A fault diagnosis method and device for a random working condition battery pack

Abstract

The present invention relates to a fault diagnosis method and device for a battery pack under random operating conditions. The method includes: performing residual calculation and feature extraction based on the obtained state parameter prediction matrix and actual observation parameter matrix of the battery cells to obtain a characteristic matrix of the battery cells; performing dimensionality reduction processing on the characteristic matrix of the battery cells using the principal component analysis method to obtain a comprehensive evaluation index of the battery cells; comparing the comprehensive evaluation index of each battery cell with a preset threshold, and determining the battery cells whose comprehensive evaluation index exceeds the preset threshold as faulty battery cells; constructing a fault type matrix of the faulty battery cells based on the characteristic matrix corresponding to the faulty battery cells and the initial state matrix of the battery cells; inputting the fault type matrix of the faulty battery cells into a trained fault type probability prediction model to obtain the trigger probability corresponding to the fault type of the faulty battery cells, so as to perform fault diagnosis on the battery pack. This solution can achieve accurate diagnosis of the fault type of the battery pack.

Description

Technical Field

[0001] This invention relates to the field of power battery safety monitoring technology, and in particular to a fault diagnosis method and device for battery packs under random operating conditions. Background Technology

[0002] Power battery systems composed of lithium-ion battery packs have advantages such as high specific energy, long lifespan, low self-discharge, and wide operating temperature range, making them the mainstream choice for electric vehicles. However, as the market share of electric vehicles increases, battery module safety failures have become a major bottleneck restricting their development. In recent years, frequent battery failures have led to countless recalls, increasing user safety anxiety and the economic burden on the market.

[0003] Research on lithium-ion battery fault diagnosis often focuses on analyzing and monitoring objects for specific fault types, and then designing corresponding diagnostic methods. Previous research on lithium-ion battery fault diagnosis mainly detected external short circuits, internal short circuits, electrolyte leakage, abnormal aging, and thermal runaway. Monitoring objects primarily included voltage, temperature, current, gas production, and battery appearance. Due to the limited availability of macroscopic characterization data, different degradation phenomena may exhibit similar macroscopic characteristics, most typically temperature rise and voltage drop, making accurate diagnosis of the battery pack's fault type impossible.

[0004] Therefore, there is an urgent need to provide a fault diagnosis method and device for battery packs under random operating conditions. Summary of the Invention

[0005] To address the problem that traditional fault diagnosis methods cannot accurately diagnose the fault types of battery packs, this invention provides a fault diagnosis method and apparatus for battery packs operating under random conditions.

[0006] In a first aspect, embodiments of the present invention provide a fault diagnosis method for a battery pack under random operating conditions, the method comprising:

[0007] The residual matrix of the battery cell is obtained by calculating the residual based on the predicted state parameter matrix and the actual observed parameter matrix of the battery cell; wherein, the predicted state parameter matrix of the battery cell is obtained by prediction using a pre-trained deep learning network;

[0008] The residual matrix of the battery cell is subjected to feature calculation to obtain the feature matrix of the battery cell;

[0009] Principal component analysis is used to reduce the dimensionality of the feature matrix of the battery cell to obtain the comprehensive evaluation index of the battery cell.

[0010] The comprehensive evaluation index of each battery cell is compared with a preset threshold, and the battery cells whose comprehensive evaluation index exceeds the preset threshold are identified as faulty battery cells.

[0011] Based on the feature matrix corresponding to the faulty battery cell and the initial state matrix of the battery cell, a fault type matrix of the faulty battery cell is constructed.

[0012] The fault type matrix of the faulty battery cell is input into the trained fault type probability prediction model to obtain the trigger probability corresponding to the fault type of the faulty battery cell, so as to diagnose the fault in the battery pack.

[0013] Secondly, embodiments of the present invention also provide a fault diagnosis device for a battery pack under random operating conditions, the device comprising:

[0014] The first computing unit is used to perform residual calculation based on the obtained state parameter prediction matrix and actual observation parameter matrix of the battery cell to obtain the residual matrix of the battery cell; wherein, the state parameter prediction matrix of the battery cell is obtained by prediction using a pre-trained deep learning network;

[0015] The second calculation unit is used to perform feature calculation on the residual matrix of the battery cell to obtain the feature matrix of the battery cell.

[0016] The third calculation unit is used to perform dimensionality reduction on the feature matrix of the battery cell using principal component analysis to obtain the comprehensive evaluation index of the battery cell.

[0017] The fault determination unit is used to compare the comprehensive evaluation index of each battery cell with a preset threshold, and to determine the battery cell whose comprehensive evaluation index exceeds the preset threshold as a faulty battery cell.

[0018] The fault type matrix construction unit is used to construct the fault type matrix of the faulty battery cell based on the feature matrix corresponding to the faulty battery cell and the initial state matrix of the battery cell.

[0019] The fault diagnosis unit is used to input the fault type matrix of the faulty battery cell into the trained fault type probability prediction model to obtain the trigger probability corresponding to the fault type of the faulty battery cell, so as to perform fault diagnosis on the battery pack.

[0020] Thirdly, embodiments of the present invention also provide a computing device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the method described in any embodiment of this specification.

[0021] Fourthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the methods described in any embodiment of this specification.

[0022] Fifthly, embodiments of the present invention also provide a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described above.

[0023] This invention provides a method and apparatus for fault diagnosis of battery packs under random operating conditions. After predicting the state parameter prediction matrix of individual battery cells using a pre-trained deep learning network, the method first performs residual and feature extraction with the actual observed parameter matrix to obtain a feature matrix composed of several parameters affecting the fault state of the battery pack. Then, principal component analysis is used to further filter the influencing parameters in the feature matrix, and the influence of each filtered parameter on the individual battery cells is further quantified. Based on the quantified indicators, the faulty battery cells are accurately located. Finally, the probability of the fault type corresponding to the faulty battery cell is predicted based on a trained fault type probability prediction model. Thus, this invention combines deep learning networks and massive cloud data for training, enabling probability prediction of fault types in battery packs under random operating conditions, thereby facilitating accurate diagnosis of fault types in battery packs under random operating conditions. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a flowchart of a fault diagnosis method for a battery pack under random operating conditions provided by an embodiment of the present invention;

[0026] Figure 2 This is a hardware architecture diagram of a computing device provided in an embodiment of the present invention;

[0027] Figure 3 This is a structural diagram of a fault diagnosis device for a battery pack under random operating conditions provided in an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0029] The specific implementation of the above concept is described below.

