Battery fault detection method and device, computer equipment and storage medium

Abstract

The invention relates to a battery fault detection method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring target battery parameters of a to-be-detected battery pack in a historical time period and actual battery voltage of each single battery in the to-be-detected battery pack; inputting the target battery parameters into the voltage prediction model to obtain a predicted average battery voltage; and determining a fault detection result of each single battery according to a difference value between the predicted average battery voltage and the actual battery voltage of each single battery. On one hand, more dimensional features can be obtained, so that the extracted battery features are more accurate, and the battery fault detection result is improved; and on the other hand, the fault detection result of the battery can be determined more concisely and efficiently.

Description

Technical Field

[0001] The present application relates to the field of vehicle technology, and in particular to a battery fault detection method, device, computer equipment and storage medium. Background Art

[0002] The battery of a new energy vehicle is its core component, which directly affects the vehicle's performance, mileage, and user experience. Therefore, detecting whether the vehicle's battery has a fault is crucial to ensure the normal operation of the vehicle.

[0003] In traditional technology, the prediction of battery failure is usually made by manually extracting battery parameter features. Since the manual extraction of battery parameter features is limited by human subjective factors, there are problems such as inaccurate extraction of battery parameter features, which leads to low reliability of battery fault detection results. Summary of the invention

[0004] Based on this, it is necessary to provide a battery fault detection method, device, computer equipment and storage medium to improve the accuracy of battery fault detection in response to the above technical problems.

[0005] In a first aspect, the present application provides a battery fault detection method, comprising:

[0006] Obtaining target battery parameters of the battery pack to be detected in a historical period and actual battery voltages of each single battery in the battery pack to be detected; wherein the target battery parameters include at least a target total voltage, a maximum target total voltage, a historical charging state, and a historical driving mileage;

[0007] Inputting the target battery parameters into a voltage prediction model to obtain a predicted average battery voltage;

[0008] The fault detection result of each single cell is determined according to the difference between the predicted average battery voltage and the actual battery voltage of each single cell.

[0009] In one of the embodiments, the voltage prediction model includes a temporal convolutional neural network module, a bidirectional long short-term memory network module, an attention mechanism module and an activation function module;

[0010] The step of inputting the target battery parameter into a voltage prediction model to obtain a predicted average battery voltage includes:

[0011] Inputting the target battery parameters into the temporal convolutional neural network module to extract local features and obtain local battery feature parameters;

[0012] Inputting the local battery characteristic parameters into the bidirectional long short-term memory network module to extract sequence characteristics to obtain sequence battery characteristic parameters;

[0013] Inputting the sequence battery characteristic parameters into the attention mechanism module for attention allocation to obtain the attention battery characteristic parameters;

[0014] The attention battery characteristic parameters are input into the activation function module for voltage prediction to obtain the predicted average battery voltage.

[0015] In one of the embodiments, the temporal convolutional neural network module includes a convolution layer, a weight normalization layer, an activation function layer and a first descent layer connected in sequence;

[0016] The bidirectional long short-term memory network module includes at least two memory network layers connected end to end; wherein each memory network layer includes a bidirectional long short-term memory network unit and a second descending layer connected in sequence.

[0017] In one embodiment, obtaining target battery parameters of the battery pack to be tested within a historical period includes:

[0018] Obtain the historical total voltage of the battery group to be tested at each historical moment in the historical period;

[0019] Obtain high-frequency components and low-frequency components of each historical total voltage;

[0020] Taking each low-frequency component as the target total voltage;

[0021] The maximum target total voltage is taken as the maximum target total voltage.

[0022] In one embodiment, determining the fault detection result of each single cell according to the difference between the predicted average battery voltage and the actual battery voltage of each single cell includes:

[0023] Determine a difference average value and a difference standard deviation according to a first difference between the predicted average battery voltage and the actual battery voltage of each single battery;

[0024] The fault detection result of each single cell is determined based on the size relationship between the target ratio corresponding to each single cell and the preset threshold; wherein the target ratio is the ratio of the second difference corresponding to the single cell to the standard deviation of the difference; and the second difference is the difference between the first difference corresponding to the single cell and the average value of the difference.

[0025] In one embodiment, the preset threshold includes a first preset threshold and a second preset threshold; the second preset threshold is greater than the first preset threshold;

[0026] The step of determining the fault detection result of each single cell according to the size relationship between the target ratio corresponding to each single cell and the preset threshold value includes:

[0027] For each single cell, if the target ratio corresponding to the single cell is less than or equal to the first preset threshold, it is determined that the fault detection result of the single cell is normal;

[0028] If the target ratio corresponding to the single cell is greater than the first preset threshold and less than the second preset threshold, then determining the fault detection result of the single cell according to the target ratio corresponding to the single cell in a future preset period and the preset threshold;

[0029] If the target ratio corresponding to the single cell is greater than or equal to the second preset threshold, it is determined that the fault detection result of the single cell is abnormal.

[0030] In one embodiment, determining the fault detection result of the single cell according to the target ratio corresponding to the single cell in a future preset period of time and the preset threshold value includes:

[0031] Obtaining the actual battery voltage of the single battery within a future preset time period;

[0032] determining a third difference value according to the predicted average battery voltage and the actual battery voltage of the single battery in a future preset time period;

[0033] If the target ratio corresponding to the single cell in a future preset time period is less than or equal to the first preset threshold, the fault detection result of the single cell is determined to be normal; wherein the target ratio corresponding to the single cell in a future preset time period is the ratio of the third difference corresponding to the single cell to the standard deviation of the difference.

