CN114879050A - Intelligent and rapid power battery service life testing method and system based on cloud edge cooperation - Google Patents
Intelligent and rapid power battery service life testing method and system based on cloud edge cooperation Download PDFInfo
- Publication number
- CN114879050A CN114879050A CN202210669341.2A CN202210669341A CN114879050A CN 114879050 A CN114879050 A CN 114879050A CN 202210669341 A CN202210669341 A CN 202210669341A CN 114879050 A CN114879050 A CN 114879050A
- Authority
- CN
- China
- Prior art keywords
- power battery
- capacity
- data
- cloud
- edge
- 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.)
- Pending
Links
- 238000012360 testing method Methods 0.000 title claims abstract description 106
- 238000012549 training Methods 0.000 claims abstract description 35
- 230000015556 catabolic process Effects 0.000 claims abstract description 27
- 238000006731 degradation reaction Methods 0.000 claims abstract description 27
- 230000006854 communication Effects 0.000 claims description 9
- 238000003062 neural network model Methods 0.000 claims description 8
- 238000004891 communication Methods 0.000 claims description 7
- 125000004122 cyclic group Chemical group 0.000 claims description 7
- 230000015654 memory Effects 0.000 claims description 4
- 230000001360 synchronised effect Effects 0.000 claims 1
- 239000012141 concentrate Substances 0.000 abstract 1
- 238000000034 method Methods 0.000 description 27
- 239000011159 matrix material Substances 0.000 description 8
- 240000007651 Rubus glaucus Species 0.000 description 5
- 235000011034 Rubus glaucus Nutrition 0.000 description 5
- 235000009122 Rubus idaeus Nutrition 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 5
- 238000009826 distribution Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 description 4
- 229910001416 lithium ion Inorganic materials 0.000 description 4
- 238000005070 sampling Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000005087 graphitization Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 230000014759 maintenance of location Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000009827 uniform distribution Methods 0.000 description 2
- CIWBSHSKHKDKBQ-JLAZNSOCSA-N Ascorbic acid Chemical compound OC[C@H](O)[C@H]1OC(=O)C(O)=C1O CIWBSHSKHKDKBQ-JLAZNSOCSA-N 0.000 description 1
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 206010011906 Death Diseases 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000007599 discharging Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 229910002804 graphite Inorganic materials 0.000 description 1
- 239000010439 graphite Substances 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000007773 negative electrode material Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011056 performance test Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/378—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC] specially adapted for the type of battery or accumulator
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
-
- 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
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
- Y02E60/10—Energy storage using batteries
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Secondary Cells (AREA)
Abstract
The invention discloses an intelligent and rapid power battery service life testing method and system based on cloud edge cooperation, which comprises the following steps: carrying out a cycle charge and discharge test on the power battery to be tested at the edge end to obtain battery capacity test time sequence data within set time; uploading the battery capacity test time series data to the cloud end to serve as training data, and performing retraining on the primarily trained power battery life prediction basic model; and according to the battery capacity testing time sequence data in the set time and the retrained power battery life prediction basic model, performing capacity prediction at the edge end to obtain the whole capacity degradation trend and the residual life of the power battery to be tested. The invention concentrates the task of training the life prediction basic model on the cloud end, and carries out the life prediction task on the edge side. The remaining service life of the battery can be predicted by using only partial capacity data.
Description
Technical Field
The invention relates to the technical field of power battery service life testing and prediction, in particular to a power battery service life intelligent rapid testing method and system based on cloud edge cooperation.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
The battery life refers to the number of charge and discharge times when the available capacity falls to 80% of the initial value. The battery is subjected to cycle life test, so that the characteristics of the battery can be further known, whether the battery reaches a design target or not is demonstrated, better management and control are realized in the process of using the battery, and the capability of meeting the requirements of different application scenes of the battery can be evaluated.
At present, the national standard GB/T31484-. In this standard, when a standard cycle life test of a battery is performed, the battery needs to be continuously charged and discharged to the end-of-life condition, i.e., "once-test-to-end". This test is often very time consuming. The cycle times of charging and discharging of the lithium ion battery can reach thousands of times or even more than ten thousand times, the testing time reaches 1 year or even longer, and the cycle life of the battery needs to pass the test of a full period, a long time and high cost. The new energy automobile industry requires that a power battery can be subjected to technical iteration and product upgrading quickly, but the existing method for quickly testing the cycle life of the battery is lacked, so that the quick and high-quality development of the new energy automobile industry is severely restricted. Therefore, how to shorten the life test time, test and evaluate the battery cycle life quickly, accurately and efficiently becomes an important means for breaking through the bottleneck of the key technology for the rapid development of power batteries and related industries.
