Lithium battery fault identification method and system based on recursive graph and improved AlexNet model, and storage medium

Abstract

The application relates to the technical field of deep learning, in particular to a lithium battery fault identification method and system based on a recurrent graph and an improved AlexNet model and a storage medium. The method comprises the following steps: collecting one-dimensional time sequence data from various sensors of a lithium battery; preprocessing the one-dimensional time sequence data to generate original fault waveform data; converting the original fault waveform data into a recurrent graph; inputting the recurrent graph into a pre-trained improved AlexNet model to extract features, wherein the improved AlexNet model processes feature vectors in the recurrent graph through a multi-head self-attention mechanism, and inputs the processed feature vectors into a classification layer to predict the fault type of the battery; and obtaining a battery fault type prediction result output by the classification layer. The application aims to solve the problem of how to predict possible faults in a lithium battery.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of deep learning, in particular to a lithium battery fault identification method and system based on a recurrent graph and an improved AlexNet model and a storage medium. BACKGROUND

[0002] Lithium-ion batteries are one of the most important energy storage devices today, and are widely used in various fields from consumer electronics to electric vehicles to large-scale energy storage systems. Such batteries have the advantages of high energy density, long cycle life and high charging and discharging efficiency. However, as the use scenarios of lithium-ion batteries become more diverse and complex, various faults such as capacity degradation, internal short circuit, overcharging, overdischarging, thermal runaway, etc. may occur after long-term use. These faults not only seriously affect the life and performance of lithium batteries, but also may cause safety accidents, especially in electric vehicles and energy storage systems, the consequences of lithium battery faults are particularly serious. Therefore, how to accurately, timely and effectively identify lithium battery faults has become an important issue in battery research and application.

[0003] Currently, the traditional identification of lithium battery faults usually relies on manually set thresholds or simple physical models, which is difficult to adapt to fault characteristics under complex working conditions, and has the defect of low fault identification accuracy. Moreover, the traditional identification of lithium battery faults is based on the symptoms of the faults that have occurred, and lacks the ability to warn early and hidden faults, especially when the internal problems of the battery have not yet manifested as abnormal external parameters, early detection is not possible, and there is a defect of detection lag.

[0004] The above content is only used to assist in understanding the technical solutions of the application and does not represent the acknowledgement of the above content as prior art. SUMMARY

[0005] The main purpose of the present application is to provide a lithium battery fault identification method based on a recurrent graph and an improved AlexNet model, which aims to solve the problem of how to predict possible faults in lithium batteries.

[0006] To achieve the above purpose, the lithium battery fault identification method based on a recurrent graph and an improved AlexNet model provided by the present application comprises:

[0007] Collecting one-dimensional time series data from various sensors of the lithium battery;

[0008] Preprocessing the one-dimensional time series data to generate original fault waveform data;

[0009] Converting the original fault waveform data into a recurrent graph;

[0010] input the recursive graph into a pre-trained improved AlexNet model for feature extraction, wherein the improved AlexNet model processes feature vectors in the recursive graph through a multi-head self-attention mechanism, and inputs the processed feature vectors into a classification layer for battery fault type prediction;

[0011] obtain a battery fault type prediction result output by the classification layer.

[0012] Optionally, the step of converting the original fault waveform data into a recursive graph comprises:

[0013] calculating the Euclidean distance between each point in the original fault waveform data;

[0014] calculating the difference between a preset threshold and the Euclidean distance, and generating a recursive graph according to the difference;

[0015] wherein each element R in the recursive graph ij is defined as follows:

[0016] R ij = Θ (ε - d ij )

[0017] wherein Θ is a step function, R ij = 1 when ∈ ≥ d ij , and R ij = 0 otherwise.

[0018] Optionally, the convolutional layers in the improved AlexNet model are frozen, and the hyperparameters in the fully connected layers in the improved AlexNet model are fine-tuned.

[0019] Optionally, the classification layer in the battery fault type prediction is calculated according to the following formula:

[0020]

[0021] wherein, is the predicted battery fault type, M is the feature vector output by the multi-head self-attention mechanism, W is the weight matrix, and b is the bias term.