[0034] In a second aspect, the present application also provides a battery fault detection device, comprising:

[0035] An acquisition module, used for acquiring target battery parameters of a battery pack to be detected in a historical period and actual battery voltages of each single battery in the battery pack to be detected;

[0036] A prediction module, used for inputting the target battery parameters into a voltage prediction model to obtain a predicted average battery voltage; wherein the voltage prediction model includes a temporal convolutional neural network module, a bidirectional long short-term memory network module, an attention mechanism module and an activation function module;

[0037] The determination module is used to determine the fault detection result of each single cell according to the difference between the predicted average battery voltage and the actual battery voltage of each single cell.

[0038] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0039] Obtaining target battery parameters of the battery pack to be detected in a historical period and actual battery voltages of each single battery in the battery pack to be detected; wherein the target battery parameters include at least a target total voltage, a maximum target total voltage, a historical charging state, and a historical driving mileage;

[0040] Inputting the target battery parameters into a voltage prediction model to obtain a predicted average battery voltage;

[0041] The fault detection result of each single cell is determined according to the difference between the predicted average battery voltage and the actual battery voltage of each single cell.

[0042] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0043] Obtaining target battery parameters of the battery pack to be detected in a historical period and actual battery voltages of each single battery in the battery pack to be detected; wherein the target battery parameters include at least a target total voltage, a maximum target total voltage, a historical charging state, and a historical driving mileage;

[0044] Inputting the target battery parameters into a voltage prediction model to obtain a predicted average battery voltage;

[0045] The fault detection result of each single cell is determined according to the difference between the predicted average battery voltage and the actual battery voltage of each single cell.

[0046] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0047] Obtaining target battery parameters of the battery pack to be detected in a historical period and actual battery voltages of each single battery in the battery pack to be detected; wherein the target battery parameters include at least a target total voltage, a maximum target total voltage, a historical charging state, and a historical driving mileage;

[0048] Inputting the target battery parameters into a voltage prediction model to obtain a predicted average battery voltage;

[0049] The fault detection result of each single cell is determined according to the difference between the predicted average battery voltage and the actual battery voltage of each single cell.

[0050] The above-mentioned battery fault detection method, device, computer equipment and storage medium obtain the target battery parameters of the battery group to be detected in the historical period and the actual battery voltage of each single battery in the battery group to be detected; and input the target battery parameters into the voltage prediction model to obtain the predicted average battery voltage; and then determine the fault detection result of each single battery according to the difference between the predicted average battery voltage and the actual battery voltage of each single battery. The above scheme, on the one hand, by inputting the target battery parameters into the voltage prediction model to predict the battery voltage, compared with the manual extraction of battery parameter features, the voltage prediction model can obtain more dimensional features, and make the extracted battery features more accurate, thereby improving the battery fault detection results; on the other hand, according to the difference between the predicted average battery voltage and the actual battery voltage of each single battery, the fault detection result of each single battery is determined, which can determine the battery fault detection result more concisely and efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 is a flow chart of a battery failure detection method in one embodiment;

[0053] Figure 2 A schematic diagram of the structure of a voltage prediction model in one embodiment;

[0054] Figure 3 A schematic diagram of a process of obtaining a predicted average battery voltage through a voltage prediction model in one embodiment;

[0055] Figure 4 is a schematic diagram of the structure of a voltage prediction model in another embodiment;

[0056] Figure 5 A schematic diagram of a process for obtaining target battery parameters in one embodiment;

[0057] Figure 6 A schematic diagram of a process for determining a fault detection result of each single battery in an embodiment;

[0058] Figure 7 A schematic diagram of a flow chart for determining a fault detection result of each single cell in another embodiment;

[0059] Figure 8A schematic diagram of a process for determining a single cell fault detection result in another embodiment;

[0060] Fig. 9 is a flow chart of a battery failure detection method in another embodiment;

[0061] Fig.10 is a structural block diagram of a battery fault detection device in one embodiment;

[0062] Fig.11 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0064] The battery fault detection method provided in the embodiment of the present application can be applied to the application scenario of fault detection of the battery of the vehicle. The method can be executed by a server or by a terminal with a certain computing power.

[0065] The server may be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services. The terminal may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, IoT devices, and portable wearable devices. IoT devices may be smart speakers, smart TVs, smart air conditioners, smart car devices, and projection devices. Portable wearable devices may be smart watches, smart bracelets, head-mounted devices, and the like. Head-mounted devices may be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, and the like.

[0066] In an exemplary embodiment, Figure 1 As shown, a battery failure detection method is provided, which is described by taking the method applied to a server as an example, and includes the following steps:

[0067] S101, obtaining target battery parameters of a battery pack to be detected in a historical period and actual battery voltages of each single battery in the battery pack to be detected.

[0068] Among them, the battery pack to be tested can be a battery pack on the vehicle, and the battery pack can be composed of multiple single cells. The historical time period can be set according to the specific situation, which can be within a few years, within a year, within half a year, etc., and no specific limitation is made here for the historical time period. The target battery parameters at least include the target total voltage, the maximum target total voltage, the historical charging state and the historical mileage; among them, the target battery parameters are the battery parameters of the battery pack to be tested in the historical period; the target total voltage is the sum of the voltages of each single cell in the battery pack to be tested; the maximum target total voltage is the maximum target total voltage among the target total voltages in the historical period; the historical charging state represents the remaining power of the battery; the historical mileage can be understood as the total historical history of the battery to be tested in the historical period.