[0030] Please refer to Figure 1 This invention provides a fault diagnosis method for battery packs under random operating conditions, the method comprising:

[0031] Step 100: Based on the obtained state parameter prediction matrix and actual observation parameter matrix of the battery cell, residual calculation is performed to obtain the residual matrix of the battery cell; wherein, the state parameter prediction matrix of the battery cell is obtained by prediction using a pre-trained deep learning network;

[0032] Step 102: Perform feature calculation on the residual matrix of the battery cell to obtain the feature matrix of the battery cell;

[0033] Step 104: Use principal component analysis to reduce the dimensionality of the feature matrix of the battery cell to obtain the comprehensive evaluation index of the battery cell.

[0034] Step 106: Compare the comprehensive evaluation index of each battery cell with a preset threshold, and identify battery cells whose comprehensive evaluation index exceeds the preset threshold as faulty battery cells.

[0035] Step 108: Construct a fault type matrix for the faulty battery cell based on the feature matrix corresponding to the faulty battery cell and the initial state matrix of the battery cell.

[0036] Step 110: Input the fault type matrix of the faulty battery cell into the trained fault type probability prediction network model to obtain the trigger probability corresponding to the fault type of the faulty battery cell, so as to perform fault diagnosis on the battery pack.

[0037] In this embodiment of the invention, after predicting the state parameter prediction matrix of a battery cell using a pre-trained deep learning network, the residual and feature extraction are first performed between the predicted matrix and the actual observed parameter matrix to obtain a feature matrix composed of several parameters affecting the fault state of the battery pack. Then, principal component analysis is used to further filter the several influencing parameters in the feature matrix, and the influence of each filtered influencing parameter on the battery cell is further quantified. Thus, the faulty battery cell is accurately located based on the quantified indicators. Finally, the probability corresponding to the fault type of the faulty battery cell is predicted based on a trained fault type probability prediction model. In this way, by combining deep learning networks and massive cloud data for training, this embodiment of the invention can realize the probability prediction of battery pack fault types under random operating conditions, which is beneficial to the accurate diagnosis of battery pack fault types under random operating conditions.

[0038] For step 100:

[0039] In this embodiment of the invention, the state prediction matrix and the actual observation parameter matrix of the battery cell are both composed of the voltage value and state of charge value of the battery cell within a sampling period.

[0040] In some implementations, in step 100, the deep learning network includes a filter, a temporal network prediction model, and a physical augmentation encoder connected in sequence.

[0041] The state parameter prediction matrix of the battery cell is predicted in the following way:

[0042] For each current moment, execute:

[0043] Step S1: Based on the voltage parameters of the individual battery cells obtained at the current moment, construct the battery pack state model and the individual battery cell deviation model.

[0044] Step S2: Use filters to predict the state of the battery pack and the individual battery cell deviation model at the current moment, respectively, to obtain the filter prediction results of the battery pack and the individual battery cells at the next moment; wherein, the filter prediction results include the voltage and state of charge of the battery pack and the individual battery cells at the next moment.

[0045] Step S3: Input the battery state parameters and vehicle state parameters obtained at the current moment into the trained temporal network prediction model to obtain the temporal network prediction result of the battery pack at the next moment; wherein, during the training process of the temporal network prediction model, the filter prediction result of the battery pack is used as a loss function to adjust the temporal network prediction result in reverse;

[0046] Step S4: Input the time-series network prediction results of the battery pack and the filter prediction results of the individual battery cells into the trained physical augmentation encoder to obtain the state parameter prediction matrix of the individual battery cells.

[0047] In this embodiment of the invention, a battery pack state model and a battery cell deviation model are first constructed based on the voltage parameters of individual battery cells obtained from a cloud data platform. Then, filters are used to predict the states of the battery pack and individual battery cells respectively, thereby obtaining prediction results with good stability. Next, a pre-trained time-series network prediction model is used to make a secondary prediction of the battery pack state. In the training of the time-series network prediction model, the prediction results of the filter are used to adjust the prediction results of the time-series network model in reverse, thereby enhancing the accuracy of the prediction results. Finally, a pre-trained physical augmentation encoder is used to further compress and decompress the prediction results of the filter and the prediction results of the time-series network model, thereby obtaining battery cell prediction results with both good stability and accuracy.

[0048] Thus, in this embodiment of the invention, filters, a temporal network prediction model, and a physical enhancement encoder are used in sequence to predict the state of the battery pack, and massive cloud data is used as a training sample set, so that the prediction results can take into account both robustness and accuracy, thereby facilitating the accurate diagnosis of the fault state of the battery pack under random operating conditions.

[0049] Regarding step S1:

[0050] In some implementations, the process of constructing the battery pack state model and the individual cell deviation model also includes:

[0051] The inverse Gaussian distribution is used as the probability distribution of the battery parameters of the battery cell to calculate the voltage distribution probability of the battery cell.

[0052] Based on the voltage distribution probability of the individual battery cells, the current voltage of the battery pack and the deviation voltage of the individual battery cells are obtained.

[0053] Based on the battery pack voltage and the individual cell deviation voltage, a battery pack state model and an individual cell deviation model are constructed.

[0054] In this embodiment of the invention, due to differences in the production batches of battery cells or the mutual influence between cells in a local circuit, the distribution of battery cell voltage often exhibits a skewed pattern. Therefore, in this embodiment of the invention, an inverse Gaussian distribution is selected as the probability distribution fitting for the battery cell voltage parameters. This allows for a detailed description of the distribution characteristics of battery cells without increasing the number of parameters to be solved.

