A method for abnormal recognition of lithium batteries based on multi-source signal reconstruction

Abstract

The present invention relates to the technical field of abnormal identification of lithium batteries, and discloses a method for abnormal identification of lithium batteries based on multi-source signal reconstruction. First, a BMS simulation system is built to obtain the multi-source signal curves of lithium battery charging and discharging and perform noise reduction processing; after obtaining multi-source signal feature data points through second-order sensitivity and Gaussian distribution weight selection and performing normalization processing, they are stored through differential compression; a multi-input multi-output LSTM model is constructed; the stored data is extracted and de-normalized to obtain a lithium battery feature data set, which is combined with the processed multi-source signal curves as the training samples of the LSTM model. The alpha evolution optimization algorithm is used to optimize the number of nodes in each layer, the time series length, and the initial learning rate in the LSTM hidden layer. Finally, a battery multi-source signal reconstruction model is obtained to reconstruct and predict the signals of the entire life cycle of the battery, realizing the identification of short-term battery abnormal states, the identification of long-term battery abnormal states, and the backtracking and analysis of the fault process.

Description

Technical Field

[0001] The present invention relates to the technical field of abnormal identification of lithium batteries, and specifically to a method for abnormal identification of lithium batteries based on multi-signal reconstruction. Background Art

[0002] During the operation of lithium batteries, various abnormal states and faults often occur. These abnormalities can lead to the attenuation of battery life, and in severe cases, may cause accidents such as fires. Electric vehicles use a battery management system (BMS) to collect various signals of the battery (such as voltage, current, and temperature) in real time, detect the abnormal state of the battery, and take measures. Battery state abnormalities and faults are often the result of the combined action of multiple complex factors, and the early signs are often not obvious, making it difficult to distinguish through instantaneous parameters. Therefore, historical data is of great significance for identifying the abnormal state of the battery and predicting potential faults.

[0003] However, limited by hardware costs and system complexity, the BMS usually comes with a small-capacity EEPROM memory, which cannot store all the battery historical data, easily resulting in the inability to capture early abnormal signs. This limitation has become a bottleneck in data utilization within the framework of existing technologies. With the in-depth research, pressure and strain, as signals that can characterize the internal state of the battery, have become new bases for battery abnormal determination. The introduction of these new signals further increases the burden of data collection and storage.

[0004] Most of the existing technologies use segments for battery state estimation, without giving the basis for selecting segments, and do not solve the problem of the small amount of data stored in the BMS. Moreover, most of the existing technologies focus on the use of voltage signals and do not involve signals such as temperature, pressure, and strain, having certain limitations. Therefore, how to optimize the data storage strategy, extract key multi-signals of the battery under limited hardware resources, restore the complete operating state through reconstruction technology when necessary, and perform abnormal identification of the battery is an urgent problem to be solved. Summary of the Invention

[0005] The purpose of the present invention is to solve the problem of how to optimize the data storage strategy, extract key multi-signals of the battery under limited hardware resources, restore the complete operating state through reconstruction technology when necessary, and perform abnormal identification of the battery, and to propose a method for abnormal identification of lithium batteries based on multi-signal reconstruction.

[0006] The technical solution of the present invention to solve the above technical problems is as follows:

[0007] A method for abnormal identification of lithium batteries based on multi-signal reconstruction includes the following steps:

[0008] S1. Build a BMS simulation system, obtain the multi-source signals of lithium battery charging and discharging, and perform noise reduction processing on the multi-source signals to obtain the processed multi-source signal curve;

[0009] S2. Select the multi-source signal feature data points in the processed multi-source signal curve by using second-order sensitivity and Gaussian distribution weight;

[0010] S3. After normalizing the multi-source signal feature data points, store them in the EEPROM memory of the BMS simulation system through differential compression;

[0011] S4. Construct a multi-input multi-output LSTM model;

[0012] S5. Extract the data from the EEPROM memory for inverse normalization processing to obtain the lithium battery feature data set. Combine it with the processed multi-source signal curve as the training sample of the LSTM model. Use the alpha evolution optimization algorithm to optimize the number of nodes in each layer, the time series length, and the initial learning rate in the LSTM hidden layer, and finally obtain the lithium battery multi-source signal reconstruction model;

[0013] S6. Use the lithium battery multi-source signal reconstruction model to reconstruct and predict the full life cycle signal of the lithium battery, and realize the identification of short-term lithium battery abnormal states, the identification of long-term lithium battery abnormal states, and the backtracking and analysis of the fault process.