[0069] For example, the target battery parameters of the battery pack to be tested and the actual battery voltage of each single cell in the battery pack to be tested in the historical period can be obtained from the battery management system (BMS) or other data recording equipment. Among them, the BMS is a system responsible for monitoring, controlling and managing the performance and status of the single cells in the battery pack. It can record the voltage, current, temperature and other parameters of the single cell in real time or periodically.

[0070] Exemplarily, the target battery parameter may be in the form of an N×M matrix, where N represents the length of the time series in the target battery parameter, and M represents the target battery parameter at each time point in the time series.

[0071] S102, inputting the target battery parameters into a voltage prediction model to obtain a predicted average battery voltage.

[0072] See also Figure 2 , Figure 2A schematic diagram of the structure of a voltage prediction model is provided. The voltage prediction model may include a temporal convolutional neural network module, a bidirectional long short-term memory network module, an attention mechanism module, and an activation function module. The temporal convolutional neural network module (TCN) is a convolutional neural network for processing time series data. The bidirectional long short-term memory network (BiLSTM) module is an improved recurrent neural network (RNN) that can capture bidirectional dependencies in a sequence. The attention mechanism module is a mechanism used in a deep learning model to enhance the model's ability to focus on key information in input data. The activation function module introduces nonlinear factors so that the neural network can solve nonlinear problems. For example, the activation function can be Sigmoid, Tanh, ReLU, etc.

[0073] For example, the target battery parameters can be input into the voltage prediction model, and the predicted average battery voltage can be obtained through the TCN module, the BiLSTM module, the attention mechanism module and the activation function module. The predicted average battery voltage can be understood as the predicted average battery voltage of each single battery in the battery pack to be tested.

[0074] S103, determining the fault detection result of each single cell according to the difference between the predicted average battery voltage and the actual battery voltage of each single cell.

[0075] Then, the fault detection result of each single cell can be determined based on the difference between the predicted average battery voltage and the actual battery voltage of each single cell. For example, if the difference is within a certain range, it can be determined that each single cell is normal; if the difference exceeds a certain range, it can be determined that each single cell is abnormal. Alternatively, a corresponding relationship between each difference and the fault detection result can be established in advance, and then the fault detection result of each single cell can be determined based on the difference and the corresponding relationship.

[0076] The above-mentioned battery fault detection method obtains the target battery parameters of the battery group to be detected in the historical period and the actual battery voltage of each single battery in the battery group to be detected; and inputs the target battery parameters into the voltage prediction model to obtain the predicted average battery voltage; wherein the voltage prediction model includes a temporal convolutional neural network module, a bidirectional long short-term memory network module, an attention mechanism module and an activation function module; and then determines the fault detection result of each single battery according to the difference between the predicted average battery voltage and the actual battery voltage of each single battery. In the above scheme, on the one hand, by inputting the target battery parameters into the voltage prediction model to predict the battery voltage, compared with the manual extraction of battery parameter features, the voltage prediction model can obtain more dimensional features, and make the extracted battery features more accurate, thereby improving the battery fault detection results; on the other hand, according to the difference between the predicted average battery voltage and the actual battery voltage of each single battery, the fault detection result of each single battery is determined, which can determine the battery fault detection result more concisely and efficiently.

[0077] In some optional implementations, see Figure 3 , Figure 3 A schematic diagram of a process for obtaining a predicted average battery voltage through a voltage prediction model is provided, which specifically includes the following steps:

[0078] S301, inputting the target battery parameters into a temporal convolutional neural network module to extract local features and obtain local battery feature parameters.

[0079] Exemplarily, the target battery parameters may be input into the TCN module for local feature extraction to obtain local battery feature parameters. For example, the target battery parameters may be input into the TCN module for local feature extraction to obtain local battery feature parameters of the target battery parameters at each historical moment.

[0080] S302, inputting the local battery characteristic parameters into the bidirectional long short-term memory network module to extract sequence characteristics and obtain sequence battery characteristic parameters.

[0081] Exemplarily, the local battery characteristic parameters can be input into the BiLSTM module for sequence feature extraction to obtain the sequence battery characteristic parameters. For example, the target battery parameters are input into the BiLSTM module for time series feature extraction to obtain the sequence battery characteristic parameters of the target battery parameters at each historical moment in the time series.

[0082] S303, inputting the sequence battery characteristic parameters into the attention mechanism module for attention allocation to obtain the attention battery characteristic parameters.

[0083] Furthermore, the sequence battery characteristic parameters can be input into the attention mechanism module for attention allocation to obtain the attention battery characteristic parameters. Among them, the attention battery characteristic parameters can be understood as the weight information corresponding to each battery parameter in the sequence battery parameters. The attention mechanism can focus on more important information features in the battery voltage prediction task among many pieces of information, reduce the attention to other information, filter out irrelevant information, and improve the efficiency and accuracy of battery voltage prediction.

[0084] S304, inputting the attention battery characteristic parameters into the activation function module for voltage prediction to obtain the predicted average battery voltage.

[0085] Then, the attention battery characteristic parameters are input into the activation function module for voltage prediction to obtain the predicted average battery voltage. For example, the activation function module uses an activation function to calculate the attention battery characteristic parameters to obtain the predicted average battery voltage.