[0055] In some preferred embodiments, the voltage distribution probability of a single battery cell is calculated using the following formula:

[0056] f(U i )=[λ / (2πU i 3 )] 1 / 2 *exp[-λ(U i -μ) 2 / (2μ 2 U i )]

[0057] In the formula, μ=U m λ=μ 3 / Var(U i ), U i U is the voltage of the battery cell. m The average voltage of the battery cell;

[0058] The voltage of the battery pack is corrected based on the obtained voltage distribution probability of the individual battery cells, thereby obtaining the battery pack voltage and the individual battery cell deviation voltage. In some preferred embodiments, the battery pack voltage and the individual battery cell deviation voltage are calculated using the following formulas:

[0059] U T，p =(f(U) i )*U i ) / ∑f(U i )

[0060] ΔU i =U i -U T，p

[0061] In the formula, U T，p The battery pack voltage, f(U) i ) represents the voltage distribution probability of a single battery cell, ΔU i The deviation voltage of the battery cell.

[0062] In some implementations, the battery pack state model and the individual cell deviation state model are respectively as follows:

[0063]

[0064] ΔU i =ΔU OCV,i (ΔSOC i )-ΔR i I

[0065] In the formula, U T,p R is the terminal voltage of the battery pack state model. P,p C is the polarization internal resistance of the battery pack. P,p U is the polarization capacitor of the battery pack. P,p R is the polarization voltage of the battery pack.0,p Let I be the ohmic resistance of the battery pack, and U be the instantaneous current. OCV，p ΔU is the open-circuit voltage of the battery pack. i Let ΔU be the terminal voltage of the single-cell deviation state model. OCV，i ΔSOC is the open-circuit voltage of the battery cell. i The state of charge of the battery cell is ΔR. i Ω represents the ohmic internal resistance of the battery cell.

[0066] Because the working circuit of the battery pack is extremely complex in actual vehicle operation and there are many factors that cause changes in the battery pack state, this embodiment uses the battery pack voltage and the individual battery cell deviation obtained after distributed optimization as model output parameters to construct the above-mentioned battery pack and individual battery cell deviation state model. This model can serve as an equivalent model of complex battery circuits to dynamically describe the dynamic process of the battery pack system. Based on the above state model, a filter is further combined to achieve online closed-loop prediction of the battery pack and individual battery cell states.

[0067] Regarding step S2:

[0068] In some preferred embodiments, step S2 includes:

[0069] The coefficients of the battery pack state model and the battery cell deviation model at the current time are identified using the least multiplier method, so as to obtain the state parameters of the battery pack state model and the state parameters of the battery cell deviation model at the current time.

[0070] The state parameters of the battery pack state model and the state parameters of the battery cell deviation model at the current moment are respectively input into the filter to obtain the first prediction result of the battery pack at the next moment and the first prediction result of the battery cell at the next moment.

[0071] In this embodiment of the invention, it can be seen from the battery pack state model and the battery cell deviation state model that the state parameters of the battery pack state model include the polarization internal resistance, polarization capacitance, and ohmic resistance of the battery pack, while the state parameters of the battery cell deviation model include the ohmic internal resistance of the battery cell. In order to facilitate the identification of the state parameters of the battery pack state model and the battery cell deviation model, it is necessary to preprocess the battery pack state model and the battery cell deviation model.

[0072] Regarding step S3:

[0073] The method of predicting the state of the battery pack and individual cells by using a filter combined with a state model in step S2 has good stability, but the accuracy of its prediction results is relatively poor. Furthermore, in the state prediction results of the battery pack and individual cells obtained in step S3, the state of the battery pack has a greater impact on the entire battery pack. In this embodiment of the invention, a trained time-series network model is used to further predict the battery pack state prediction results output from the filter. This can enhance the accuracy of the battery pack system prediction, thereby making the prediction results both robust and accurate.

[0074] In step S3, the temporal network prediction model is trained in the following manner:

[0075] Construct a pre-defined temporal network learning model;

[0076] The hyperparameters and loss function of the temporal network learning model are set; wherein the loss function is as follows:

[0077]

[0078] In the formula, loss1 is the loss function, and t is the sampling time. The prediction results of the temporal network learning model. For the first prediction result of the battery pack, U T，p，k+1 Let x be the terminal voltage of the battery pack at time k+1. T，p，k+1 The state of charge of the battery pack at time k+1;

[0079] The known vehicle state parameters and battery state parameters are used as the input sample set, and the known battery pack voltage and state of charge parameters are used as the output sample set. The Adam algorithm is used to train the time-series network learning model. The vehicle state parameters include vehicle speed, vehicle mileage and vehicle operating status, and the battery state parameters include battery pack total voltage, current, battery pack average temperature and battery pack state of charge.

[0080] During training, the loss and accuracy of the temporal network learning model are calculated, and the network parameters of the temporal network learning model are iterated until the model converges or reaches the preset number of iterations, thus obtaining the temporal network prediction model.

[0081] Regarding step S4:

[0082] In some preferred embodiments, the physical enhancement encoder in step S4 includes an encoding module, a latent space sampling module, and a decoding module connected in sequence; wherein:

[0083] The encoding module is used to perform feature fusion and dimensionality reduction operations on the first prediction result of the input battery cell and the second prediction result of the battery pack to obtain latent variables;

[0084] The latent space sampling module is used to sum the latent variables output by the encoding module and the first prediction result of the battery cell to obtain sampling points corresponding to the number of battery cells.

[0085] The decoding module is used to map the sampling points corresponding to the number of battery cells output by the latent space sampling module to obtain the second prediction matrix of the battery cell.

[0086] In this embodiment of the invention, a physical enhancement encoder is first constructed and trained. This physical encoder is then used to compress and decompress the second prediction result of the battery pack and the first prediction result of the individual battery cells, and to predict the state of the individual battery cells. This allows for accurate identification of battery pack system faults using the predicted state of the individual battery cells.

