Fault processing method and device, equipment and storage medium

Abstract

The invention discloses a fault processing method and apparatus, a device and a storage medium. The method comprises the steps of obtaining to-be-processed fault information; performing feature extraction on the to-be-processed fault information based on a first fault processing model to obtain fault feature information; and performing identification processing on the fault feature information based on a second fault processing model to obtain a fault identification result. The method can effectively improve the fault recognition rate of the battery pack.

Description

Technical Field

[0001] The present invention relates to the field of fault handling, and particularly to a fault handling method, device, equipment and storage medium. Background Art

[0002] The voltage of a parallel battery pack is a key parameter of a power battery. Battery faults may lead to out-of-control of the battery pack, and if not solved in time, it may even cause combustion and explosion accidents. In related technologies, generally, the voltage and current at both ends of the battery are measured to determine whether the battery has a short circuit or a short circuit problem. However, for a battery pack, there may be more types of faults, such as single-cell battery fault problems. Through the above method, different battery pack faults cannot be identified, resulting in a low fault recognition rate of the battery pack. Summary of the Invention

[0003] Embodiments of the present invention provide a fault handling method, device, equipment and storage medium, aiming to effectively improve the fault recognition rate of an energy storage device.

[0004] In a first aspect, embodiments of the present invention provide a fault handling method, the method including:

[0005] Obtain fault information to be processed;

[0006] Extract feature information of the fault based on a first fault handling model to obtain fault feature information;

[0007] Perform recognition processing on the fault feature information based on a second fault handling model to obtain a fault recognition result.

[0008] Optionally, the extracting feature information of the fault based on the first fault handling model to obtain fault feature information includes:

[0009] Update the fault information to be processed based on the weight coefficients corresponding to each target data feature in the fault information to be processed;

[0010] Process the updated fault information to be processed based on the first fault handling model to obtain the fault feature information.

[0011] Optionally, the weight coefficients are obtained based on the following steps:

[0012] Obtain multiple target data features of the fault information to be processed;

[0013] Calculate the similarity between the target data feature and each data feature in the fault information to be processed;

[0014] Determine the weight coefficients corresponding to each target data feature according to the similarity corresponding to each target data feature.

[0015] Optionally, before obtaining the fault information to be processed, the method further includes:

[0016] Inputting the energy storage device data to be detected into an encoding layer and a decoding layer corresponding to the encoding layer in sequence to obtain reconstructed energy storage device data;

[0017] Determining an abnormal risk coefficient of the energy storage device data based on a reconstruction error between the energy storage device data and the reconstructed energy storage device data;

[0018] When the abnormal risk coefficient is greater than an error threshold, using an output result of the encoding layer as the fault information to be processed.

[0019] Optionally, before using the output result of the encoding layer as the fault information to be processed when the abnormal risk coefficient is greater than the error threshold, the method further includes:

[0020] Obtaining a plurality of verified energy storage device data;

[0021] Inputting the verified energy storage device data into the encoding layer and the decoding layer in sequence to obtain reconstructed verified energy storage device data corresponding to the verified energy storage device data;

[0022] Determining the error threshold based on all the verified energy storage device data and the reconstructed verified energy storage device data.

[0023] Optionally, determining the error threshold based on all the verified energy storage device data and the reconstructed verified energy storage device data includes:

[0024] For each of the verified energy storage device data, determining a verification reconstruction error between each of the verified energy storage device data and the corresponding reconstructed verified energy storage device data;

[0025] Based on the verification reconstruction errors, counting the number of verification reconstruction errors in each error interval;

[0026] Determining a target error interval according to the number of verification reconstruction errors in each error interval, and determining the error threshold according to the target error interval.

[0027] Optionally, the second fault processing model is obtained based on the following steps:

[0028] Obtaining a plurality of training fault information corresponding to different fault recognition results, and determining positive and negative categories corresponding to the training fault information based on an initial second fault processing model;

[0029] For the target training fault information of each positive and negative category, determine the importance of the target training fault information according to the difference between the target training fault information and the central estimation point of all training fault information belonging to the same positive and negative category;

[0030] According to the importance, the weight value of the positive and negative categories, and the optimization model, determine the adjustment variable based on the initial second fault handling model;

[0031] Optimize the initial second fault handling model according to the adjustment variable to obtain the second fault handling model.

[0032] In a second aspect, an embodiment of the present invention provides a fault handling device, and the fault handling device includes:

[0033] An acquisition module, configured to acquire fault information to be processed;

[0034] An extraction module, configured to perform feature extraction on the fault information to be processed based on a first fault handling model to obtain fault feature information;

[0035] A processing module, configured to perform identification processing on the fault feature information based on a second fault handling model to obtain a fault identification result.