[0086] In an embodiment of the present application, by adopting a temporal convolutional neural network module, a bidirectional long short-term memory network module, an attention mechanism module and an activation function module, feature extraction and processing of target battery parameters can be performed to improve the accuracy of predicting the average battery voltage.

[0087] In some optional implementations, the TCN module may include a convolutional layer, a weight normalization layer, an activation function layer, and a first descent layer connected in sequence; the BiLSTM module may include at least two memory network layers connected end to end; wherein each memory network layer includes a bidirectional long short-term memory network unit connected in sequence and a second descent layer.

[0088] Among them, the TCN module uses residual units to connect between each layer of the network, which can realize the transmission of information across layers, thereby accelerating the feedback and convergence of the deep network, enhancing the generalization ability of the model, and avoiding the phenomenon of "gradient disappearance" or "gradient explosion" caused by multi-layer back propagation of error signals. The convolution layer of TCN can extract the correct battery parameter features; the weight normalization layer is conducive to accelerating the convergence of the model and can prevent the input and output from having different widths. The 1×1 convolution in the weight normalization layer ensures that the tensor is the same when the next layer receives it; the activation function layer is a feature processing layer, which speeds up the training speed by normalization; the first dropout layer (Dropout) effectively prevents overfitting problems in the signal preprocessing process.

[0089] Among them, the bidirectional long short-term memory network unit is used to learn the characteristics of data in time series, and the second dropout layer (Dropout) is used to prevent overfitting during training.

[0090] The embodiment of the present application takes a bidirectional long short-term memory network module including three end-to-end memory network layers as an example to introduce a voltage prediction model.

[0091] For example, see Figure 4 , Figure 4 A structural schematic diagram of another voltage prediction model is provided.

[0092] Among them, the target battery parameters can be input into the convolution layer for convolution with the hole causal convolution kernel to obtain a multi-level group feature vector; then the multi-level group feature vector is input into the weight normalization layer to normalize the weight of the convolution kernel to obtain a normalized feature vector; then the normalized feature vector is input into the activation function layer, where the activation function is Sigmoid, to obtain an activated feature vector; then the activated feature vector is input into the first dropout layer for processing to prevent overfitting during training. Then the local battery feature parameters output by the first dropout layer are input into the BiLSTM module.

[0093] Among them, the target battery parameters can be input into the convolution layer for convolution with the hole causal convolution kernel. The convolution operation is as follows:

[0094]

[0095] Among them, F(s) is the output of the hole convolution at position s. Indicates that the input sequence x is at position The value at . Represents the convolution kernel The weight at position i. d represents the dilation rate, which determines the step size of the convolution kernel sliding on the input sequence.

[0096] Exemplarily, the local battery characteristic parameters output by the first descent layer can be input into the bidirectional long short-term memory network unit of the first memory network layer in the BiLSTM module to learn the characteristics of the data in the time series, and the learned characteristics in the time series are input into the second descent layer to avoid overfitting during the training process; the output data of the first memory network layer is then processed in the same way through two identical memory network layers to finally obtain the sequence battery characteristic parameters.

[0097] Furthermore, the sequence battery characteristic parameters can be input into the attention mechanism module for attention allocation, wherein the attention mechanism module can determine the weight matrix of the battery characteristic parameters by using the probability allocation weight method. The specific calculation formula is as follows:

[0098]

[0099]

[0100]

[0101] Among them, e t 、a t They are the attention scores and weights corresponding to different features of the input data at time t, h t represents the output of the BiLSTM module at time t, b is the bias coefficient, u and is the weight coefficient, C t Represents the output of the attention mechanism module at time t, that is, the characteristic parameters of the attention battery at time t.

[0102] Furthermore, the characteristic parameters of the attention battery at time t can be input into the activation function module, and the activation function module performs voltage prediction. The specific calculation formula for determining the predicted total voltage is as follows:

[0103]

[0104] Among them, y t represents the predicted average battery voltage output at time t, represents the weight coefficient, b o Represents the bias term.

[0105] In an embodiment of the present application, by adopting a temporal convolutional neural network module including a sequentially connected convolutional layer, a weight normalization layer, an activation function layer and a first descent layer, and a bidirectional long short-term memory network module including at least two end-to-end memory network layers, feature extraction and processing of target battery parameters are performed, the accuracy of predicting the average battery voltage can be improved.

[0106] In some optional implementations, in order to avoid the influence of noise signals on the predicted average battery voltage, the target battery parameters may be denoised.

[0107] For example, see Figure 5 , Figure 5 A schematic diagram of a process for obtaining target battery parameters is provided, which specifically includes the following steps:

[0108] S501, obtaining the historical total voltage of the battery pack to be detected at each historical moment in a historical period.

[0109] Exemplarily, the historical total voltage of the battery pack to be tested at each historical moment in the historical period can be obtained in the BMS, that is, the total voltage of the battery pack to be tested at each historical moment in the historical period is obtained. The historical total voltage obtained here includes a noise signal, and the historical total voltage needs to be denoised.

[0110] S502, obtaining high-frequency components and low-frequency components of each historical total voltage.

[0111] Furthermore, each historical total voltage may be subjected to denoising to obtain high-frequency components and low-frequency components of each historical total voltage. For example, each historical total voltage may be subjected to variational mode decomposition (VMD) to obtain high-frequency components and low-frequency components.