[0087] During the training of the physical augmentation encoder, the MSE function is used as a loss function to back-adjust the prediction results of the physical augmentation encoder; wherein, the loss function is:

[0088]

[0089] In the formula, loss2 is the MSE function. The second prediction result of the battery cell is the output of the physical enhancement encoder, where k is the sampling time. Here is the actual observation matrix of a single battery cell, x n，k For the nth battery cell, U n，k Let be the terminal voltage of the nth battery cell.

[0090] Regarding step 102:

[0091] In some embodiments, the feature matrix of the battery cell is as follows:

[0092] X=[ewu,ewx,zu1(k),zx1(k),zu2(k),zx2(k)],

[0093]

[0094] In the formula, zu1(k) and zx1(k) are the maximum values ​​of the outlier coefficients of the corresponding variables, zu2(k) and zx2(k) are the deviations between the maximum and second largest outlier coefficients, and k is the sampling time. This represents the voltage value in the first row of the battery cell residual matrix. This represents the average voltage value in the first row of the battery cell residual matrix. This represents the state of charge value in the second row of the battery cell residual matrix. This represents the standard deviation of the state-of-charge values ​​in the second row of the battery cell residual matrix. This represents the standard deviation of the voltage values ​​in the first row of the battery cell residual matrix. This represents the average state-of-charge values ​​in the second row of the battery cell residual matrix. This represents the maximum voltage value in the first row of the residual matrix of a single battery cell. This represents the maximum value of the state of charge (SOC) value in the second row of the residual matrix of a single battery cell. This represents the second largest voltage value in the first row of the residual matrix of a single battery cell. is the second largest value of the state of charge (SOC) value in the second row of the residual matrix of the battery cell, ewu is the exponentially weighted moving average filtered signal of the maximum voltage value in the first row of the residual matrix, and ewx is the exponentially weighted moving average filtered signal of the maximum SOC value in the second row of the residual matrix.

[0095] To identify battery pack fault states, most related technologies compare the residual values ​​of the calculated state parameter prediction matrix of individual battery cells with a manually set threshold. However, in the actual operation of a battery pack, the operating conditions are relatively random, and the state of each individual cell is unstable. Simply comparing the magnitude of the residual value with the preset threshold cannot accurately reflect the fault state of the battery pack. Therefore, in this embodiment of the invention, after obtaining the residual matrix of the individual battery cells, feature extraction and calculation are further performed on the residual matrix to obtain a six-dimensional feature matrix composed of parameters closely related to the battery pack fault state. This not only helps to improve the accuracy of battery pack fault state diagnosis under random operating conditions, but also enables accurate identification of battery pack fault types.

[0096] Regarding steps 104 and 106:

[0097] In this embodiment of the invention, principal component analysis is used to perform dimensionality reduction and screening on the feature matrix of individual battery cells, thereby extracting the main feature parameters affecting the fault state of the battery pack from the feature matrix, reducing the influence of redundant information, and obtaining a low-dimensional feature matrix. Further, a comprehensive evaluation index is designed to quantify the low-dimensional feature matrix. Finally, based on the quantified index and a preset threshold, accurate diagnosis of the fault state of the battery pack can be achieved quickly and efficiently.

[0098] In some implementations, step 104 includes:

[0099] Principal component analysis was used to reduce the dimensionality of the feature matrix of the battery cell to obtain a low-dimensional feature matrix.

[0100] Based on the feature matrix of the battery cell and the low-dimensional feature matrix, the principal component space statistics and residual space statistics are obtained.

[0101] Based on the principal component space statistics and residual space statistics, a comprehensive evaluation index for the battery cell is obtained.

[0102] In this embodiment of the invention, the method for dimensionality reduction of the feature matrix of a single battery cell using principal component analysis is as follows:

[0103] First, the feature matrix of a single battery cell is centered. Each feature element in the feature matrix is ​​centered using the following formula:

[0104]

[0105] In the formula, X ij For the centered feature element, x ij The feature element in the i-th row and j-th column of the feature matrix of a single battery cell. s is the average of the characteristic elements in the j-th column. j Let be the standard deviation of the characteristic element in column j;

[0106] Then, calculate the covariance matrix of the centered feature matrix X1:

[0107]

[0108] Then according to |R-λI n | = 0 Calculate the eigenvalues ​​and eigenvectors of the covariance matrix, obtaining the eigenvalues ​​λ = (λ1, λ2, ..., λ3). n The corresponding eigenvector matrix P = (p1, p2, ..., p) n This allows us to determine the principal component vectors.

[0109] Furthermore, the centered feature matrix X1 is decomposed as follows:

[0110]

[0111] In the formula, T is the score matrix of the feature matrix X1, and t k =X1P k Essentially, it is the projection of the characteristic matrix X1 along the load vector direction, where each column vector t of T is... k This represents the k-th principal component. The space in which E resides is called the principal component subspace, k is the number of principal components, and E is the residual matrix in the principal component space, which contains information that has not been explained in the principal component model. The space in which E resides is called the residual subspace.

[0112] The number of principal components, k, is determined based on the cumulative contribution rate (CPV) of the principal components.

[0113]

[0114] For example, the number of principal components can be obtained by selecting the k principal components corresponding to the feature values ​​with a cumulative contribution rate of over 80%.

[0115] In some implementations, the principal component space statistics and the residual space statistics are calculated using the following formulas:

[0116] T 2 =X T PS -1 P T X

[0117] SPE = X(I-PP) T )X T

[0118] In the formula, T 2 Let X be the principal component space statistic, SPE be the residual space statistic, X be the residual matrix of the battery cell, P be the low-dimensional feature matrix, I be the instantaneous current, and S be the sample set X. T X.

[0119] In some embodiments, the comprehensive evaluation index of the battery cell is calculated using the following formula:

[0120]

[0121] In the formula, δ is the sum of the evaluation indicators for the battery cells. 2 These are the control limits for the principal component space statistics. SPE is the control limit of the residual space statistic, and T is the residual space statistic. 2 The principal component space statistics are... P is a symmetric positive definite matrix. k Let be the low-dimensional feature matrix, and Λ be a diagonal matrix composed of the eigenvalues ​​of the covariance matrix of the test samples.