A prediction-based method for rapidly testing battery life is an excellent solution to the above-mentioned problem, namely, "combination of estimation". By replacing the test with prediction, a large amount of time is saved, and the efficient and accurate evaluation of the cycle life of the battery can be realized. Many researchers have conducted extensive research, for example, in the prior art, a battery is subjected to short-term cycle performance tests with different cycle times, the cycle times and the capacity retention rate are recorded, then the battery after different cycle times is disassembled, the graphitization degree of the graphite negative electrode material is tested by using an XRD method, and prediction is conducted according to three data, namely the cycle times, the capacity retention rate and the graphitization degree. The method needs to disassemble and destroy the battery, and has certain limitation. The prior art discloses a method for predicting the cycle life of a lithium ion battery, which comprises the steps of installing a pressure sensor on the surface of the battery, recording capacity information of the battery within a certain cycle number, and fitting according to the cycle number, the discharge capacity and voltage data in the pressure sensor to realize the prediction of the battery life. The method needs an additional pressure sensor, actually increases the cost and has certain limitations. The prior art also discloses a method for rapidly predicting the cycle life of lithium ions, which is implemented by placing a battery under different multiplying powers to carry out 500 times of cycle tests, and fitting an equation to obtain a prediction equation to predict the life of the battery. The method does not need precise test equipment and complex theoretical calculation, but still needs long-time cycle test, and has low practicability.
The data-driven prediction does not need to disassemble the battery or consider the internal physical and chemical changes of the battery, corresponding analysis is directly carried out on the basis of test data to predict the service life of the battery, and the method is simple and rapid. For example, the prior art discloses a method for predicting the remaining life of a battery based on data driving, which includes the steps of recording data such as charging time, temperature and charging times, sampling the data to be used as training data of a neural network model, and predicting the data through a self-set data wrapper. The prior art discloses a service life prediction method based on combination of data driving and battery characteristics, which extracts characteristic factors and calculates SOH (sequence of events) under different states through operation data of a large number of different batteries, and performs service life prediction by establishing a battery and user portrait algorithm relation model. The prior art also discloses a method for predicting the service life of a power battery based on big data, wherein big data analysis relational expressions are obtained by analyzing big data of parameters of the battery under the conditions of different discharge depths, different temperatures, different discharge multiplying powers and the like, and the service life of the battery is predicted by using a throughput method. Although the above method can obtain the cycle life of the battery, the amount of data to be used is large, the implementation speed is slow, and the required cost is high.
Although much research is currently directed to life prediction, intelligent and rapid testing methods for battery life based on prediction are still missing at the present stage.
Disclosure of Invention
In order to solve the problems that the traditional testing method is long in period, high in cost and required to disassemble batteries, the invention provides an intelligent and rapid testing method and system for the service life of a power battery based on cloud-edge cooperation, and the intelligent and rapid testing for the service life of the power battery is realized by combining a cloud-edge cooperation technology and an 'testing and estimating combination' technology and replacing testing with prediction. And (4) concentrating the task of training the life prediction basic model at the cloud end, and performing the life prediction task at the edge side. The degradation capacity and the residual service life of the battery to be tested can be predicted only by using partial capacity data, and the requirement of intelligent and rapid test on the service life of the power battery is met.
In some embodiments, the following technical scheme is adopted:
a cloud edge cooperation-based intelligent and rapid power battery life testing method comprises the following steps:
carrying out a cycle charge and discharge test on the power battery to be tested at the edge end to obtain battery capacity test time sequence data within set time;
uploading the battery capacity test time series data to a cloud end to serve as training data, and performing retraining on the primarily trained power battery life prediction basic model;
and according to the battery capacity testing time sequence data in the set time and the retrained power battery life prediction basic model, performing capacity prediction at the edge end to obtain the whole capacity degradation trend and the residual life of the power battery to be tested.
In other embodiments, the following technical solutions are adopted:
a power battery life intelligent rapid test system based on cloud edge cooperation comprises:
the data acquisition module is used for performing a cyclic charge and discharge test on the power battery to be tested at the edge end to acquire battery capacity test time sequence data within set time;
the model training module is used for uploading the battery capacity test time series data to a cloud end to serve as training data, and retraining the primarily trained power battery life prediction basic model;
and the capacity degradation trend prediction module is used for predicting the capacity at the edge end according to the battery capacity test time sequence data in the set time and the retrained power battery life prediction basic model to obtain the whole capacity degradation trend and the residual life of the power battery to be tested.
In other embodiments, the following technical solutions are adopted:
a power battery life intelligent rapid test system based on cloud edge cooperation comprises:
the edge terminal equipment is used for performing a cyclic charge-discharge test on the power battery to be tested at the edge terminal to obtain battery capacity test time sequence data within set time;
moreover, according to the battery capacity testing time sequence data in the set time and the retrained power battery life prediction basic model, capacity prediction is carried out at the edge end to obtain the whole capacity degradation trend and the residual life of the power battery to be tested;
and the cloud equipment is used for training the power battery life prediction basic model based on the existing battery capacity data and the received battery capacity test time sequence data sent by the edge terminal.
Compared with the prior art, the invention has the beneficial effects that:
(1) the invention realizes the intelligent and rapid test of the service life of the power battery based on the prediction method, replaces the test with the prediction, combines the test and the estimation, saves a large amount of test time, and can realize the high-efficiency and accurate estimation of the cycle life of the battery.
(2) Based on a large amount of data of the existing battery, the task of training the life prediction basic model is concentrated in the cloud, and the life prediction task is carried out on the edge side; the cloud computing and the edge computing are reasonably combined, so that the data transmission delay is reduced, and the data security is enhanced.