[0112] Among them, VMD is an adaptive, non-recursive signal mode decomposition method that uses iterative search for the optimal solution of the variational model to determine the center frequency and bandwidth of each decomposed component, thereby achieving effective separation of the signal frequency domain. The VMD method can decompose the historical total voltage in the following steps:

[0113] Initialization: Initialize each mode and its center frequency, which is usually set by random generation or based on prior knowledge.

[0114] Iterative update: The alternating direction method of multipliers (ADMM) is used for iterative update. In each iteration, each mode and its center frequency are updated separately.

[0115] Stopping Condition: The iteration process will continue until a preset stopping condition is reached. This is usually based on whether the update of the mode shape and center frequency is less than a preset threshold.

[0116] Output results: When the iteration stops, the final modes and their center frequencies are output. These modes represent different components of the historical total voltage, and the center frequency reflects the frequency domain characteristics of these components. Thus, the high-frequency and low-frequency components of the historical total voltage are obtained.

[0117] S503: taking each low-frequency component as a target total voltage.

[0118] Among them, the low-frequency component can be considered as the steady-state component of the voltage, so the low-frequency component can be used as the target total voltage, thus achieving denoising and avoiding the influence of noise on the predicted average battery voltage.

[0119] S504: Taking the maximum target total voltage as the maximum target total voltage.

[0120] Furthermore, the maximum target total voltage can be used as the maximum target total voltage. Since the target total voltage has been subjected to denoising, the obtained maximum target total voltage has also been subjected to denoising, further avoiding the influence of noise on the predicted average battery voltage.

[0121] In the embodiment of the present application, by denoising the target total voltage, the target total voltage and the maximum target total voltage are free of noise, thereby avoiding the influence of noise on the predicted average battery voltage, thereby improving the accuracy of the predicted average battery voltage.

[0122] In some optional implementations, when a battery fails, the battery voltage change is the most rapid response feature, and thus whether a single cell is abnormal can be determined based on the difference between the predicted average battery voltage and the actual battery voltage of each single cell.

[0123] For example, see Figure 6 , Figure 6 A schematic diagram of a process for determining a fault detection result of each single cell is provided, which specifically includes the following steps:

[0124] S601, determining a difference average and a difference standard deviation according to a first difference between the predicted average battery voltage and the actual battery voltage of each single battery.

[0125] Exemplarily, the difference average value and the difference standard deviation may be determined based on the first difference between the predicted average battery voltage and the actual battery voltage of each single battery. For example, if there are n single batteries, the first difference between the actual battery voltage of each single battery and the predicted average battery voltage is averaged to obtain the difference average value; then the average value of the square of the difference between the first difference corresponding to each single battery and the difference average value is determined, and the square root of the average value is taken to obtain the difference standard deviation.

[0126] S602, determining the fault detection result of each single cell according to the size relationship between the target ratio corresponding to each single cell and a preset threshold.

[0127] The target ratio is the ratio of the second difference corresponding to the single cell to the standard deviation of the difference; the second difference is the difference between the first difference corresponding to the single cell and the average difference. That is, the difference between the first difference corresponding to the single cell and the average difference can be calculated and used as the second difference.

[0128] Then, the Z score can be introduced to determine the fault detection result of each single cell. The target ratio corresponding to each single cell can be used as the Z score of each single cell, and the calculation formula of the Z score of each single cell is as follows:

[0129]

[0130] Where Z(i) is the Z score of the ith single cell; E is the difference between the first difference and the average difference of the ith single cell, E ave is the mean value of the difference, is the standard deviation of the difference.

[0131] The fault detection result of each single cell can be determined according to the size relationship between the target ratio corresponding to each single cell and the preset threshold. The preset threshold can be set according to empirical data and actual needs, for example, it can be set to 3.5, and no specific limitation is made here.

[0132] For example, single cell battery fault detection rules can be set in advance, where the battery fault detection rule can be that the target ratio corresponding to the single cell battery exceeds a preset threshold, i.e., the single cell battery is determined to be faulty; or, the part of the target ratio corresponding to the single cell battery that exceeds the preset threshold is within a certain range, i.e., the single cell battery is determined to be normal, and the exceeding part is outside a certain range, i.e., the single cell battery is determined to be abnormal.

[0133] In the embodiment of the present application, the fault detection result of each single cell is determined by comparing the target ratio corresponding to each single cell with the preset threshold value, so that the fault detection result of each single cell can be determined more efficiently.

[0134] In some optional implementations, the preset threshold may include a first preset threshold and a second preset threshold; the second preset threshold is greater than the first preset threshold. For example, the first preset threshold may be 3.5, and the second preset threshold may be 4.

[0135] Based on this, see Figure 7 , Figure 7 Another flow chart for determining the fault detection result of each single battery is provided, which specifically includes the following steps:

[0136] S701 , for each single cell, if the target ratio corresponding to the single cell is less than or equal to a first preset threshold, then determining that the fault detection result of the single cell is normal.

[0137] Exemplarily, if the ratio of the second difference corresponding to the single cell to the standard deviation of the difference is less than or equal to the first preset threshold, ie, |Z(i)|≤3.5, it can be determined that the fault detection result of the single cell is normal.

[0138] S702, if the target ratio corresponding to the single cell is greater than the first preset threshold and less than the second preset threshold, determine the fault detection result of the single cell according to the target ratio corresponding to the single cell in a future preset period and the preset threshold.