[0022] Optionally, after the step of obtaining the battery fault type prediction result output by the classification layer, the method further comprises:

[0023] generating a battery health status report, wherein the battery health status report comprises the current operating status of the battery, potential fault risks, and maintenance recommendations.

[0024] Optionally, the preprocessing comprises denoising and normalization processing.

[0025] Optionally, the formula for normalization processing is:

[0026]

[0027] wherein, m i is a data point in one-dimensional time series data, max(M) and min(M) represent the maximum and minimum values of the time series data respectively, and M is the input time series.

[0028] In addition, to achieve the above object, the application further provides a lithium battery fault identification system, comprising a memory, a processor and a lithium battery fault identification program based on a recurrent graph and an improved AlexNet model stored on the memory and executable on the processor, wherein the lithium battery fault identification program based on the recurrent graph and the improved AlexNet model implements the steps of the lithium battery fault identification method based on the recurrent graph and the improved AlexNet model as described above when executed by the processor.

[0029] In addition, to achieve the above object, the application further provides a computer readable storage medium having a lithium battery fault identification program based on a recurrent graph and an improved AlexNet model stored thereon, wherein the lithium battery fault identification program based on the recurrent graph and the improved AlexNet model implements the steps of the lithium battery fault identification method based on the recurrent graph and the improved AlexNet model as described above when executed by a processor.

[0030] The application has at least the following beneficial effects:

[0031] 1. Traditional lithium battery fault identification often relies on manually set thresholds or simple physical models, which is difficult to adapt to complex working conditions and leads to low accuracy of fault identification. Therefore, the application introduces a recurrent graph and an improved deep learning model (AlexNet) and combines a multi-head attention mechanism to enhance the extraction of complex fault features and improve the accuracy of fault identification.

[0032] 2. Traditional fault detection methods are usually based on existing fault symptoms and lack the ability to detect early and hidden faults. In particular, when internal battery problems have not yet manifested as external parameter abnormalities, existing methods are difficult to detect early. Therefore, the application can effectively identify early signs of failure through the time series feature capturing ability of the recurrent graph, thereby achieving early fault detection.

[0033] 3. As the application scenarios of lithium batteries become more diverse, the working environment of the battery becomes increasingly complex, such as temperature and load changes. Traditional methods have obvious limitations in dealing with complex working conditions. Therefore, the application introduces a multi-head attention mechanism to enhance the ability to capture local important features, allowing it to maintain high accuracy in fault detection under complex working conditions. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 The hardware running environment architecture diagram of the lithium battery fault identification system involved in the embodiment of the present application is shown in the figure.

[0035] Figure 2 The flowchart of the first embodiment of the lithium battery fault identification method based on the recurrent graph and the improved AlexNet model of the present application is shown in the figure.

[0036] Figure 3 The architecture diagram of the improved AlexNet model involved in the embodiment of the present application is shown in the figure.

[0037] Figure 4 The flowchart of the second embodiment of the lithium battery fault identification method based on the recurrent graph and the improved AlexNet model of the present application is shown in the figure.

[0038] The implementation, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0039] In order to better understand the above technical solutions, the exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0040] As an implementation scheme, Figure 1 The hardware running environment architecture diagram of the lithium battery fault identification system involved in the embodiment of the present application is shown in the figure.

[0041] As Figure 1 shown, the lithium battery fault identification system can include a processor 1001, such as a CPU, a memory 1005, a user interface 1003, a network interface 1004, and a communication bus 1002. The communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 can include a display screen (Display), an input unit such as a keyboard (Keyboard), and an optional user interface 1003 can also include a standard wired interface, a wireless interface. The network interface 1004 can optionally include a standard wired interface, a wireless interface (such as a WI-FI interface). The memory 1005 can be a high-speed RAM memory, or a stable memory (non-volatile memory) such as a magnetic disk memory. The memory 1005 can also be an independent storage device from the aforementioned processor 1001.