(3) According to the invention, only partial data is tested at the beginning stage, and the residual service life of the battery can be predicted by using partial test data without disassembling the power battery, so that the test time is shortened, and the intelligent and rapid detection of the power battery is realized.
Additional features and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
Fig. 1 is a flowchart of an intelligent rapid testing method for service life of a power battery based on cloud-edge coordination in an embodiment of the invention;
fig. 2 is a cloud edge coordination-based power battery life intelligent rapid test system architecture in the embodiment of the present invention;
FIG. 3 is a schematic diagram of an LSTM structure in an embodiment of the present invention;
FIG. 4 is a capacity degradation curve of four batteries in an embodiment of the invention;
FIG. 5 is a schematic diagram of Sobol sequence sample generation in an embodiment of the present invention;
FIG. 6 is a graph of expanded degraded capacity in an embodiment of the present invention;
FIG. 7 shows the predicted results of training the base model using 8 sets of battery data according to an embodiment of the present invention.
Detailed Description
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
Example one
The essence of cloud computing is to provide ultra-strong and efficient computing power based on a large amount of centralized computing resources, to implement parallel computing, distributed computing, load balancing, and the like, and the high intensification thereof can bring about reduction of unit computing power cost. The cloud server used in the cloud computing has a large number of CPUs and GPUs, can receive battery test data collected by a local user side, extracts massive cycle life information and evaluates the continuity of the use of the cycle life information. However, as the curtain of the big data era of the power battery is pulled open, the transmission bandwidth and power of data, the storage space of the cloud server and the computing resources will increase sharply. The data that can be stored and processed by the cloud server is limited, and long-term high-capacity and high-speed data uploading cannot be supported.
The edge calculation is a distributed open platform which integrates the network, calculation, storage and other capabilities at the network edge side, and in the edge calculation, the service life test data of the battery can be directly stored and calculated at a user side instead of being uploaded to a cloud. The data transmission time and cost are greatly reduced, and meanwhile, compared with a traditional centralized deployment mode of cloud computing, better support is provided for services with high bandwidth and low time delay.
Cloud computing and edge computing are both long, cloud computing is good at large data processing and analysis with long period, non-real time and global property, and edge computing can play an important role in nearby real-time intelligent processing of user-side services. The cloud computing and the edge computing are reasonably combined, and intelligent and rapid testing of the service life of the power battery can be achieved.
Based on this, in one or more embodiments, a cloud-edge-coordination-based intelligent and rapid power battery life testing method is disclosed, which specifically includes the following processes in combination with fig. 1:
(1) carrying out a cycle charge and discharge test on the power battery to be tested at the edge end to obtain battery capacity test time sequence data within set time;
the method of the embodiment can complete the whole battery capacity degradation trend and the residual life prediction by only collecting partial data (about the data of the first 10 percent of the battery capacity) of the tested power battery.
(2) Uploading the battery capacity test time series data to a cloud end to serve as training data, and performing retraining on the primarily trained power battery life prediction basic model;
in the embodiment, the service life prediction basic model of the power battery adopts an LSTM model, and the LSTM records historical information and controls the influence degree of current time information on the historical information by introducing a memory control gate, so that the problems of long-term dependence, gradient disappearance and the like of the traditional RNN neural network are effectively solved. LSTM structurally contains three parts, respectively: an input gate, a forgetting gate and an output gate. A "gate" is a unique feature of the LSTM that has selective access to information. The three gates are mutually matched, so that the historical information is efficiently extracted and utilized for prediction. The basic structure of the network is shown in FIG. 3; wherein x is t 、h t-1 And C t-1 The input at the time t, the output at the time t-1 and the cell state output at the time t-1 are respectively; sigma is a sigmoid activation function, and tanh is a hyperbolic tangent activation function.
With t as a cycle period, the forgetting gate of LSTM can be represented by equation 1.
f t ＝sigmoid(W f ·[h t-1 ,x t ]+b f ) (1)
Wherein, W f And b f Weight matrix and offset vector of forgetting gate, [ h ] respectively t-1 ,x t ]Representing the input vector. W f Dimension of and h t-1 And x t In this regard, it can be expressed by formula (2).
Similarly, the input gate can be expressed as equation (3).
i t ＝sigmoid(W i ·[h t-1 ,x t ]+b i ) (3)
Wherein, W i And b i Are respectively inputsA weight matrix for the gate and an offset vector. The currently inputted cell state is calculated by the output at the previous time and the input this time, as shown in formula (4).
C′ t ＝tanh(W c ·[h t-1 ,x t ]+b c ) (4)
Wherein, W c And b c Respectively, the weight matrix and the offset vector of the currently input cell state. Further, by cell state C t-1 And calculating the unit state at the current moment according to the currently input unit state. Formula (5) gives C t The calculation formula of (2).
C t ＝f t ⊙C t-1 +i t ⊙C′ t (5)
New cell state C t From current unit C' t And a long-term memory cell C t-1 The two units are respectively controlled by a forgetting door and an input door. The forgetting gate has the capability of storing previous information, and the existence of the input gate can prevent the current irrelevant content from influencing the memory process. The output of the LSTM is determined by the output gate and the update status unit.