[0139] For example, if the target ratio corresponding to the single cell is greater than the first preset threshold value and less than the second preset threshold value, that is, 3.5<|Z(i)|<4, it means that the single cell has an abnormal voltage and needs to be continuously monitored, that is, the target ratio corresponding to the single cell in the future preset period can be monitored, and the fault detection result of the single cell can be determined based on the target ratio corresponding to the single cell in the future preset period and the preset threshold value. Among them, the preset time length can be set according to empirical data or actual needs, for example, it can be set to 1 minute, and no specific limitation is made here.

[0140] Exemplarily, it can be judged according to the duration that the target ratio corresponding to the single cell is greater than the first preset threshold and less than the second preset threshold. For example, if the target ratio corresponding to the single cell is greater than the first preset threshold and the duration that is less than the second preset threshold is less than the duration of the preset period or the duration threshold, it can be determined that the single cell is normal; if the target ratio corresponding to the single cell is greater than the first preset threshold and the duration that is less than the second preset threshold is greater than or equal to the duration of the preset period or the duration threshold, it can be determined that the single cell is abnormal. Wherein, the duration threshold is less than the duration of the preset period.

[0141] S703: If the target ratio corresponding to the single cell is greater than or equal to a second preset threshold, it is determined that the fault detection result of the single cell is abnormal.

[0142] Exemplarily, if the ratio of the second difference corresponding to the single cell to the standard deviation of the difference is greater than or equal to the second preset threshold, ie, |Z(i)|≥4, then the fault detection result of the single cell is determined to be abnormal.

[0143] In the embodiment of the present application, by setting a first preset threshold and a second preset threshold, and determining whether a single cell battery is abnormal based on the ratio of the second difference corresponding to the single cell battery to the standard deviation of the difference and the relationship between the first preset threshold and the second preset threshold, the efficiency and accuracy of determining whether a single cell battery is abnormal can be improved.

[0144] In some optional implementations, when the ratio of the second difference corresponding to a single cell to the standard deviation of the difference is greater than the first preset threshold and less than the second preset threshold, that is, an abnormal voltage occurs in the single cell, the single cell needs to be continuously monitored to determine whether the single cell is actually abnormal.

[0145] Based on this, see Figure 8 , Figure 8 A further flow chart for determining a single cell battery fault detection result is provided, which specifically includes the following steps:

[0146] S801, obtaining the actual battery voltage of the single battery in a future preset period of time.

[0147] Exemplarily, the actual battery voltage of the single battery in a future preset period of time may be obtained, so as to determine the third difference value according to the actual battery voltage of the single battery in the future preset period of time and the predicted average battery voltage.

[0148] S802, determining a third difference value according to the predicted average battery voltage and the actual battery voltage of the single battery in a future preset time period.

[0149] The difference between the actual battery voltage of the single battery in the future preset period and the predicted average battery voltage may be used as the third difference.

[0150] S803: If the target ratio corresponding to the single cell in a future preset time period is less than or equal to a first preset threshold, it is determined that the fault detection result of the single cell is normal.

[0151] The target ratio corresponding to the single cell in the future preset time period is the ratio of the third difference corresponding to the single cell to the standard deviation of the difference.

[0152] For example, if the ratio of the third difference to the standard deviation of the difference of the single battery is less than or equal to the first preset threshold value within the future preset period, the fault detection result of the single battery can be determined to be normal. If not, the fault detection result of the single battery can be determined to be abnormal.

[0153] In the embodiment of the present application, after an abnormal voltage appears in a single cell battery, the voltage of the single cell battery is continuously monitored, and whether the single cell battery has an abnormality is determined based on the voltage of the single cell battery in a future time period, thereby avoiding misjudgment of battery abnormality and improving the reliability of the fault detection results of the single cell battery.

[0154] In some optional implementations, see Figure 4 ,exist Figure 4 Based on Fig. 9 , Fig. 9 A flowchart of another battery fault detection method is provided, which specifically includes the following steps:

[0155] S901, obtaining the historical total voltage, historical charging state and historical driving mileage of the battery pack to be detected at each historical moment in a historical period.

[0156] S902, obtaining high-frequency components and low-frequency components of each historical total voltage.

[0157] S903: taking each low-frequency component as a target total voltage.

[0158] S904: Taking the maximum target total voltage as the maximum target total voltage.

[0159] S905, inputting the target total voltage, the maximum target total voltage, the historical charging state and the historical mileage into a time series convolutional neural network module to extract local features and obtain local battery feature parameters.

[0160] S906, inputting the local battery characteristic parameters into the bidirectional long short-term memory network module to extract sequence characteristics and obtain sequence battery characteristic parameters.

[0161] S907, input the sequence battery characteristic parameters into the attention mechanism module for attention allocation to obtain the attention battery characteristic parameters.

[0162] S908, input the attention battery characteristic parameters into the activation function module for voltage prediction to obtain the predicted average battery voltage.

[0163] S909 , determining a difference average and a difference standard deviation according to a first difference between the predicted average battery voltage and the actual battery voltage of each single battery.

[0164] S910, determining a fault detection result of each single cell according to a magnitude relationship between a target ratio corresponding to each single cell and a preset threshold.