[0042] Those skilled in the art can understand that Figure 1 The lithium battery fault identification system architecture shown in the foregoing embodiments does not constitute a limitation on the lithium battery fault identification system, and can include more or fewer components than those shown, or combine certain components, or different component arrangements.

[0043] As shown in Figure 1 The memory 1005 as a storage medium can include an operating system, a network communication module, a user interface module, and a lithium battery fault identification program based on a recurrent graph and an improved AlexNet model. Among them, the operating system is a program that manages and controls the hardware and software resources of the lithium battery fault identification system, and the running of the lithium battery fault identification program based on the recurrent graph and the improved AlexNet model and other software or programs.

[0044] In the lithium battery fault identification system shown in Figure 1 In the lithium battery fault identification system shown in

[0045] In this embodiment, the lithium battery fault identification system includes a memory 1005, a processor 1001, and a lithium battery fault identification program based on a recurrent graph and an improved AlexNet model stored on the memory and executable on the processor, wherein:

[0046] When the processor 1001 calls the lithium battery fault identification program based on the recurrent graph and the improved AlexNet model stored in the memory 1005, the following operations are performed:

[0047] Collecting one-dimensional time series data from various sensors of the lithium battery;

[0048] Pretreating the one-dimensional time series data to generate original fault waveform data;

[0049] Converting the original fault waveform data into a recurrent graph;

[0050] Inputting the recurrent graph into a pre-trained improved AlexNet model for feature extraction, wherein the improved AlexNet model processes feature vectors in the recurrent graph through a multi-head self-attention mechanism, and inputs the processed feature vectors into a classification layer for battery fault type prediction;

[0051] Obtaining the battery fault type prediction result output by the classification layer.

[0052] When the processor 1001 invokes the lithium battery fault identification program based on the recurrent graph and the improved AlexNet model stored in the memory 1005, the following operations are performed:

[0053] The Euclidean distance between each point in the original fault waveform data is calculated.

[0054] The difference between the preset threshold and the Euclidean distance is calculated, and the recurrent graph is generated according to the difference.

[0055] When the processor 1001 invokes the lithium battery fault identification program based on the recurrent graph and the improved AlexNet model stored in the memory 1005, the following operations are performed:

[0056] A battery health status report is generated, which includes the current operating status of the battery, potential fault risk and maintenance recommendations.

[0057] Based on the hardware architecture of the lithium battery fault identification system based on the above deep learning technology, the embodiments of the lithium battery fault identification method based on the recurrent graph and the improved AlexNet model are proposed.

[0058] First embodiment

[0059] Reference Figure 2 In the first embodiment, the lithium battery fault identification method based on the recurrent graph and the improved AlexNet model includes the following steps:

[0060] Step S10, collecting one-dimensional time series data from various sensors of the lithium battery;

[0061] In this embodiment, the system first collects key one-dimensional time series data in the lithium battery.

[0062] Optionally, the one-dimensional time series data includes parameters such as voltage, current and temperature.

[0063] Step S20, preprocessing the one-dimensional time series data to generate original fault waveform data;

[0064] Further, the one-dimensional time series data is preprocessed to eliminate noise interference and unify the data scale, thereby obtaining high-quality original fault waveform data.

[0065] Optionally, the preprocessing method includes denoising and normalization processing.

[0066] Further and optionally, the formula of the normalization processing is:

[0067]

[0068] Wherein, m iis a data point in one-dimensional time series data, max(M) and min(M) represent the maximum and minimum values of the time series data respectively, and M is the input time series.

[0069] Step S30, converting the original fault waveform data into a recurrence plot;

[0070] In this embodiment, the recurrence plot is used for feature extraction, and the system converts one-dimensional time series data into a two-dimensional image to show the similarity and repeatability of the data at different time points. The recurrence plot can effectively capture potential fault features in the operation of the battery, especially in the process of gradual development of the fault, and it can reveal early signals of the fault.

[0071] It should be noted that the recurrence plot can effectively capture the time series features in the operation of the lithium battery, help to extract key time-dependent information in the fault evolution process, and improve the recognition ability of early faults and hidden faults.