[0014] Based on the above technical solutions, the present invention can also be improved as follows.

[0015] Preferably, the specific steps of S2 are as follows:

[0016] S2.1. Use to represent the multi-source signal, where x i is the sampling point position, f(x i ) is the value corresponding to the signal at the x i position, and then calculate the second-order sensitivity ω i of f(x);

[0017] S2.2. Use the Gaussian distribution function to weight the second-order sensitivity ω i to obtain W i ;

[0018] S2.3. Set the threshold T, and select the multi-source signal feature data points according to the judgment formula W i <T.

[0019] Preferably, the multi-source signal includes voltage signal, temperature signal, pressure signal and strain signal.

[0020] Preferably, the specific steps of S3 are as follows:

[0021] S3.1. Set identifiers for voltage signal, temperature signal, pressure signal, and strain signal respectively to represent the signal types;

[0022] S3.2. Normalize the multi - signal feature data points obtained in S2 respectively according to the signal types;

[0023] S3.3. Integrate the multi - signal feature data points together according to the timestamps, and store the data using the differential compression method.

[0024] Preferably, the specific steps of S5 are as follows:

[0025] S5.1. Extract the differentially compressed data in S3, perform denormalization processing, restore it to multi - signal feature data points, arrange them in the order of sampling timestamps, and form a lithium - battery feature data set;

[0026] S5.2. Take the multi - signal matrix S recorded within the timestamps in the lithium - battery feature data set as the input layer of the LSTM model. The matrix S is:

[0027] ;

[0028] where S vol ,S tem ,S for ,S str represent the feature data points of voltage, temperature, pressure, and strain signals respectively;

[0029] S5.3. Use the lithium - battery feature data set and the processed multi - signal curves to train the LSTM model, and use the alpha - evolution optimization algorithm to optimize the number of nodes in each layer, the length of the time series, and the initial learning rate in the LSTM hidden layer, and finally obtain a lithium - battery multi - signal reconstruction model.

[0030] Preferably, the calculation formula of the second - order sensitivity ω i in S2.1 is as follows:

[0031] ;

[0032] where f ,, (x i )is the second - order derivative of f(x i ).

[0033] Preferably, the weighting formula in S2.2 is as follows:

[0034] ;

[0035] where μ is the mean value of the second - order sensitivity ω i and σ is the second - order sensitivity ωi Standard deviation

[0036] Preferably, the specific steps of S2.3 are as follows:

[0037] Set a threshold T. According to Wi < T, select the multi-signal feature data points. The index idx of the multi-signal feature data points is:

[0038] ;

[0039] Respectively select the marked data points in the index from the multi-signal, and record the values and timestamps of these data points as the multi-signal feature data points.

[0040] Preferably, the difference formula in S3.3 is as follows:

[0041] ;

[0042] where Δy nor is the difference value, and y nor (k) and y nor (k - 1) are the normalized values of the k-th and (k - 1)-th feature data points respectively.

[0043] Preferably, the specific steps of integrating the multi-signal feature data points together according to the timestamp in S3.3 are as follows:

[0044] Integrate the feature data points of the voltage signal, temperature signal, pressure signal, and strain signal together according to the timestamp, sharing the same timestamp. For a certain type of signal that does not exist within the timestamp, use an empty character to replace it. The data structure for storage is:

[0045] ;

[0046] where t l is the timestamp, and v l is the multi-signal data set within the timestamp.

[0047] Compared with the prior art, the technical solution of the present application has the following beneficial technical effects:

[0048] 1) The present invention can accurately extract the data points highly relevant to the operating state of the lithium battery and store them, effectively alleviating the limitation of the BMS system in terms of storage resources, enabling it to store and utilize important historical data for a long time.

[0049] 2) The present invention can achieve efficient data utilization under the condition of limited BMS hardware resources, avoiding the introduction of large-capacity memories or high-performance computing units, and significantly reducing the development and operation costs of the system.