[0165] Exemplarily, for each single cell, if the target ratio corresponding to the single cell is less than or equal to the first preset threshold, it is determined that the fault detection result of the single cell is normal;

[0166] If the target ratio corresponding to the single cell battery is greater than the first preset threshold value and less than the second preset threshold value, the actual battery voltage of the single cell battery in the future preset time period can be obtained; and the third difference is determined based on the predicted average battery voltage and the actual battery voltage of the single cell battery in the future preset time period; if the target ratio corresponding to the single cell battery in the future preset time period is less than or equal to the first preset threshold value, the fault detection result of the single cell battery is determined to be normal.

[0167] If the target ratio corresponding to the single cell is greater than or equal to the second preset threshold, it is determined that the fault detection result of the single cell is abnormal.

[0168] The specific process from S901 to S910 can refer to the description of the above method embodiment, and its implementation principle and technical effect are similar, which will not be repeated here.

[0169] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0170] Based on the same inventive concept, the embodiment of the present application also provides a battery fault detection device for implementing the battery fault detection method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more battery fault detection device embodiments provided below can refer to the limitations of the battery fault detection method above, and will not be repeated here.

[0171] In an exemplary embodiment, Fig.10 As shown, a battery fault detection device is provided, comprising:

[0172] The acquisition module 10 is used to acquire the target battery parameters of the battery pack to be detected in the historical period and the actual battery voltage of each single battery in the battery pack to be detected; wherein the target battery parameters at least include the target total voltage, the maximum target total voltage, the historical charging state and the historical mileage;

[0173] A prediction module 20, for inputting target battery parameters into a voltage prediction model to obtain a predicted average battery voltage;

[0174] The determination module 30 is used to determine the fault detection result of each single cell according to the difference between the predicted average battery voltage and the actual battery voltage of each single cell.

[0175] The battery fault detection device obtains the target battery parameters of the battery group to be detected in the historical period and the actual battery voltage of each single battery in the battery group to be detected; and inputs the target battery parameters into the voltage prediction model to obtain the predicted average battery voltage; wherein the voltage prediction model includes a temporal convolutional neural network module, a bidirectional long short-term memory network module, an attention mechanism module and an activation function module; and then determines the fault detection result of each single battery according to the difference between the predicted average battery voltage and the actual battery voltage of each single battery. In the above scheme, on the one hand, by inputting the target battery parameters into the voltage prediction model to predict the battery voltage, compared with manually extracting the battery parameter features, the voltage prediction model can obtain more dimensional features, and make the extracted battery features more accurate, thereby improving the battery fault detection results; on the other hand, according to the difference between the predicted average battery voltage and the actual battery voltage of each single battery, the fault detection result of each single battery is determined, which can determine the battery fault detection result more concisely and efficiently.

[0176] In one embodiment, the voltage prediction model includes a sequential convolutional neural network module, a bidirectional long short-term memory network module, an attention mechanism module and an activation function module; the prediction module 20 is specifically used for:

[0177] The target battery parameters are input into the temporal convolutional neural network module for local feature extraction to obtain local battery feature parameters; the local battery feature parameters are input into the bidirectional long short-term memory network module for sequence feature extraction to obtain sequence battery feature parameters; the sequence battery feature parameters are input into the attention mechanism module for attention allocation to obtain attention battery feature parameters; the attention battery feature parameters are input into the activation function module for voltage prediction to obtain the predicted average battery voltage.

[0178] In one embodiment, the temporal convolutional neural network module includes a convolutional layer, a weight normalization layer, an activation function layer and a first descent layer connected in sequence; the bidirectional long short-term memory network module includes at least two memory network layers connected end to end; wherein each memory network layer includes a bidirectional long short-term memory network unit connected in sequence and a second descent layer.

[0179] In one embodiment, the acquisition module 10 is specifically used for:

[0180] Obtain the historical total voltage of the battery pack to be tested at each historical moment in the historical period; obtain the high-frequency component and the low-frequency component of each historical total voltage; use each low-frequency component as the target total voltage; and use the maximum target total voltage as the maximum target total voltage.

[0181] In one embodiment, the determination module 30 specifically includes:

[0182] A first determining unit, configured to determine a difference average value and a difference standard deviation according to a first difference between the predicted average battery voltage and the actual battery voltage of each single battery;

[0183] The second determination unit is used to determine the fault detection result of each single cell according to the size relationship between the target ratio corresponding to each single cell and the preset threshold; wherein the target ratio is the ratio of the second difference corresponding to the single cell to the standard deviation of the difference; and the second difference is the difference between the first difference corresponding to the single cell and the average value of the difference.

[0184] In one embodiment, the preset threshold includes a first preset threshold and a second preset threshold; the second preset threshold is greater than the first preset threshold; and the second determining unit specifically includes:

[0185] A first determination subunit is used for determining, for each single cell, that the fault detection result of the single cell is normal if the target ratio corresponding to the single cell is less than or equal to a first preset threshold value;

[0186] A second determination subunit is used to determine the fault detection result of the single cell according to the target ratio corresponding to the single cell in a future preset period of time and the preset threshold value if the target ratio corresponding to the single cell is greater than the first preset threshold value and less than the second preset threshold value;

[0187] The third determination subunit is used to determine that the fault detection result of the single cell is abnormal if the target ratio corresponding to the single cell is greater than or equal to the second preset threshold.