Wherein, W o And b o Respectively, a weight matrix and an offset vector for the output gates.
In this embodiment, the input of the LSTM neural network model is power battery capacity data at the first N times, and the output is power battery capacity data at the N +1 th time.
The training process of the LSTM neural network model is carried out at the cloud, firstly, a model training data set is constructed by utilizing the capacity data of a large number of existing power batteries of the same type at different time, and a power battery life prediction basic model is preliminarily trained. And then, further training the preliminarily trained model by using the received battery capacity test time sequence data within the set time uploaded by the edge terminal.
(3) And according to the battery capacity testing time sequence data in the set time and the retrained power battery life prediction basic model, performing capacity prediction at the edge end to obtain the whole capacity degradation trend and the residual life of the power battery to be tested.
Specifically, the edge end obtains an LSTM neural network model trained by the cloud end, and the LSTM neural network model inputs battery capacity testing time sequence data within a set time into the LSTM neural network model, so that the overall capacity degradation trend of the power battery to be tested can be obtained.
With reference to fig. 2, the overall architecture for implementing the method of the present embodiment is composed of an edge and a cloud. The edge end mainly comprises battery testing and edge computing equipment, is distributed at power battery stacks of different types and materials, is responsible for collecting, processing, uploading and the like of small-part capacity degradation and cycle life data of the battery, and sends the cycle life and capacity degradation results of the battery, state information of the battery testing equipment and the like to the cloud. The cloud end is mainly responsible for receiving historical data of the edge end, storing known cycle life data of a large number of batteries, training a battery life prediction basic model, visualizing a battery life prediction result and the like.
One cloud can deploy multiple power cell stack edge terminals. Herein, and subject to the constraints, the cloud is simulated by a dell tower workstation, which is basically configured as follows: intel to strong W-22458C 3.9GHz DDR4-2933 central processing unit; great RTX6000 display card. The edge is simulated by the raspberry group, and includes extension modules such as bluetooth and WIFI, in addition attaches 64 GB's RAM card and is used for storing procedure and backup data. The operating system is a customized system based on Linux, the version of the system is newer, and the support for the novel raspberry group is complete. The raspberry group obtains battery test information collected by the sensor through a serial port. In order to guarantee the possible Bluetooth expansion requirement of the system, the serial port is remapped to mini UART.
And the communication between the cloud end and the edge end is realized based on the FTP protocol. The multithreading processing capability of the cloud workstation enables the cloud workstation to be synchronously connected with a plurality of edge devices. In the coverage range of the communication network, after the edge-end equipment sends an application to the cloud end, the application is distributed to a designated port of the workstation, and the communication process is monitored by a unique independent thread. The communication content is written into a workstation for storage in a csv format, and the files are synchronously read in sequence by a preprocessing and capacity feature extraction program at the cloud end, so that a basic model for predicting the service life of the power battery is trained. The size of a file generated by the cloud in communication with the edge device is strictly limited, and when the size of the file exceeds the upper limit of a workstation memory, the early file is cleared so as to ensure that the available space of the cloud is sufficient.
The framework of the embodiment can realize the interactive communication between the uplink data of the edge end and the downlink data of the cloud end. Firstly, at the edge device, an Arbin battery testing device is used for carrying out a power battery cyclic charge-discharge test, and battery capacity testing historical data are uploaded to the cloud. Secondly, on a cloud workstation, preprocessing (data cleaning, data removing and the like) the obtained battery test data and the existing massive battery capacity and service life data of different models, and building and training a power battery service life prediction basic model. And finally, downloading a cloud-trained model on the edge raspberry, and quickly predicting the service life of the power battery through about 10% of the previous data.
As an optional implementation manner, based on the existing capacity data of the same type of power battery at different times, the data expansion is performed on the model training data set, specifically:
C gen ＝a×C 1 +b×C 2 +...+c×C N
wherein, C 1 -C N Respectively representing the degradation capacity data of different power batteries; a. b and c are corresponding capacity coefficients respectively; wherein the capacity coefficient is selected from samples generated based on the Sobol sequence.
The Sobol sequence is described below:
for all algorithms requiring sampling, uniformly distributed random numbers mean more excellent sample distribution. For the optimization problem of unknown solution distribution, the initial values of the individuals should be distributed uniformly in the data space as much as possible, so as to maintain the diversity of the population at a higher level. Thus, instead of requiring new generators to simulate a true uniform distribution, it is desirable that arbitrarily sized samples (especially small samples) meet low variability.
All modern CPU-based random number generation algorithms are pseudo-random (quasi-random). They are limited to one cycle. When the period is exceeded, the random numbers are repeated and are not independent of each other. The final limit of this period is determined by the number of bits in the computer, so that none of the built-in random numbers are "truly" random.
Sobol sampling uses different approaches to sampling. Sobol sequences emphasize over random numbers producing a uniform distribution in probability space. However, this is not simply filled using a grid, but rather a substantially random, but smart approach to "filling" the probability space, i.e., the random numbers generated later are distributed to areas that were not previously sampled. The principle of generation of Sobol is as follows.