[0050] 3) The present invention predicts the operating state of a lithium battery throughout its life cycle through a multi-signal reconstruction model of the lithium battery. By comparison, it can accurately predict potential faults of the lithium battery, effectively reducing safety risks. Even if a small number of faults occur suddenly without prior signs, the method proposed by the present invention can still reconstruct the operating data of the faulty lithium battery to achieve the backtracking and analysis of the fault process. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0052] Figure 2 It is a diagram showing the selection result of voltage signal characteristic data points under constant current charge and discharge of the present invention;

[0053] Figure 3 It is a diagram showing the selection result of pressure signal characteristic data points under constant current charge and discharge of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0054] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present invention.

[0055] A method for identifying lithium battery anomalies based on multi-signal reconstruction includes the following steps:

[0056] S1. Build a BMS simulation system, obtain multi-signals of lithium battery charge and discharge, and perform noise reduction processing on the multi-signals to obtain a processed multi-signal curve;

[0057] The specific steps of S1 are as follows:

[0058] Under laboratory conditions, design a full-life cycle experiment of a lithium battery, adopt a constant current-constant voltage charging and constant current discharging strategy until the lithium battery reaches the end of its life. Build a BMS simulation system using an MCU chip and an EEPROM memory. The MCU chip integrates a voltage sensor, a thermocouple, a pressure thin film sensor, and a strain gauge sensor to collect the voltage, temperature, pressure, and strain signals of the lithium battery in real time. Then, perform noise reduction processing on the multi-signals to obtain a full-life cycle data set of the lithium battery;

[0059] S2. Obtain multi-signal characteristic data points in the processed multi-signal curve through second-order sensitivity and Gaussian distribution weight selection;

[0060] S3. After normalizing the multi-signal characteristic data points, store them in the EEPROM memory of the BMS simulation system through differential compression;

[0061] S4. Construct a multi-input multi-output LSTM model;

[0062] S5. Extract data from the EEPROM memory for denormalization processing to obtain a lithium battery feature dataset. Combine the processed multi-signal curve as the training sample of the LSTM model. Use the alpha evolution optimization algorithm to optimize the number of nodes in each layer, the time series length, and the initial learning rate in the LSTM hidden layer, and finally obtain a lithium battery multi-signal reconstruction model;

[0063] S6. Use the lithium battery multi-signal reconstruction model to reconstruct and predict the signals in the entire life cycle of the lithium battery, and realize the identification of short-term lithium battery abnormal states, the identification of long-term lithium battery abnormal states, and the backtracking and analysis of the fault process.

[0064] The specific steps of S6 are as follows:

[0065] S6.1 Use the lithium battery multi-signal reconstruction model to predict the operation data of the lithium battery in this cycle, and compare it with the operation data collected by the BMS in real time to realize the identification of short-term lithium battery abnormal states.

[0066] S6.2 Use the model to predict the multi-signal curve of the entire life cycle of the lithium battery, and compare it with the multi-signal curve reconstructed from the characteristic data stored in the BMS to realize the identification of long-term lithium battery abnormal states.

[0067] S6.3 If there is no obvious abnormality before the fault occurs, then reconstruct all the operation signals of the faulty lithium battery through the model to backtrack and analyze the fault.

[0068] The specific steps of S2 are as follows:

[0069] S2.1. Use to represent the multi-signal, where x i is the sampling point position, f(x i ) is the value corresponding to the signal at the x i position, and then calculate the second-order sensitivity ω i of f(x);

[0070] S2.2. Use the Gaussian distribution function to weight the second-order sensitivity ω i to obtain W i ;

[0071] S2.3. Set a threshold T, and select the multi-signal feature data points according to the judgment formula W i <T.

[0072] The multi-signal includes voltage signal, temperature signal, pressure signal and strain signal.

[0073] The specific steps of S3 are as follows:

[0074] S3.1. Set identifiers for the voltage signal, temperature signal, pressure signal, and strain signal respectively to represent the signal types;

[0075] S3.2. Normalize the multi-source signal feature data points obtained in S2 according to the signal types respectively;

[0076] S3.3. Integrate the multi-source signal feature data points together according to the timestamps, and store the data using the differential compression method.