[0072] Step S40, inputting the recurrence plot into a pre-trained improved AlexNet model for feature extraction, wherein the improved AlexNet model processes the feature vectors in the recurrence plot through a multi-head self-attention mechanism, and inputs the processed feature vectors into a classification layer for battery fault type prediction.

[0073] In this embodiment, the converted recurrence plot is input into a pre-trained improved AlexNet model for feature extraction.

[0074] The AlexNet model is a convolutional neural network, and the network structure mainly consists of convolutional layers, pooling layers, local response normalization layers, fully connected layers and activation function modules.

[0075] Convolutional Layers: Convolutional layers are responsible for extracting local features in the recurrence plot, such as edges, textures, etc. Convolution operation realizes feature extraction by sliding convolution kernel and input image region for element-wise weighted sum. The specific convolution calculation formula is:

[0076]

[0077] where I is the input recurrence plot, W is the convolution kernel, and x, y are the positions in the feature map. Convolution operation extracts local features from the recurrence plot by sliding the convolution kernel.

[0078] Activation Function: After the convolutional layer, AlexNet uses ReLU (Rectified Linear Unit) as the activation function, which aims to introduce nonlinearity so that the network can learn complex patterns. The calculation formula of ReLU is as follows:

[0079] f(x) = max(0, x)

[0080] The non-linear characteristics of ReLU enable the network to learn more rich features in different feature spaces, while avoiding the gradient vanishing problem of traditional activation functions such as sigmoid and tanh. max(0, x): represents the maximum value between the input value x and 0. If x is less than 0, the ReLU function outputs 0; if x is greater than 0, it directly outputs x.

[0081] Pooling Layers: Pooling layers are used to down-sample feature maps and reduce computational complexity. The commonly used is Max Pooling, whose formula is:

[0082]

[0083] Where R represents the size of the pooling window.

[0084] It should be noted that the improved AlexNet model after pre-training needs to be pre-trained on a large-scale image dataset to have good image feature extraction capability.

[0085] In this embodiment, referring to Figure 3 The improved AlexNet model architecture diagram shown in the figure, in order to further improve the feature extraction capability of the model under complex working conditions, the traditional AlexNet model architecture is improved in this embodiment, and an improved AlexNet model architecture feature is further provided:

[0086] A multi-head self-attention mechanism (MSA) is introduced in the output layer of the model. MSA can simultaneously focus on different regions in the recursive graph through multiple parallel attention heads, and automatically identify the most critical features for fault identification. Multi-head self-attention mechanism helps the model to capture more representative fault patterns under complex working conditions, and improves the accuracy and robustness of fault detection.

[0087] The calculation steps of the multi-head self-attention mechanism are as follows:

[0088] First, each attention head calculates the attention weight based on the input query vector (Query), key vector (Key) and value vector (Value), and the formula is as follows:

[0089]

[0090] Where Q, K and V are query, key and value vectors, respectively, d kis the dimension of the vector. Through the self-attention mechanism, the model can assign different weights to different regions in the recursive graph, so as to more accurately capture the fault features.

[0091] Secondly, multiple attention heads process different feature subspaces in parallel, and finally the outputs of these heads are spliced and linearly transformed to generate the final output, as follows:

[0092] MSA(Q, K, V) = Concat(head1,..., head h )W O

[0093] Where head i is the output of each attention head, and w o is the linear transformation matrix.

[0094] Step S50, obtaining the battery fault type prediction result output by the classification layer.

[0095] In this embodiment, the battery fault type prediction result output by the classification layer is obtained, so as to analyze whether the lithium battery has a potential fault risk according to the battery fault type prediction result.

[0096] In the technical scheme provided in this embodiment, by introducing the recursive graph and the improved deep learning model (AlexNet), combining the multi-head attention mechanism, the extraction ability of complex fault features is enhanced, and the accuracy of fault recognition is improved; through the time sequence feature capturing ability of the recursive graph, the early signs of fault occurrence can be effectively identified, so as to realize the detection of early faults; in addition, by introducing the multi-head attention mechanism, the capturing ability of local important features is strengthened, so that it can still maintain a high accuracy of fault detection under complex working conditions.