[0122] Considering that principal component space statistics and residual space statistics can reflect the degree to which data deviates from the predicted value of the principal component model in the residual subspace and in the principal component space, respectively, in this embodiment of the invention, the comprehensive index of the battery pack system at the current moment is calculated based on the principal component space statistics and residual space statistics, which is beneficial for accurately diagnosing whether there is a fault in the battery pack system at the current moment.

[0123] In this embodiment of the invention, the comprehensive evaluation index of a single battery cell conforms to a chi-square distribution with h degrees of freedom and g coefficients, based on the probability function of the chi-square distribution. The comprehensive evaluation index of a single battery cell can be calculated, including:

[0124]

[0125] In the formula, S represents the training sample set.

[0126] It should be noted that the preset threshold in step 106 can be set by methods such as expert experience threshold and 3σ criterion, and the absolute value of the index and the frequency of exceeding the threshold can be monitored to achieve quantitative and graded alarm for battery pack faults.

[0127] Regarding step 108:

[0128] In some implementations, step 108 includes:

[0129] Based on the feature matrix corresponding to the faulty battery cell, calculate the contribution of each element in the feature matrix to the principal component space statistics and the residual space statistics.

[0130] Based on the contribution of each element in the feature matrix to the principal component space statistics and the residual space statistics, the contribution of each element in the feature matrix to the comprehensive index is obtained.

[0131] Based on the contribution of each element in the feature matrix to the comprehensive index and the initial state matrix of the battery cell, a fault type matrix of the faulty battery cell is constructed.

[0132] In this embodiment of the invention, in order to determine the cause of a battery cell failure, the contribution of each feature element in the feature matrix corresponding to the failure cell to the comprehensive index is first calculated. Then, based on the contribution of each feature element to the comprehensive index and further combined with the initial state matrix of the battery cell, classification features are extracted for different failure types to obtain the failure type matrix of the failure battery cell. This matrix is ​​then used as the input matrix of the failure type probability prediction model, thereby enabling accurate estimation of the failure type probability of the failure battery cell.

[0133] In some implementations, the contribution of each element in the feature matrix to the overall index is calculated using the following formula:

[0134]

[0135]

[0136] CONTq = e 2

[0137] In the formula, CONT represents the contribution of each element in the feature matrix to the comprehensive index, CONTq represents the contribution of each element in the feature matrix to the residual space statistic, CONTj represents the contribution of each element in the feature matrix to the principal component space statistic, and t... i λ is the score of the observation value in the i-th dimension. i Let P(i,j) be the eigenvalue corresponding to the i-th principal component, and let P(i,j) be the element in the i-th row and j-th column of the low-dimensional matrix. j Let u be the j-th variable in the characteristic matrix of a single battery cell. j It is a column vector composed of the mean of the j-th column in the feature matrix of a single battery cell;

[0138] In this embodiment of the invention, it is necessary to calculate the total contribution of the feature matrix corresponding to the faulty battery cell to the principal component space statistics. Specifically, the standardized score of each actual observation matrix is ​​first calculated. Assuming the preset threshold is T α ,by After determining the r scores that cause the battery pack failure state and identifying the time of failure occurrence, extract the principal component P at that time. i Calculate the score t in the i-th dimension. i :

[0139] t i =X T P i

[0140] Next, the contribution of each feature element in the feature matrix corresponding to the faulty battery cell, cont(i,j), to the score relative to the time of the fault occurrence is calculated:

[0141]

[0142] In the formula, λ i Let P(i,j) be the eigenvalue corresponding to the i-th principal component, and let P(i,j) be the element in the i-th row and j-th column of the k-th low-dimensional matrix. j Let u be the j-th variable in the characteristic matrix of a single battery cell. j Let be the mean of the j-th variable in the characteristic matrix of a single battery cell;

[0143] Finally, the calculated j process variables x j The contributions of each feature matrix to the principal component space statistics are summed to obtain the total contribution of the feature matrix to the statistics.

[0144]

[0145] Based on this, the fault type matrix of the faulty battery cell is constructed as follows:

[0146]

[0147] In the formula, CONT represents the contribution of each element in the feature matrix to the comprehensive index. T represents the maximum value of the comprehensive evaluation index of a single battery cell within a sampling period, ra represents the proportion of columns corresponding to the maximum value of the comprehensive evaluation index of a single battery cell that are always at their maximum value within 3000 consecutive sampling points, and T represents the percentage of columns corresponding to the maximum value of the comprehensive evaluation index of a single battery cell. m For in [t m -3000, t m The maximum temperature value during the sampling period, t m r represents the time corresponding to the maximum value of the comprehensive evaluation index of a single battery cell within a sampling period. isc U represents the insulation resistance value of the actual battery cell collected. min This represents the lowest voltage value of a single battery cell within a sampling period.

[0148] Regarding step 110:

[0149] In this embodiment of the invention, the prediction network model structure consists of linear layers and activation functions, and outputs the probability of whether it is a corresponding fault type by evaluating the similarity with the labeled sample. Specifically, based on the fault types of battery cells in the discrete cloud data, each fault type has corresponding trainable weights and biases. By using the fault types of battery cells and the probability values ​​corresponding to each fault type as the input and output sample sets of the fault type probability learning model, respectively, a fault type probability prediction model is obtained. Thus, by using the fault type matrix of the faulty battery cells constructed in step 108 as the input of the fault type probability prediction model, the probability of each fault type causing a battery cell fault can be estimated.