[0077] The specific steps of S5 are as follows:

[0078] S5.1. Extract the differentially compressed data in S3, perform denormalization processing, restore it to multi-source signal feature data points, arrange them in the order of sampling timestamps, and form a lithium battery feature data set;

[0079] S5.2. Use the multi-source signal matrix S recorded within the timestamps in the lithium battery feature data set as the input layer of the LSTM model. The matrix S is:

[0080] ;

[0081] where S vol ,S tem ,S for ,S str represent the feature data points of the voltage, temperature, pressure, and strain signals respectively;

[0082] S5.3. Use the lithium battery feature data set and the processed multi-source signal curve to train the LSTM model, and use the alpha evolutionary optimization algorithm to optimize the number of nodes in each layer of the LSTM hidden layer, the time series length, and the initial learning rate, and finally obtain a lithium battery multi-source signal reconstruction model.

[0083] The specific steps for optimizing the number of nodes in each layer of the LSTM hidden layer, the time series length, and the initial learning rate using the alpha evolutionary optimization algorithm are as follows:

[0084] (1. Set the input parameters of the optimization algorithm, including the number of candidate solutions N, the dimension dim of the optimization parameters, the upper and lower bounds [ub, lb] of the variables, and the maximum number of iterations Imax (the candidate solutions correspond to the number of nodes in each layer of the LSTM hidden layer, the time series length, and the initial learning rate);

[0085] (2. Select the root mean square error (RMSE) as the fitness function. The optimization goal is to minimize the fitness. Generate an initial candidate solution matrix. The candidate solutions include three dimensions: the number of nodes in each layer of the LSTM hidden layer, the length of the time series, and the initial learning rate. Iteratively optimize during the LSTM training process until the maximum number of iterations is reached;

[0086] (3. Output the best combination of the number of nodes in each layer of the LSTM hidden layer, the length of the time series, and the initial learning rate, and the model training is completed.

[0087] The second-order sensitivity ω i in S2.1 is calculated as follows:

[0088] ;

[0089] where f ,, (x i ) is the second derivative of f(x i ).

[0090] The weighting formula in S2.2 is as follows:

[0091] ;

[0092] where μ is the mean of the second-order sensitivity ω i and σ is the standard deviation of the second-order sensitivity ω i ).

[0093] The specific steps of S2.3 are as follows:

[0094] Set a threshold T. According to Wi < T, select the multi-signal feature data points. The index idx of the multi-signal feature data points is:

[0095] ;

[0096] Select the marked data points in the index from the multi-signal respectively, and record the values and timestamps of these data points as the multi-signal feature data points.

[0097] The difference formula in S3.3 is as follows:

[0098] ;

[0099] where Δy nor is the difference value, and y nor (k) and y nor (k - 1) are the normalized values of the kth and (k - 1)th feature data points respectively.

[0100] The specific steps of integrating the multi-signal feature data points together by timestamp in S3.3 are as follows:

[0101] Integrate the characteristic data points of voltage signals, temperature signals, pressure signals, and strain signals according to the time stamp, sharing a single time stamp. For a certain type of signal that does not exist within the time stamp, use an empty character to replace it. The data structure for storage is as follows:

[0102] ;

[0103] where t l is the time stamp, and v l is the multi-signal data set within the time stamp.

[0104] Combined with Figure 1 and Figure 2 , using the method of combining second-order sensitivity with Gaussian distribution weights, more than 200 characteristic data points can be selected from the original data points with a length of more than 7000. When the BMS stores them, a large amount of space can be saved, and the trained LSTM can restore the original voltage and pressure curves based on these more than 200 data points for subsequent identification of abnormal lithium batteries.

[0105] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the existence of additional identical elements in the process, method, article or device comprising the element.