[0097] Second embodiment

[0098] As an optional embodiment, the present embodiment provides a method for generating a recursive graph, which is based on the first embodiment, and refers to Figure 4 , and the step S20 specifically comprises:

[0099] Step S21, calculating the Euclidean distance between each point in the original fault waveform data;

[0100] Specifically, in this embodiment, first, the Euclidean distance between each point in the time series data is calculated, which is defined as follows:

[0101] d ij =||m i -m j ||

[0102] Where m i and m jdenotes the Euclidean distance.

[0103] In step S22, a difference between the preset threshold and the Euclidean distance is calculated, and a recurrence graph is generated according to the difference.

[0104] wherein each element R ij is defined as follows:

[0105] R ij = Θ (e-d ij )

[0106] wherein Θ is a step function, when e≥d ij , R ij = 1, otherwise R ij = 0.

[0107] Third embodiment

[0108] As an optional embodiment, in this embodiment, based on the first embodiment, the first few layers of the improved AlexNet model are frozen by the method of transfer learning, and the hyperparameters in the fully connected layer of the improved AlexNet model are fine-tuned.

[0109] Since the lithium battery fault detection is different from the general image classification task, the last few layers of the model (such as the fully connected layer) are fine-tuned to adapt to the specific fault detection task. After freezing the first few layers of the convolutional layer, only the subsequent fully connected layer is fine-tuned, and the fine-tuned features are calculated by the formula of the fully connected layer:

[0110]

[0111] wherein, is the predicted battery fault type, M is the feature vector output by the multi-head self-attention mechanism, W is the weight matrix, and b is the bias term.

[0112] Fourth embodiment

[0113] As an optional embodiment, in this embodiment, based on the first embodiment, after the step S40, the method further comprises:

[0114] In step S50, a battery health status report is generated, and the battery health status report includes the current operating state of the battery, the potential fault risk, and the maintenance suggestion.

[0115] In this embodiment, after the battery fault detection result prediction is completed, the system generates a battery health status report, which includes the current operating state of the battery, the potential fault risk, and the maintenance suggestion, helping users optimize the use of the battery, prevent the fault from further deteriorating, and prolong the service life of the battery.

[0116] Further, it is understood by those skilled in the art that all or part of the processes in the method of the above embodiments can be completed by instructing the relevant software by a computer program. The computer program includes program instructions, and the computer program can be stored in a storage medium, which is a computer readable storage medium. The program instructions are executed by at least one processor in the lithium battery fault identification system to realize the process steps of the above method embodiments.

[0117] Therefore, the application also provides a computer readable storage medium, which stores a lithium battery fault identification program based on a recurrent graph and an improved AlexNet model. The lithium battery fault identification program based on the recurrent graph and the improved AlexNet model realizes each step of the lithium battery fault identification method based on the recurrent graph and the improved AlexNet model when executed by a processor.

[0118] The computer readable storage medium can be a U disk, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk, and various computer readable storage media that can store program codes.

[0119] It should be noted that the storage medium provided by the embodiments of the application is a storage medium used to implement the method of the embodiments of the application. Therefore, based on the method introduced in the embodiments of the application, those skilled in the art can understand the specific structure and modification of the storage medium, and therefore it is not repeated here. Any storage medium used by the method of the embodiments of the application belongs to the scope of protection of the application.

[0120] Those skilled in the art should understand that the embodiments of the application can be provided as a method, a system or a computer program product. Therefore, the application can be in the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program codes.

[0121] The application is described with reference to flowcharts and / or block diagrams of the method, device (system) and computer program product according to the embodiments of the application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general purpose computer, a special purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a machine that implements the functions described in the flowcharts and / or block diagrams.Figure 1 one or more processes and / or blocks Figure 1 means for performing the function specified by the block or blocks.

[0122] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the Figure 1 one or more processes and / or blocks Figure 1 means for performing the function specified by the block or blocks.

[0123] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the Figure 1 one or more processes and / or blocks Figure 1 means for performing the function specified by the block or blocks.