[0150] like Figure 2 , Figure 3 As shown, this embodiment of the invention provides a fault diagnosis device for a battery pack under random operating conditions. The device embodiment can be implemented through software, hardware, or a combination of both. From a hardware perspective, as... Figure 2 The diagram shown is a hardware architecture diagram of a computing device housing a fault diagnosis device for a battery pack under random operating conditions, as provided in an embodiment of the present invention. (Except for...) Figure 2 In addition to the processor, memory, network interface, and non-volatile memory shown, the computing device in the embodiment may also include other hardware, such as a forwarding chip responsible for processing packets. Taking software implementation as an example, such as... Figure 3 As shown, a device in a logical sense is formed by the CPU of its computing device reading the corresponding computer program from non-volatile memory into memory and running it. This embodiment provides a fault diagnosis device for a battery pack under random operating conditions, the device comprising:

[0151] The first computing unit 301 is used to perform residual calculation based on the obtained state parameter prediction matrix and actual observation parameter matrix of the battery cell to obtain the residual matrix of the battery cell; wherein, the state parameter prediction matrix of the battery cell is obtained by prediction using a pre-trained deep learning network;

[0152] The second calculation unit 302 is used to perform feature calculation on the residual matrix of the battery cell to obtain the feature matrix of the battery cell.

[0153] The third calculation unit 303 is used to perform dimensionality reduction processing on the feature matrix of the battery cell using principal component analysis to obtain the comprehensive evaluation index of the battery cell.

[0154] The fault determination unit 304 is used to compare the comprehensive evaluation index of each battery cell with a preset threshold, and to determine the battery cell whose comprehensive evaluation index exceeds the preset threshold as a faulty battery cell.

[0155] The fault type matrix construction unit 305 is used to construct the fault type matrix of the faulty battery cell based on the feature matrix corresponding to the faulty battery cell and the initial state matrix of the battery cell.

[0156] The fault diagnosis unit 306 is used to input the fault type matrix of the faulty battery cell into the trained fault type probability prediction model to obtain the trigger probability corresponding to the fault type of the faulty battery cell, so as to perform fault diagnosis on the battery pack.

[0157] In one embodiment of the present invention, the deep learning network in the first computing unit 301 includes a filter, a temporal network prediction model and a physical enhancement encoder connected in sequence;

[0158] The state parameter prediction matrix of the battery cell is predicted in the following way:

[0159] Based on the voltage parameters of the individual battery cells obtained at the current moment, construct the battery pack state model and the individual battery cell deviation model;

[0160] The state prediction of the battery pack and the individual battery cell deviation model at the current moment is performed using filters, and the filter prediction results of the battery pack and the individual battery cells at the next moment are obtained respectively; wherein, the filter prediction results include the voltage and state of charge of the battery pack and the individual battery cells at the next moment.

[0161] The battery state parameters and vehicle state parameters obtained at the current moment are input into the trained temporal network prediction model to obtain the temporal network prediction result of the battery pack at the next moment. During the training process of the temporal network prediction model, the filter prediction result of the battery pack is used as a loss function to adjust the temporal network prediction result in reverse.

[0162] The temporal network prediction results of the battery pack and the filter prediction results of the individual battery cells are input into the trained physical augmentation encoder to obtain the state parameter prediction matrix of the individual battery cells.

[0163] In one embodiment of the present invention, the feature matrix of the battery cell in the first calculation unit 302 is as follows:

[0164] X=[ewu,ewx,zu1(k),zx1(k),zu2(k),zx2(k)],

[0165]

[0166] In the formula, zu1(k) and zx1(k) are the maximum values ​​of the outlier coefficients of the corresponding variables, respectively, and zu2(k) and zx2(k) are the deviations between the maximum and second-largest outlier coefficients. This represents the voltage value in the first row of the battery cell residual matrix. This represents the average voltage value in the first row of the battery cell residual matrix. This represents the state of charge value in the second row of the battery cell residual matrix. This represents the standard deviation of the state-of-charge values ​​in the second row of the battery cell residual matrix. This represents the standard deviation of the voltage values ​​in the first row of the battery cell residual matrix. This represents the average state-of-charge values ​​in the second row of the battery cell residual matrix. This represents the maximum voltage value in the first row of the residual matrix of a single battery cell. This represents the maximum value of the state of charge (SOC) value in the second row of the residual matrix of a single battery cell. This represents the second largest voltage value in the first row of the residual matrix of a single battery cell. is the second largest value of the state of charge (SOC) value in the second row of the residual matrix of the battery cell, ewu is the exponentially weighted moving average filtered signal of the maximum voltage value in the first row of the residual matrix, and ewx is the exponentially weighted moving average filtered signal of the maximum SOC value in the second row of the residual matrix.

[0167] In one embodiment of the present invention, when the third calculation unit 303 performs dimensionality reduction processing on the feature matrix of the battery cell using principal component analysis, it performs the following operation:

[0168] Principal component analysis was used to reduce the dimensionality of the feature matrix of the battery cell to obtain a low-dimensional feature matrix.

[0169] Based on the feature matrix of the battery cell and the low-dimensional feature matrix, the principal component space statistics and residual space statistics are obtained.

[0170] Based on the principal component space statistics and residual space statistics, a comprehensive evaluation index for the battery cell is obtained.

[0171] In one embodiment of the present invention, in the third calculation unit 303, the principal component space statistics and the residual space statistics are calculated by the following formulas:

[0172] T 2 =X T PS -1 P T X

[0173] SPE = X(I-PP) T )X T

[0174] In the formula, T 2 Let SPE be the principal component space statistics, SPE be the residual space statistics, X be the residual matrix of the battery cell, P be the low-dimensional feature matrix, and I be the instantaneous current.

[0175] In one embodiment of the present invention, in the third calculation unit 303, the comprehensive evaluation index of the battery cell is calculated by the following formula:

[0176]

[0177] In the formula, δ is the sum of the evaluation indicators for the battery cells. 2 These are the control limits for the principal component space statistics. SPE is the control limit of the residual space statistic, and T is the residual space statistic. 2 The principal component space statistics are... Let P be a symmetric positive definite matrix, P be the low-dimensional feature matrix, and Λ be a diagonal matrix composed of the eigenvalues ​​of the covariance matrix of the test samples.

[0178] In one embodiment of the present invention, when constructing the fault type matrix unit 305, the following operations are performed:

[0179] Based on the feature matrix corresponding to the faulty battery cell, calculate the contribution of each element in the feature matrix to the principal component space statistics and the residual space statistics.