(1) Radial interrogation operation
Φ b,C (i)＝(b -1 ...b -M )[C(a 0 (i)...a M-1 (i)) T ] (8)
In the above formula, b is a positive integer, if any integer i is expressed as a number in the b-ary system, then the obtained number is expressed by a bit a l (i) Forming a vector, multiplying the vector by the generator matrix C to obtain a new vector, and finally mirroring the new vector to the right of the decimal point to obtain another number in the range of [0,1), which is called a radial inversion operation and recorded as phi b,C (i)。
If the above process is simplified, let C be the identity matrix, the Van der Corput sequence can be obtained
(2) Generation of Sobol sequences
Each dimension of the Sobol sequence is composed of radial inversion with base number 2, but each radial inversion of each dimension has a different generator matrix C. Namely, it is
Because the base number is completely 2, the generation of the Sobol sequence can directly use bit operation to realize radiology conversion, and the method is very efficient. The distribution of the Sobol sequence is not only uniform, but also [0,1] when the number of samples is an integer power of 2 8 There will be one and only one point in each base interval at base 2 in the interval, which means that samples of high quality distribution can be generated. The generation matrix C may be obtained by searching related websites, and the generation of the Sobol sequence is realized by using a Matlab tool box. FIG. 5 is a sample schematic diagram of the generated Sobol sequence.
The present embodiment takes NASA data set as an example to verify the method of the present embodiment:
the NASA dataset is from the national aerospace agency emms prominent prediction center. The 18650 lithium ion battery was used in the experiment, and the charge and discharge experiment was carried out under different operating conditions, and impedance was used as the damage standard, and the experiment captured a large amount of capacity degradation and life data of the battery under different operating conditions. The capacity degradation curves are plotted, for the cells 5, 6, 7, 18 in the data set, as shown in fig. 4. As can be seen from fig. 4, the 4 batteries exhibited different life distributions, but the general trend of capacity degradation was similar.
Because the data of four batteries is relatively less, under the limited experimental conditions, the data expansion is carried out in the embodiment. And then inputting the life prediction model into the cloud for life prediction basic model training.
The data expansion method of the embodiment is as follows:
C gen ＝a×C B0005 +b×C B0006 +c×C B0007 (11)
wherein, C gen To expand the uncertainty battery degradation capacity data obtained, C B0005～B0007 Degraded capacity data of No. 5-No. 7 batteryAnd a, b and c are capacity coefficients respectively. In order to ensure that enough battery data can be obtained by expansion and have strong randomness, and because the primary batteries are the same batch of batteries, in order to ensure that the expanded capacity degradation data is approximately similar to the primary batteries, a Sobol sequence is adopted to generate [ -1,1]The parameters a, b are randomly chosen in the sample population, c is 1- (a + b), then equation (11) becomes
C gen ＝a×C B0005 +b×C B0006 +(1-a-b)×C B0007 (12)
When a and b belong to [ -1,1], four value taking conditions exist, and the range and the accepting or rejecting conditions of the parameters are given in table 1;
TABLE 1 ranges of parameters and trade-offs
As can be seen from the above table, when the ranges of a and b are [ -1,1], the numeric range of c is in four cases. When c is ∈ [ -1,0] or [0,1], the requirement is met; the values of c in the other two cases are greater than 1 and are undesirably discarded. Randomly select 5 sets of data from the satisfactory parameter set as capacity coefficients, as shown in table 2. The resulting expanded cell degraded capacity curve is shown in fig. 6.
TABLE 25 set of capacity coefficients
In this embodiment, all capacity data of 8 batteries are selected to perform cloud base model training, where 3 batteries are B0005, B0006, and B0007 batteries, and all data of 5 batteries are 5 sets of battery data C obtained by expansion according to equation (11) gen (i) When i is 1,2, … 5, the time step is 3, i.e. the next time is predicted by the capacity value of the previous 3 times. Inputting the data of the 8 batteries into a cloud-training LSTM life prediction basic model, issuing the model to a side-end raspberry group after the model training is finished, and predicting the life by using the data of about the first 10% of B0018。
The prediction results are shown in fig. 7. It can be seen from fig. 7 that after 8 sets of data are used for basic model training, accurate and rapid prediction of battery life can be achieved only by using about 10% of data, and test time is greatly reduced.
TABLE 3 prediction index comparison
Table 3 gives the prediction index comparisons. As can be seen from table 3, when the basic model is trained using 1 set of data, the error is large because the data size is small and the information related to the battery capacity degradation cannot be obtained by the LSTM network training, and the Root Mean Square Error (RMSE) and the Mean Absolute Error (MAE) are 0.062 and 0.049, respectively. Meanwhile, the predicted RUL value is 85, the difference with the actual value is 12 cycles, and the error is about 12 percent. When 3 groups of data are used for training the basic model, the data volume is expanded to a certain extent, so that the error is reduced to a certain extent, the RMSE and the MAE are respectively reduced to 0.039 and 0.030, and the RUL prediction error is reduced to 4.1%. The prediction error is further reduced when 8 sets of data are used to train the model. RMSE and MAE were reduced to 0.034 and 0.021, 45% and 57.1% compared to training with group 1 data. The RUL prediction error is reduced to 2.0% and reduced by 83.7%.