[0188] In one embodiment, the second determining subunit is specifically configured to:

[0189] Obtain the actual battery voltage of the single cell in a future preset time period; determine a third difference based on the predicted average battery voltage and the actual battery voltage of the single cell in the future preset time period; if the target ratio corresponding to the single cell in the future preset time period is less than or equal to the first preset threshold value, determine that the fault detection result of the single cell is normal; wherein the target ratio corresponding to the single cell in the future preset time period is the ratio of the third difference corresponding to the single cell to the standard deviation of the difference.

[0190] Each module in the above battery fault detection device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0191] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Fig.11As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store battery parameter data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a battery fault detection method is implemented.

[0192] Those skilled in the art will understand that Fig.11 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0193] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of the battery fault detection method described in any of the above embodiments are implemented.

[0194] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the battery fault detection method described in any of the above embodiments are implemented.

[0195] In one embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps of the battery fault detection method described in any of the above embodiments are implemented.

[0196] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0197] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0198] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the present application. It should be noted that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A battery failure detection method, characterized in that: The method comprises: Obtaining target battery parameters of the battery pack to be detected in a historical period and actual battery voltages of each single battery in the battery pack to be detected; wherein the target battery parameters include at least a target total voltage, a maximum target total voltage, a historical charging state, and a historical driving mileage; Inputting the target battery parameters into a voltage prediction model to obtain a predicted average battery voltage; The fault detection result of each single cell is determined according to the difference between the predicted average battery voltage and the actual battery voltage of each single cell.

2. The method according to claim 1, characterized in that The voltage prediction model includes a temporal convolutional neural network module, a bidirectional long short-term memory network module, an attention mechanism module and an activation function module; The step of inputting the target battery parameter into a voltage prediction model to obtain a predicted average battery voltage includes: Inputting the target battery parameters into the temporal convolutional neural network module to extract local features and obtain local battery feature parameters; Inputting the local battery characteristic parameters into the bidirectional long short-term memory network module to extract sequence characteristics to obtain sequence battery characteristic parameters; Inputting the sequence battery characteristic parameters into the attention mechanism module for attention allocation to obtain the attention battery characteristic parameters; The attention battery characteristic parameters are input into the activation function module for voltage prediction to obtain the predicted average battery voltage.

3. The method according to claim 2, characterized in that The temporal convolutional neural network module includes a convolution layer, a weight normalization layer, an activation function layer and a first descent layer connected in sequence; The bidirectional long short-term memory network module includes at least two memory network layers connected end to end; wherein each memory network layer includes a bidirectional long short-term memory network unit and a second descending layer connected in sequence.

4. The method according to claim 1, characterized in that Obtain the target battery parameters of the battery pack to be tested within a historical period, including: Obtain the historical total voltage of the battery group to be tested at each historical moment in the historical period; Obtain high-frequency components and low-frequency components of each historical total voltage; Taking each low-frequency component as the target total voltage; The maximum target total voltage is taken as the maximum target total voltage.

5. The method according to claim 1, characterized in that: Determining the fault detection result of each single cell according to the difference between the predicted average battery voltage and the actual battery voltage of each single cell includes: Determine a difference average value and a difference standard deviation according to a first difference between the predicted average battery voltage and the actual battery voltage of each single battery; The fault detection result of each single cell is determined based on the size relationship between the target ratio corresponding to each single cell and the preset threshold; wherein the target ratio is the ratio of the second difference corresponding to the single cell to the standard deviation of the difference; and the second difference is the difference between the first difference corresponding to the single cell and the average value of the difference.

6. The method according to claim 5, characterized in that The preset threshold includes a first preset threshold and a second preset threshold; the second preset threshold is greater than the first preset threshold; The step of determining the fault detection result of each single cell according to the size relationship between the target ratio corresponding to each single cell and the preset threshold value includes: For each single cell, if the target ratio corresponding to the single cell is less than or equal to the first preset threshold, it is determined that the fault detection result of the single cell is normal; If the target ratio corresponding to the single cell is greater than the first preset threshold and less than the second preset threshold, then determining the fault detection result of the single cell according to the target ratio corresponding to the single cell in a future preset period and the preset threshold; If the target ratio corresponding to the single cell is greater than or equal to the second preset threshold, it is determined that the fault detection result of the single cell is abnormal.

7. The method according to claim 6, characterized in that The step of determining the fault detection result of the single cell according to the target ratio corresponding to the single cell in a future preset period of time and the preset threshold value includes: Obtaining the actual battery voltage of the single battery within a preset time period in the future; determining a third difference value according to the predicted average battery voltage and the actual battery voltage of the single battery in a future preset time period; If the target ratio corresponding to the single cell in a future preset time period is less than or equal to the first preset threshold, the fault detection result of the single cell is determined to be normal; wherein the target ratio corresponding to the single cell in a future preset time period is the ratio of the third difference corresponding to the single cell to the standard deviation of the difference.

8. A battery fault detection device, characterized in that: The device comprises: An acquisition module, used for acquiring target battery parameters of a battery pack to be detected in a historical period and actual battery voltages of each single battery in the battery pack to be detected; A prediction module, used for inputting the target battery parameters into a voltage prediction model to obtain a predicted average battery voltage; wherein the voltage prediction model includes a temporal convolutional neural network module, a bidirectional long short-term memory network module, an attention mechanism module and an activation function module; The determination module is used to determine the fault detection result of each single cell according to the difference between the predicted average battery voltage and the actual battery voltage of each single cell.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images