[0180] Based on the contribution of each element in the feature matrix to the principal component space statistics and the residual space statistics, the contribution of each element in the feature matrix to the comprehensive index is obtained.

[0181] Based on the contribution of each element in the feature matrix to the comprehensive index and the initial state matrix of the battery cell, a fault type matrix of the faulty battery cell is constructed.

[0182] In one embodiment of the present invention, in the fault type matrix construction unit 305, the contribution of each element in the feature matrix to the comprehensive index is calculated by the following formula:

[0183]

[0184] CONTq = e 2

[0185] In the formula, CONT represents the contribution of each element in the feature matrix to the comprehensive index, CONTq represents the contribution of each element in the feature matrix to the residual space statistic, CONTj represents the contribution of each element in the feature matrix to the principal component space statistic, and t... i λ is the score of the observed value of the i-th load vector. i Let P(i,j) be the eigenvalue corresponding to the i-th principal component, and let P(i,j) be the element in the i-th row and j-th column of the low-dimensional matrix. j For the j-th column of the battery cell feature matrix, u j Let be the column vector formed by the mean of the j-th column in the characteristic matrix of a single battery cell.

[0186] In one embodiment of the present invention, in the fault type matrix construction unit 305, the fault type matrix of the faulty battery cell is as follows:

[0187]

[0188] In the formula, CONT represents the contribution of each element in the feature matrix to the comprehensive index. T represents the maximum value of the comprehensive evaluation index of a single battery cell within a sampling period, ra represents the proportion of columns corresponding to the maximum value of the comprehensive evaluation index of a single battery cell that are always at their maximum value within 3000 consecutive sampling points, and T represents the percentage of columns corresponding to the maximum value of the comprehensive evaluation index of a single battery cell. m For in [t m -3000, t m The maximum temperature value during the sampling period, t m r represents the time corresponding to the maximum value of the comprehensive evaluation index of a single battery cell within a sampling period. isc U represents the insulation resistance value of the actual battery cell collected. min This represents the lowest voltage value of a single battery cell within a sampling period.

[0189] It is understood that the structures illustrated in the embodiments of the present invention do not constitute a specific limitation on a fault diagnosis device for a battery pack under random operating conditions. In other embodiments of the present invention, a fault diagnosis device for a battery pack under random operating conditions may include more or fewer components than illustrated, or combine some components, or split some components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0190] The information interaction and execution process between the modules in the above-mentioned device are based on the same concept as the method embodiment of the present invention, and the specific details can be found in the description of the method embodiment of the present invention, and will not be repeated here.

[0191] This invention also provides a computing device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements a fault diagnosis method for a battery pack under random operating conditions according to any embodiment of this invention.

[0192] This invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform a fault diagnosis method for a battery pack under random operating conditions according to any embodiment of this invention.

[0193] Specifically, a system or apparatus equipped with a storage medium may be provided, on which software program code implementing the functions of any of the embodiments described above is stored, and the computer (or CPU or MPU) of the system or apparatus may read and execute the program code stored in the storage medium.

[0194] In this case, the program code read from the storage medium can itself implement the function of any of the above embodiments, and therefore the program code and the storage medium storing the program code constitute part of the present invention.

[0195] Examples of storage media used to provide program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer via a communication network.

[0196] Furthermore, it should be clear that not only can the program code read by the computer be executed, but also the operating system or other components operating on the computer can be instructed based on the program code to perform some or all of the actual operations, thereby realizing the function of any of the embodiments described above.

[0197] Furthermore, it is understood that the program code read from the storage medium is written to the memory set in the expansion board inserted into the computer or to the memory set in the expansion module connected to the computer. Then, based on the instructions of the program code, the CPU or other components installed on the expansion board or expansion module execute some and all of the actual operations, thereby realizing the function of any of the above embodiments.

[0198] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus.

[0199] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as ROM, RAM, magnetic disk, or optical disk.

[0200] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A fault diagnosis method for a battery pack under random operating conditions, characterized in that, include: The residual matrix of the battery cell is obtained by calculating the residual based on the predicted state parameter matrix and the actual observed parameter matrix of the battery cell; the predicted state parameter matrix of the battery cell is obtained by using a pre-trained deep learning network. Deep learning networks consist of filters, temporal network prediction models, and physically enhanced encoders connected in sequence. The state parameter prediction matrix of a single battery cell is predicted in the following way: For each current moment, execute: Based on the voltage parameters of the individual battery cells obtained at the current moment, construct the battery pack state model and the individual battery cell deviation model; The state prediction of the battery pack and the individual battery cell deviation model at the current moment is performed by using filters, and the filter prediction results of the battery pack and the individual battery cells at the next moment are obtained respectively. The battery state parameters and vehicle state parameters obtained at the current moment are input into the trained temporal network prediction model to obtain the temporal network prediction result of the battery pack at the next moment. During the training process of the temporal network prediction model, the filter prediction result of the battery pack is used as a loss function to adjust the temporal network prediction result in reverse. The temporal network prediction results of the battery pack and the filter prediction results of the individual battery cells are input into the trained physical augmentation encoder to obtain the state parameter prediction matrix of the individual battery cells. The residual matrix of a single battery cell is used to perform feature calculations to obtain the feature matrix of the single battery cell. Principal component analysis was used to reduce the dimensionality of the feature matrix of a single battery cell, resulting in a low-dimensional feature matrix. Based on the characteristic matrix and low-dimensional characteristic matrix of the battery cell, the principal component space statistics and residual space statistics are obtained. Based on principal component space statistics and residual space statistics, a comprehensive evaluation index for individual battery cells is obtained. The principal component space statistics and residual space statistics are calculated using the following formulas: T 2 =X T PS -1 P T X SPE=X(I-PP T )X T In the formula, T 2 , where SPE is the principal component space statistic, X is the residual space statistic, P is the low-dimensional characteristic matrix, and I is the instantaneous current. The comprehensive evaluation index of a single battery cell is calculated using the following formula: In the formula, δ is the sum of evaluation indicators for individual battery cells. 2 Control limits for principal component space statistics. SPE represents the control limit for the residual space statistic, where SPE is the residual space statistic and T is the control limit. 2 Principal component space statistics Let P be a symmetric positive definite matrix, P be a low-dimensional feature matrix, and Λ be a diagonal matrix composed of the eigenvalues ​​of the covariance matrix of the test samples. The comprehensive evaluation index of each battery cell is compared with a preset threshold, and the battery cells whose comprehensive evaluation index exceeds the preset threshold are identified as faulty battery cells. Based on the feature matrix corresponding to the faulty battery cell and the initial state matrix of the battery cell, construct the fault type matrix of the faulty battery cell; The fault type matrix of the faulty battery cell is input into the trained fault type probability prediction model to obtain the trigger probability corresponding to the fault type of the faulty battery cell, so as to diagnose the fault of the battery pack.