TABLE 4 comparison of Battery Life test times
Table 4 shows that the battery test time is greater than 300 days for the comparison of the service life test time, and this technique only uses about 10% of data to realize quick accurate prediction through the high in the clouds training model, and the test time is less than 15 days, and the time shortens by a wide margin.
Example two
In one or more embodiments, disclosed is a power battery life intelligent rapid test system based on cloud edge coordination, which specifically includes:
the data acquisition module is used for performing a cyclic charge and discharge test on the power battery to be tested at the edge end to acquire battery capacity test time sequence data within set time;
the model training module is used for uploading the battery capacity test time series data to a cloud end to serve as training data, and retraining the primarily trained power battery life prediction basic model;
and the capacity degradation trend prediction module is used for predicting the capacity at the edge end according to the battery capacity test time sequence data in the set time and the retrained power battery life prediction basic model to obtain the whole capacity degradation trend and the residual life of the power battery to be tested.
The specific implementation of the above modules has been described in detail in the first embodiment, and is not described in detail here.
EXAMPLE III
The utility model provides a quick test system of power battery life intelligence based on cloud limit is in coordination which characterized in that includes:
the edge terminal equipment is used for performing a cyclic charge-discharge test on the power battery to be tested at the edge terminal to obtain battery capacity test time sequence data within set time;
moreover, according to the battery capacity testing time sequence data in the set time and the retrained power battery life prediction basic model, capacity prediction is carried out at the edge end to obtain the whole capacity degradation trend and the residual life of the power battery to be tested;
and the cloud equipment is used for training the power battery life prediction basic model based on the existing battery capacity data and the received battery capacity test time sequence data sent by the edge terminal.
Although the embodiments of the present invention have been described with reference to the accompanying drawings, it is not intended to limit the scope of the present invention, and it should be understood by those skilled in the art that various modifications and variations can be made without inventive efforts by those skilled in the art based on the technical solution of the present invention.
Claims (9)
1. A cloud edge cooperation-based intelligent and rapid power battery life testing method comprises the following steps:
carrying out a cycle charge and discharge test on the power battery to be tested at the edge end to obtain battery capacity test time sequence data within set time;
uploading the battery capacity test time series data to a cloud end to serve as training data, and performing retraining on the primarily trained power battery life prediction basic model;
and according to the battery capacity testing time sequence data in the set time and the retrained power battery life prediction basic model, performing capacity prediction at the edge end to obtain the whole capacity degradation trend and the residual life of the power battery to be tested.
2. The intelligent and rapid power battery life testing method based on cloud edge coordination as claimed in claim 1, wherein the power battery life prediction basic model adopts an LSTM neural network model; the input of the LSTM neural network model is power battery capacity data of the first N moments, and the output is power battery capacity data of the (N + 1) th moment.
3. The intelligent and rapid power battery life testing method based on cloud-edge coordination as claimed in claim 1, wherein a model training data set is constructed by using capacity data of existing multiple power batteries of the same type at different times, and a power battery life prediction basic model is preliminarily trained.
4. The intelligent and rapid power battery life testing method based on cloud-edge coordination as claimed in claim 3, wherein based on capacity data of existing multiple power batteries of the same type at different times, data expansion is performed on a model training data set, specifically: multiplying the degradation capacity data of different power batteries by corresponding capacity coefficients respectively; wherein the capacity coefficient is selected from samples generated based on the Sobol sequence.
5. The intelligent and rapid power battery life testing method based on cloud-edge collaboration as claimed in claim 1, wherein the cloud end and the edge end are in mutual communication based on FTP protocol; the high in the clouds can many edge end equipment of synchronous connection.
6. The intelligent and rapid power battery life testing method based on cloud-edge coordination as claimed in claim 5, wherein after sending an application to the cloud, the edge-end device is allocated to a designated port of a workstation, and the communication process is monitored by a unique independent thread.
7. The intelligent and rapid power battery life testing method based on cloud-edge coordination as claimed in claim 1, wherein when the size of a file generated in the communication between the cloud and the edge device exceeds the upper limit of a workstation memory, the early stage file of the workstation is cleared.
8. The utility model provides a quick test system of power battery life intelligence based on cloud limit is in coordination which characterized in that includes:
the data acquisition module is used for performing a cyclic charge and discharge test on the power battery to be tested at the edge end to acquire battery capacity test time sequence data within set time;
the model training module is used for uploading the battery capacity test time series data to a cloud end to serve as training data, and retraining the primarily trained power battery life prediction basic model;
and the capacity degradation trend prediction module is used for predicting the capacity at the edge end according to the battery capacity test time sequence data in the set time and the retrained power battery life prediction basic model to obtain the whole capacity degradation trend and the residual life of the power battery to be tested.