[0106] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for abnormal identification of lithium batteries based on multi-source signal reconstruction, characterized in that, Including the following steps: S1. Build a BMS simulation system, obtain multi-source signals of lithium battery charging and discharging, and perform noise reduction processing on the multi-source signals to obtain a processed multi-source signal curve; S2. Select multi-source signal feature data points in the processed multi-source signal curve through second-order sensitivity and Gaussian distribution weight selection; S3. After normalizing the multi-source signal feature data points, store them in the EEPROM memory of the BMS simulation system through differential compression; S4. Construct a multi-input multi-output LSTM model; S5. Extract data from the EEPROM memory for inverse normalization processing to obtain a lithium battery feature data set. Combine it with the processed multi-source signal curve as the training sample of the LSTM model. Use the alpha evolution optimization algorithm to optimize the number of nodes in each layer, the time series length, and the initial learning rate in the LSTM hidden layer, and finally obtain a lithium battery multi-source signal reconstruction model; S6. Use the lithium battery multi-source signal reconstruction model to reconstruct and predict the full life cycle signals of the lithium battery, and realize short-term lithium battery abnormal state recognition, long-term lithium battery abnormal state recognition, and backtracking and analysis of the fault process.

2. The abnormal identification method of lithium battery based on multi-signal reconstruction according to claim 1, wherein The specific steps of S2 are as follows: S2.

1. Use to represent a multi - variable signal, where x i is the sampling point position, and f(x i ) is the value corresponding to the position signal of x i . Then calculate the second - order sensitivity ω i of f(x); S2.

2. Use the Gaussian distribution function to weight the second-order sensitivity ω i to obtain W i ; S2.

3. Set a threshold T, and select multi-source signal feature data points according to the judgment formula W i < T 3. A method for identifying anomalies in lithium batteries based on multi-source signal reconstruction according to claim 1, characterized in that The multi-source signals include voltage signals, temperature signals, pressure signals, and strain signals.

4. A method for identifying anomalies in lithium batteries based on multi-source signal reconstruction according to claim 3, characterized in that The specific steps of S3 are as follows: S3.

1. Set identifiers for voltage signals, temperature signals, pressure signals, and strain signals respectively to represent signal types; S3.

2. Perform normalization processing on the multi-source signal feature data points obtained in S2 according to signal types respectively; S3.

3. Integrate the multi-source signal feature data points together according to time stamps, and store the data using the differential compression method.

5. The method for identifying lithium battery anomalies based on multi-source signal reconstruction according to claim 4, wherein, The specific steps of S5 are as follows: S5.

1. Extract the differentially compressed data in S3, perform inverse normalization processing, restore it to multi-source signal feature data points, and arrange them in the order of sampling time stamps to form a lithium battery feature data set; S5.

2. Use the multi-source signal matrix S recorded within the time stamp in the lithium battery feature data set as the input layer of the LSTM model. The matrix S is: ； Among them, S vol , S tem , S for , S str respectively represent the characteristic data points of voltage, temperature, pressure and strain signals; S5.

3. Use the lithium battery feature data set and the processed multi-source signal curve to train the LSTM model, and use the alpha evolution optimization algorithm to optimize the number of nodes in each layer, the time series length, and the initial learning rate in the LSTM hidden layer, and finally obtain a lithium battery multi-source signal reconstruction model.

6. The abnormal recognition method of lithium battery based on multi-source signal reconstruction according to claim 2, characterized in that, The second-order sensitivity ω in S2.1 i has the following calculation formula: ； Among them, f ,, (x i ) is the second derivative of f(x i ).

7. A method for abnormal identification of lithium batteries based on multi-signal reconstruction according to claim 2, characterized in that, The weighting formula in S2.2 is as follows: ； where μ is the mean of the second-order sensitivity ω i and σ is the standard deviation of the second-order sensitivity ω i .

8. The method for identifying lithium battery anomalies based on multi-source signal reconstruction according to claim 2, characterized in that, The specific steps of S2.3 are as follows: Set a threshold T. According to Wi < T, select multi-source signal feature data points. The index idx of the multi-source signal feature data points is: ； Select the marked data points in the index in the multi-source signals respectively, and record the values and time stamps of these data points as multi-source signal feature data points.

9. The abnormal identification method of lithium battery based on multi-signal reconstruction according to claim 4, characterized in that The differential formula in S3.3 is as follows: ； Among them, Δy nor is the difference value, and y nor (k) and y nor (k - 1) are the normalized values of the k-th and (k - 1)-th characteristic data points respectively.

Images