2. The method according to claim 1, characterized in that, The feature matrix of the battery cell is as follows: X=[ewu,ewx,zu1(k),zx1(k),zu2(k),zx2(k)], In the formula, zu1(k) and zx1(k) are the maximum values ​​of the outlier coefficients of the corresponding variables, zu2(k) and zx2(k) are the deviations between the maximum and second largest outlier coefficients, and k is the sampling time. This represents the voltage value in the first row of the battery cell residual matrix. This represents the average voltage value in the first row of the battery cell residual matrix. This represents the state of charge value in the second row of the battery cell residual matrix. This represents the standard deviation of the state-of-charge values ​​in the second row of the battery cell residual matrix. This represents the standard deviation of the voltage values ​​in the first row of the battery cell residual matrix. This represents the average state-of-charge values ​​in the second row of the battery cell residual matrix. This represents the maximum voltage value in the first row of the residual matrix of a single battery cell. This represents the maximum value of the state of charge (SOC) value in the second row of the residual matrix of a single battery cell. This represents the second largest voltage value in the first row of the residual matrix of a single battery cell. is the second largest value of the state of charge (SOC) value in the second row of the residual matrix of the battery cell, ewu is the exponentially weighted moving average filtered signal of the maximum voltage value in the first row of the residual matrix, and ewx is the exponentially weighted moving average filtered signal of the maximum SOC value in the second row of the residual matrix.

3. The method according to claim 1, characterized in that, The step of constructing a fault type matrix for the faulty battery cell based on the feature matrix corresponding to the faulty battery cell and the initial state matrix of the battery cell includes: Based on the feature matrix corresponding to the faulty battery cell, calculate the contribution of each element in the feature matrix to the principal component space statistics and the residual space statistics. Based on the contribution of each element in the feature matrix to the principal component space statistics and the residual space statistics, the contribution of each element in the feature matrix to the comprehensive index is obtained. Based on the contribution of each element in the feature matrix to the comprehensive index and the initial state matrix of the battery cell, a fault type matrix of the faulty battery cell is constructed.

4. The method according to claim 1, characterized in that, The contribution of each element in the feature matrix to the comprehensive index is calculated using the following formula: CONTq=e 2 In the formula, CONT represents the contribution of each element in the feature matrix to the comprehensive index, CONTq represents the contribution of each element in the feature matrix to the residual space statistic, CONTj represents the contribution of each element in the feature matrix to the principal component space statistic, and t... i λ is the score of the observed value of the i-th load vector. i Let P(i,j) be the eigenvalue corresponding to the i-th principal component, and let P(i,j) be the element in the i-th row and j-th column of the low-dimensional matrix. j For the j-th column of the battery cell feature matrix, u j It is a column vector composed of the mean of the j-th column in the feature matrix of a single battery cell; and / or, The fault type matrix of the faulty battery cell is as follows: In the formula, CONT represents the contribution of each element in the feature matrix to the comprehensive index. T represents the maximum value of the comprehensive evaluation index of a single battery cell within a sampling period, ra represents the proportion of columns corresponding to the maximum value of the comprehensive evaluation index of a single battery cell that are always at their maximum value within 3000 consecutive sampling points, and T represents the percentage of columns corresponding to the maximum value of the comprehensive evaluation index of a single battery cell. m The maximum temperature value is defined within the sampling period [tm-3000, tm], where tm is the time corresponding to the maximum value of the comprehensive evaluation index of a single battery cell within a sampling period, and r is the maximum temperature value. isc U represents the insulation resistance value of the actual battery cell collected. min This represents the lowest voltage value of a single battery cell within a sampling period.

5. A fault diagnosis device for a battery pack under random operating conditions, used to implement the method according to any one of claims 1 to 4, characterized in that, include: The first computing unit is used to perform residual calculation based on the obtained state parameter prediction matrix and actual observation parameter matrix of the battery cell to obtain the residual matrix of the battery cell; wherein, the state parameter prediction matrix of the battery cell is obtained by prediction using a pre-trained deep learning network; The second calculation unit is used to perform feature calculation on the residual matrix of the battery cell to obtain the feature matrix of the battery cell. The third calculation unit is used to perform dimensionality reduction on the feature matrix of the battery cell using principal component analysis to obtain the comprehensive evaluation index of the battery cell. The fault determination unit is used to compare the comprehensive evaluation index of each battery cell with a preset threshold, and to determine the battery cell whose comprehensive evaluation index exceeds the preset threshold as a faulty battery cell. The fault type matrix construction unit is used to construct the fault type matrix of the faulty battery cell based on the feature matrix corresponding to the faulty battery cell and the initial state matrix of the battery cell. The fault diagnosis unit is used to input the fault type matrix of the faulty battery cell into the trained fault type probability prediction model to obtain the trigger probability corresponding to the fault type of the faulty battery cell, so as to perform fault diagnosis on the battery pack.

6. A computing device comprising a memory and a processor, wherein the memory stores a computer program, and the processor, when executing the computer program, implements the method as described in any one of claims 1-4.

7. A computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the method of any one of claims 1-4.

Images