9. The utility model provides a quick test system of power battery life intelligence based on cloud limit is in coordination which characterized in that includes:
the edge terminal equipment is used for performing a cyclic charge-discharge test on the power battery to be tested at the edge terminal to obtain battery capacity test time sequence data within set time;
moreover, according to the battery capacity testing time sequence data in the set time and the retrained power battery life prediction basic model, capacity prediction is carried out at the edge end to obtain the whole capacity degradation trend and the residual life of the power battery to be tested;
and the cloud equipment is used for training the power battery life prediction basic model based on the existing battery capacity data and the received battery capacity test time sequence data sent by the edge terminal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210669341.2A CN114879050A (en) | 2022-06-14 | 2022-06-14 | Intelligent and rapid power battery service life testing method and system based on cloud edge cooperation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210669341.2A CN114879050A (en) | 2022-06-14 | 2022-06-14 | Intelligent and rapid power battery service life testing method and system based on cloud edge cooperation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114879050A true CN114879050A (en) | 2022-08-09 |
Family
ID=82682349
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210669341.2A Pending CN114879050A (en) | 2022-06-14 | 2022-06-14 | Intelligent and rapid power battery service life testing method and system based on cloud edge cooperation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114879050A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116205379A (en) * | 2023-05-04 | 2023-06-02 | 上海叁零肆零科技有限公司 | Method, system, equipment and medium for establishing lithium-ion battery survival day prediction model |
Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160112518A1 (en) * | 2014-10-21 | 2016-04-21 | Skynet Phase 1, Inc. | Systems and methods for smart device networking |
| CN108279383A (en) * | 2017-11-30 | 2018-07-13 | 深圳市科列技术股份有限公司 | battery life predicting method, battery data server and battery data processing system |
| CN109001631A (en) * | 2018-04-27 | 2018-12-14 | 张家港莫特普数据科技有限公司 | Battery performance online evaluation system and method |
| CN109342949A (en) * | 2018-11-06 | 2019-02-15 | 长沙理工大学 | Lithium-ion-power cell remaining life on-line prediction method in charging process |
| CN109993270A (en) * | 2019-03-27 | 2019-07-09 | 东北大学 | Lithium ion battery residual life prediction technique based on grey wolf pack optimization LSTM network |
| CN110492186A (en) * | 2019-07-02 | 2019-11-22 | 北京航空航天大学 | A kind of management method of the power battery module based on cloud control technology |
| WO2020115761A1 (en) * | 2018-12-06 | 2020-06-11 | Sosaley Technologies Pvt. Ltd. | Cloud-based battery management system to predict battery life and battery health |
| CN111369329A (en) * | 2020-03-31 | 2020-07-03 | 中国科学院自动化研究所 | Electric vehicle battery management system based on block chain, leasing method and operation and maintenance method |
| CN111999648A (en) * | 2020-08-20 | 2020-11-27 | 浙江工业大学 | Lithium battery residual life prediction method based on long-term and short-term memory network |
| CN112711892A (en) * | 2020-12-07 | 2021-04-27 | 深圳先进技术研究院 | Cloud battery management system and method based on digital twin and block chain technology |
| CN112732443A (en) * | 2021-01-12 | 2021-04-30 | 徐州普罗顿氢能储能产业研究院有限公司 | Energy storage power station state evaluation and operation optimization system based on edge calculation |
| CN113536670A (en) * | 2021-07-09 | 2021-10-22 | 北京航空航天大学 | Federal transfer learning method for predicting remaining life of group products |
| CN113536671A (en) * | 2021-07-09 | 2021-10-22 | 北京航空航天大学 | Lithium battery life prediction method based on LSTM |
| CN113673115A (en) * | 2021-08-31 | 2021-11-19 | 北京航空航天大学 | Lithium battery cycle life prediction and reliability evaluation method based on digital twinning |
| CN113821875A (en) * | 2021-09-26 | 2021-12-21 | 湖南大学 | Intelligent vehicle fault real-time prediction method and system based on end cloud cooperation |
| CN114066100A (en) * | 2021-12-01 | 2022-02-18 | 深圳市新威尔电子有限公司 | Online battery cycle life prediction system and method based on federal learning |
-
2022
- 2022-06-14 CN CN202210669341.2A patent/CN114879050A/en active Pending
Patent Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160112518A1 (en) * | 2014-10-21 | 2016-04-21 | Skynet Phase 1, Inc. | Systems and methods for smart device networking |
| CN108279383A (en) * | 2017-11-30 | 2018-07-13 | 深圳市科列技术股份有限公司 | battery life predicting method, battery data server and battery data processing system |
| CN109001631A (en) * | 2018-04-27 | 2018-12-14 | 张家港莫特普数据科技有限公司 | Battery performance online evaluation system and method |
| CN109342949A (en) * | 2018-11-06 | 2019-02-15 | 长沙理工大学 | Lithium-ion-power cell remaining life on-line prediction method in charging process |
| WO2020115761A1 (en) * | 2018-12-06 | 2020-06-11 | Sosaley Technologies Pvt. Ltd. | Cloud-based battery management system to predict battery life and battery health |
| CN109993270A (en) * | 2019-03-27 | 2019-07-09 | 东北大学 | Lithium ion battery residual life prediction technique based on grey wolf pack optimization LSTM network |
