimizing the prediction model through training data to obtain a parameter prediction model; and the pump pressure data output by the parameter prediction model is brought into a preset fault evaluation algorithm for carrying out the processBarrier prediction, i.e.If so, it is determined that the hydraulic pump has a failure. And the set value in the fault evaluation algorithm is adjusted by combining the fault prediction result of the fault evaluation algorithm and the actual fault condition, so that the fault prediction accuracy of the fault evaluation algorithm meets the preset precision requirement.
In this example, when w=20% and threshold=20%, a prediction accuracy that satisfies the requirement can be obtained.
In other specific embodiments, the person skilled in the art can correspondingly adjust the set value according to the specific data condition, so that the set value meets the precision requirement, and the details are not repeated here.
And then, the collected real-time parameters can be input into a parameter prediction model, and pump pressure data output by the parameter prediction model is brought into a fault evaluation algorithm with a well-regulated set value, so that pump pressure fault prediction is carried out on the equipment. And if the pump pressure fault prediction result of the equipment is that the pump pressure fault will occur, sending out early warning.
Finally, it should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the technical solution, and those skilled in the art should understand that modifications and equivalents may be made to the technical solution of the present invention without departing from the spirit and scope of the present invention, and all such modifications and equivalents are included in the scope of the claims.
Claims (7)
1. The hydraulic pump fault prediction method based on big data of the Internet of things and machine learning is characterized by comprising the following steps of:
s1, selecting characteristic parameters for carrying out fault prediction and a corresponding prediction model according to the type of the hydraulic pump fault to be predicted; the characteristic parameter includes a pump pressure; the prediction model is used for predicting pump pressure data of a preset time in the future according to the actual parameter data;
s2, installing corresponding sensors at corresponding positions of the equipment according to the selected characteristic parameters, and acquiring corresponding data of the selected characteristic parameters; after installing the corresponding sensor at the corresponding position of the equipment, establishing long connection between the sensor and the platform of the Internet of things;
s3, carrying out quality processing on the data acquired by the sensor according to the quality requirement of the parameter data to obtain operation data;
s4, carrying out preset pretreatment on the operation data to obtain clean data meeting training requirements;
s5, performing feature processing on clean data meeting training requirements according to the features of the data required by the selected prediction model to obtain training data meeting the model requirements; s5 comprises the following steps:
s501, screening clean data of equipment in a working state;
s502, performing characteristic transformation processing on the category type variable through single-heat coding;
s503, combining a working model of the equipment, and performing variable recombination operation on the clean data to obtain new parameter data;
s503, carrying out correlation analysis on each parameter in the clean data and the newly obtained parameter, and if the parameters with correlation exceeding 85% exist, only preserving one parameter; obtaining training data;
s6, training and optimizing the prediction model through training data to obtain a parameter prediction model; the pump pressure data output by the parameter prediction model is brought into a preset fault evaluation algorithm to perform fault prediction, and a set value in the fault evaluation algorithm is adjusted by combining a fault prediction result of the fault evaluation algorithm and an actual fault condition, so that the fault prediction accuracy of the fault evaluation algorithm meets a preset accuracy requirement; the preset fault evaluation algorithm is as follows:
if it isDetermining that the hydraulic pump has a fault; wherein N is the predicted data amount of the pump pressure on the same day; y is i Pump pressure data of the ith time of the day predicted by the parameter prediction model; y is r Is a preset pump pressure standard value; w is a set deviation value; threshold is set; the set value comprisesSetting a deviation value and a threshold value;
s7, inputting the collected real-time parameters into a parameter prediction model, and then taking the output pumping pressure data of the parameter prediction model into a fault evaluation algorithm with a well-regulated set value to predict pumping pressure faults of the equipment;
the parameter prediction model and the fault evaluation algorithm are both arranged on the Internet of things platform.
2. The hydraulic pump fault prediction method based on internet of things big data and machine learning as set forth in claim 1, wherein: in S2, the corresponding positions of the apparatus include node positions of the steering, radial clearance, axial clearance, oil contamination and/or oil delivery of the hydraulic pump.
3. The hydraulic pump fault prediction method based on internet of things big data and machine learning as set forth in claim 1, wherein: s3 comprises the following steps:
s301, processing discrete variable parameters to enable the discrete variable parameters to be changed into dummy variable parameters;
s302, maintaining and processing the dummy variable parameters and the continuity variable parameters according to preset quality requirements to obtain operation data; the quality requirements include integrity, normalization, consistency, accuracy, and relevance.
4. The hydraulic pump failure prediction method based on internet of things big data and machine learning as set forth in claim 3, wherein: s4 comprises the following steps:
s401, checking operation data, and converting checked parameters into a data format meeting preset requirements to obtain clean data;
s402, performing exception processing on clean data, marking the data exceeding the actual value range of the field as an abnormal value, and deleting the data;
s403, performing deletion processing on the clean data;
s404, performing protocol processing on continuous variable data in the clean data by adopting a Zscore standardization method, and performing protocol processing on discrete variable data in the clean data by adopting single-heat coding to obtain the clean data meeting training requirements.
5. The hydraulic pump fault prediction method based on internet of things big data and machine learning as set forth in claim 4, wherein: in S401, the checking includes: check whether the data type of each field accords with logic and whether the value range accords with reality.