[0124] It is noted that in the claims the word "comprising" does not exclude not having other parts than those specified in the claim. The word "a" or "an" preceding a plural reference to elements does not exclude a larger number for these elements than specified in the claim. The application can be implemented by means of both hardware and software, and any combination thereof. The use of the word "about" in relation to a numerical value specifies the precision of that value, and the use of the word "substantially" in relation to a geometrical or spatial relationship specifies the closeness of that relationship. The word "comprising" does not exclude other elements being present in addition to those elements recited. The word "a" or "an" preceding a list of elements does not exclude the presence of more than one of such element. The application can be implemented by means of both hardware and software, and any combination thereof. In the claims, the word "comprising" does not exclude other elements being present in addition to those elements recited. The word "a" or "an" preceding the commencement of a list of elements does not exclude the presence of more than one of such element. The word "first", "second", "third", etc. does not imply any order but is used for naming purposes only. The use of numbering or letters in the claims is not to be construed as limiting the scope of the claims to a certain number of such elements. The use of the terms first, second and third, etc. does not denote any order. These terms can be understood as naming.

[0125] Although the preferred embodiments of the application have been described, those skilled in the art will recognize that many modifications and variations of the described implementation can be made without departing from the spirit or scope of the application. Accordingly, it is intended that there be included within the scope of the application, whatever falls within the scope of the appended claims and their equivalents.

[0126] Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A lithium battery fault identification method based on a recursive graph and an improved AlexNet model, characterized in that, The method comprises the following steps: Collecting one-dimensional time series data from various sensors of a lithium battery; Preprocessing the one-dimensional time series data to generate raw fault waveform data; Converting the raw fault waveform data into a recurrent graph; Inputting the recurrent graph into a pre-trained improved AlexNet model for feature extraction, wherein the improved AlexNet model processes feature vectors in the recurrent graph through a multi-head self-attention mechanism and inputs the processed feature vectors into a classification layer for battery fault type prediction; Obtaining the battery fault type prediction result output by the classification layer; The step of converting the raw fault waveform data into a recurrent graph comprises: Calculating the Euclidean distance between each point in the raw fault waveform data; Calculating the difference between the preset threshold and the Euclidean distance, and generating a recurrent graph according to the difference; wherein each element of the recurrence graph is defined as follows: ； where Θ is a step function, when ≥ , , otherwise ; The convolution layer in the improved AlexNet model is frozen, and the hyperparameters in the fully connected layer in the improved AlexNet model are fine-tuned; The calculation formula for battery fault type prediction in the classification layer is as follows: ； wherein, For the predicted battery failure type, M is a feature vector output by the multi-head self-attention mechanism, W is a weight matrix, and b is a bias term.

2. The method of claim 1, wherein, After the step of obtaining the battery fault type prediction result output by the classification layer, the method further comprises: Generating a battery health status report, which includes the current operating status, potential fault risk, and maintenance recommendations of the battery.

3. The method of claim 1, wherein, The preprocessing includes denoising and normalization processing.

4. The method of claim 3, wherein, The formula for the normalization processing is: ； wherein is a data point in one-dimensional time series data, max(M) and min(M) represent the maximum and minimum values of the time series data, respectively, and M is the input time series.

5. A lithium battery failure identification system characterized by, The lithium battery fault identification system comprises a memory, a processor, and a lithium battery fault identification program based on a recurrent graph and an improved AlexNet model stored on the memory and executable on the processor, wherein the lithium battery fault identification program based on the recurrent graph and the improved AlexNet model implements the steps of the lithium battery fault identification method based on the recurrent graph and the improved AlexNet model according to any one of claims 1 to 4 when executed by the processor.

6. A computer readable storage medium characterized by, The computer-readable storage medium stores a lithium battery fault identification program based on a recurrent graph and an improved AlexNet model, wherein the lithium battery fault identification program based on the recurrent graph and the improved AlexNet model implements the steps of the lithium battery fault identification method based on the recurrent graph and the improved AlexNet model according to any one of claims 1 to 4 when executed by the processor.

Images