| CN110492186A (en) * | 2019-07-02 | 2019-11-22 | 北京航空航天大学 | A kind of management method of the power battery module based on cloud control technology |
| CN111369329A (en) * | 2020-03-31 | 2020-07-03 | 中国科学院自动化研究所 | Electric vehicle battery management system based on block chain, leasing method and operation and maintenance method |
| CN111999648A (en) * | 2020-08-20 | 2020-11-27 | 浙江工业大学 | Lithium battery residual life prediction method based on long-term and short-term memory network |
| CN112711892A (en) * | 2020-12-07 | 2021-04-27 | 深圳先进技术研究院 | Cloud battery management system and method based on digital twin and block chain technology |
| CN112732443A (en) * | 2021-01-12 | 2021-04-30 | 徐州普罗顿氢能储能产业研究院有限公司 | Energy storage power station state evaluation and operation optimization system based on edge calculation |
| CN113536670A (en) * | 2021-07-09 | 2021-10-22 | 北京航空航天大学 | Federal transfer learning method for predicting remaining life of group products |
| CN113536671A (en) * | 2021-07-09 | 2021-10-22 | 北京航空航天大学 | Lithium battery life prediction method based on LSTM |
| CN113673115A (en) * | 2021-08-31 | 2021-11-19 | 北京航空航天大学 | Lithium battery cycle life prediction and reliability evaluation method based on digital twinning |
| CN113821875A (en) * | 2021-09-26 | 2021-12-21 | 湖南大学 | Intelligent vehicle fault real-time prediction method and system based on end cloud cooperation |
| CN114066100A (en) * | 2021-12-01 | 2022-02-18 | 深圳市新威尔电子有限公司 | Online battery cycle life prediction system and method based on federal learning |
Non-Patent Citations (3)
| Title |
|---|
| 屈力刚: "基于Sobol序列采样点分布策略的研究与应用", 《锻压装备与制造技术》, 6 January 2020 (2020-01-06), pages 1 - 5 * |
| 左伟;冯金富;张佳强;: "制导弹药允许发射区参数模型设计", 兵工学报, no. 05, 15 May 2011 (2011-05-15) * |
| 成庶;甘沁洁;赵明;毕福亮;王家捷;王国良;于天剑;: "镉镍蓄电池寿命预测的PF-LSTM建模方法研究", 铁道科学与工程学报, no. 07, 15 July 2020 (2020-07-15) * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116205379A (en) * | 2023-05-04 | 2023-06-02 | 上海叁零肆零科技有限公司 | Method, system, equipment and medium for establishing lithium-ion battery survival day prediction model |
| CN116205379B (en) * | 2023-05-04 | 2023-07-07 | 上海叁零肆零科技有限公司 | Method, system, equipment and medium for establishing lithium-ion battery survival day prediction model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | A deep learning method for lithium-ion battery remaining useful life prediction based on sparse segment data via cloud computing system | |
| CN111584952B (en) | Method and system for online evaluation of electrochemical cells of energy storage power station | |
| CN110752410B (en) | Method for rapidly sorting and recombining retired lithium battery | |
| CN113777496B (en) | Lithium ion battery residual life prediction method based on time convolution neural network | |
| Zhou et al. | Battery health prognosis using improved temporal convolutional network modeling | |
| Haris et al. | Degradation curve prediction of lithium-ion batteries based on knee point detection algorithm and convolutional neural network | |
| Li et al. | An intelligent transient stability assessment framework with continual learning ability | |
| CN113484787B (en) | Electrochemical impedance spectrum-based rapid sorting and recombining method for retired lithium ion battery | |
| CN111880099B (en) | Method and system for predicting service life of battery monomer in energy storage power station | |
| Zheng et al. | State of health estimation for lithium battery random charging process based on CNN-GRU method | |
| CN112666480B (en) | Battery life prediction method based on characteristic attention of charging process | |
| CN111707956B (en) | Method for predicting health state and residual life of multi-type lithium ion battery pack management system | |
| CN108009585B (en) | Lead-acid battery health state prediction method based on local information fusion | |
| Tian et al. | Data‐driven battery degradation prediction: Forecasting voltage‐capacity curves using one‐cycle data | |
| CN113361692A (en) | Lithium battery residual life combined prediction method | |
| CN112666479A (en) | Battery service life prediction method based on charging cycle fusion | |
| Zheng et al. | A novel method for lithium-ion battery remaining useful life prediction using time window and gradient boosting decision trees | |
| Lyu et al. | A semiparametric clustering method for the screening of retired Li-ion batteries from electric vehicles | |
| CN116736172A (en) | Lithium battery pack health state evaluation method based on generation type countermeasure network | |
| CN114879050A (en) | Intelligent and rapid power battery service life testing method and system based on cloud edge cooperation | |
| Zhu et al. | A novel fast estimation and regroup method of retired lithium‐ion battery cells | |
| Hassanieh et al. | Orthogonal autoencoders for long-term state-of-charge forecasting of Li-ion battery cells | |
| Li et al. | Big data driven Lithium-ion battery modeling method: a Cyber-Physical System approach | |
| Zhang et al. | Battery health state prediction based on lightweight neural networks: A review | |
| Zhao et al. | Capacity prediction and remaining useful life diagnosis of lithium-ion batteries using CNN-LSTM hybrid neural network |
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 |