6. The hydraulic pump fault prediction method based on internet of things big data and machine learning as set forth in claim 4, wherein: s403 includes:
s4031, deleting columns with a null value of > 70%;
s4032, deleting columns with only one value except the null value;
s4033, deleting rows with a null of > 70%;
s403, filling the missing value by using an upward filling method.
7. The hydraulic pump fault prediction method based on internet of things big data and machine learning as set forth in claim 1, wherein: and S7, if the pump pressure fault prediction result of the equipment is that the pump pressure fault will occur, sending out early warning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211371106.3A CN115907104B (en) | 2022-11-03 | 2022-11-03 | Hydraulic pump fault prediction method based on big data of Internet of things and machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211371106.3A CN115907104B (en) | 2022-11-03 | 2022-11-03 | Hydraulic pump fault prediction method based on big data of Internet of things and machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115907104A CN115907104A (en) | 2023-04-04 |
| CN115907104B true CN115907104B (en) | 2023-12-26 |
Family
ID=86475568
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211371106.3A Active CN115907104B (en) | 2022-11-03 | 2022-11-03 | Hydraulic pump fault prediction method based on big data of Internet of things and machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115907104B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109189659A (en) * | 2018-08-28 | 2019-01-11 | 郑州云海信息技术有限公司 | A kind of server failure prediction meanss and method |
| CN111241629A (en) * | 2020-01-08 | 2020-06-05 | 沈阳航空航天大学 | Intelligent prediction method for performance change trend of airplane hydraulic pump based on data driving |
| CN111765075A (en) * | 2020-05-20 | 2020-10-13 | 天津市天锻压力机有限公司 | Hydraulic forging press pump source fault prediction method and system |
| CN112943595A (en) * | 2021-02-07 | 2021-06-11 | 三一重工股份有限公司 | Hydraulic pump fault prediction method, hydraulic pump fault prediction device, electronic equipment and storage medium |
| CN114810572A (en) * | 2022-04-19 | 2022-07-29 | 鲁东大学 | Hydraulic pump fault diagnosis method and device |
-
2022
- 2022-11-03 CN CN202211371106.3A patent/CN115907104B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109189659A (en) * | 2018-08-28 | 2019-01-11 | 郑州云海信息技术有限公司 | A kind of server failure prediction meanss and method |
| CN111241629A (en) * | 2020-01-08 | 2020-06-05 | 沈阳航空航天大学 | Intelligent prediction method for performance change trend of airplane hydraulic pump based on data driving |
| CN111765075A (en) * | 2020-05-20 | 2020-10-13 | 天津市天锻压力机有限公司 | Hydraulic forging press pump source fault prediction method and system |
| CN112943595A (en) * | 2021-02-07 | 2021-06-11 | 三一重工股份有限公司 | Hydraulic pump fault prediction method, hydraulic pump fault prediction device, electronic equipment and storage medium |
| CN114810572A (en) * | 2022-04-19 | 2022-07-29 | 鲁东大学 | Hydraulic pump fault diagnosis method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115907104A (en) | 2023-04-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9921577B2 (en) | Method and system for diagnostic rules for heavy duty gas turbines | |
| CN110333074B (en) | Multi-measuring-point transmission mechanism fault diagnosis method and system based on convolutional neural network | |
| CN108491860A (en) | Shale gas pressure break equips Life cycle reliability in time dynamic assessment method and device | |
| CN112487910A (en) | Fault early warning method and system for nuclear turbine system | |
| CN110046409B (en) | ResNet-based steam turbine component health state evaluation method | |
| CN112598172A (en) | Wind turbine bearing temperature early warning method | |
| CN116670608A (en) | Hybrid ensemble method for predictive modeling of Internet of things | |
| CN115907104B (en) | Hydraulic pump fault prediction method based on big data of Internet of things and machine learning | |
| CN110554682B (en) | Fault detection reasoning method based on fault correlation analysis | |
| US20200293877A1 (en) | Interactive assistant | |
| CN116662920B (en) | Abnormal data identification method, system, equipment and medium for drilling and blasting method construction equipment | |
| CN110333694A (en) | Numerical control device method for diagnosing faults based on Fuzzy Petri Net | |
| CN105760672B (en) | A kind of Trouble Diagnostic Method of Machinery Equipment | |
| CN111842759B (en) | Heavy forging equipment remote operation and maintenance system architecture based on industrial internet platform | |
| CN117948319A (en) | Hydraulic system state prediction method, device, equipment and storage medium | |
| CN117094164A (en) | Sensor layout optimization method based on mixed information entropy | |
| CN113313365A (en) | Degradation early warning method and device for primary air fan | |
| Yacout | Logical analysis of maintenance and performance data of physical assets, ID34 | |
| CN118309644A (en) | Pipeline pump operation flow monitoring method and system based on digital twin | |
| CN117076886A (en) | Petrochemical pipe gallery data analysis and decision system and operation method thereof | |
| AU2022201697B2 (en) | Method and system of multi-objective and multi-dimensional online joint monitoring for nuclear turbine | |
| CN116955955A (en) | Pipeline defect model prediction method, system, terminal equipment and storage medium | |
| CN116628544A (en) | Fault diagnosis method based on hydraulic winch state on-line monitoring | |
| CN112326255B (en) | Engine remote health monitoring system and monitoring method | |
| CN113279812A (en) | Method and system for monitoring state and predicting residual life of mine main drainage equipment |
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 |