[0036] Preferably, the extraction module performing feature extraction on the fault information to be processed based on a first fault handling model to obtain fault feature information includes:

[0037] Update the fault information to be processed based on the weight coefficients corresponding to the target data features in the fault information to be processed;

[0038] Perform processing on the updated fault information to be processed based on the first fault handling model to obtain the fault feature information;

[0039] Preferably, the weight coefficient is obtained by the extraction module based on the following steps:

[0040] Acquire a plurality of target data features in the fault information to be processed;

[0041] Calculate the similarity between the target data features and each data feature in the fault information to be processed;

[0042] Determine the weight coefficients corresponding to the target data features according to the similarities corresponding to the target data features;

[0043] Preferably, before the acquisition module acquires the fault information to be processed, it further includes:

[0044] Input the energy storage device data to be detected into the encoding layer and the decoding layer corresponding to the encoding layer in sequence to obtain the reconstructed energy storage device data;

[0045] Determine the abnormal risk coefficient of the energy storage device data based on the reconstruction error between the energy storage device data and the reconstructed energy storage device data;

[0046] When the abnormal risk coefficient is greater than the error threshold, use the output result of the encoding layer as the fault information to be processed;

[0047] Preferably, before the obtaining module uses the output result of the encoding layer as the fault information to be processed when the abnormal risk coefficient is greater than the error threshold, it further includes:

[0048] Obtain multiple verified energy storage device data;

[0049] Input the verified energy storage device data into the encoding layer and the decoding layer in sequence to obtain the reconstructed verified energy storage device data corresponding to the verified energy storage device data;

[0050] Determine the error threshold based on all the verified energy storage device data and the reconstructed verified energy storage device data;

[0051] Preferably, when the obtaining module determines the error threshold based on all the verified energy storage device data and the reconstructed verified energy storage device data, it includes:

[0052] For each of the verified energy storage device data, determine the verification reconstruction error between each of the verified energy storage device data and the corresponding reconstructed verified energy storage device data;

[0053] Based on the verification reconstruction error, count the number of verification reconstruction errors in each error interval;

[0054] Determine the target error interval according to the number of verification reconstruction errors in each error interval, and determine the error threshold according to the target error interval;

[0055] Preferably, the second fault processing model is obtained by the processing module based on the following steps:

[0056] Obtain multiple training fault information corresponding to different fault recognition results, and determine the positive and negative categories corresponding to the training fault information based on the initial second fault processing model;

[0057] For the target training fault information of each positive and negative category, determine the importance of the target training fault information according to the difference between the target training fault information and the central estimation point of all the training fault information belonging to the same positive and negative category;

[0058] Determine an adjustment variable based on the initial second fault handling model according to the importance, the weight values of the positive and negative categories, and the optimization model;

[0059] Optimize the initial second fault handling model according to the adjustment variable to obtain the second fault handling model.

[0060] In a third aspect, an embodiment of the present invention further provides a fault handling device, including a memory storing multiple instructions; a processor loads instructions from the memory to execute the steps of any one of the fault handling methods provided by the embodiments of the present invention.

[0061] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium storing multiple instructions, and the instructions are suitable for being loaded by a processor to execute the steps of any one of the fault handling methods provided by the embodiments of the present invention.

[0062] The present invention first obtains fault information to be processed; extracts fault feature information from the fault information to be processed based on a first fault handling model; and performs identification processing on the fault feature information based on a second fault handling model to obtain a fault identification result. By processing the fault information to be identified through the first fault handling model, more prominent fault feature information is obtained, and then the second fault handling model is used to perform identification processing on the more prominent fault feature information, so that the fault types of the energy storage device that cannot be directly determined through the energy storage device data can be distinguished, and a more accurate fault identification result can be obtained, thereby improving the accuracy of identifying the faults of the energy storage device. Description of the Drawings

[0063] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for description in the embodiments. Obviously, the following drawings are only some embodiments of the present invention, and those skilled in the art can obtain other drawings without creative efforts based on these drawings.

[0064] Figure 1 is a schematic flowchart of an embodiment of the fault handling method provided in the embodiment of the present invention;

[0065] Figure 2 is a schematic flowchart of another embodiment of the fault handling method provided in the embodiment of the present invention;

[0066] Figure 3 is a schematic circuit diagram of a simulated energy storage device provided in the embodiment of the present invention;

[0067] Figure 4 is a schematic flowchart of yet another embodiment of the fault handling method provided in the embodiment of the present invention;

[0068] Figure 5 It is a schematic flow chart of obtaining reconstruction error provided in an embodiment of the present invention;

[0069] Figure 6 It is a schematic application flow chart of the fault handling method provided in an embodiment of the present invention;

[0070] Figure 7 It is a schematic structural diagram of the fault handling device provided in an embodiment of the present invention;

[0071] Figure 8 It is a schematic structural diagram of the fault handling equipment provided in an embodiment of the present invention. Detailed implementation manners

[0072] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts belong to the scope of protection of the present invention. At the same time, in the description of the embodiments of the present invention, terms such as "first" and "second" are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more features. In the description of the embodiments of the present invention, the meaning of "a plurality" is two or more, unless otherwise clearly and specifically defined.

[0073] An embodiment of the present invention provides a fault handling method, device, equipment and storage medium.

[0074] Specifically, this embodiment will be described from the perspective of the fault handling device, which can be specifically integrated in the fault handling equipment, that is, the fault handling method in the embodiment of the present invention can be executed by the fault handling equipment.

[0075] The following will be described in detail with reference to the accompanying drawings. In this embodiment, the execution entity is the fault handling equipment as an example. It should be noted that the description order of the following embodiments does not limit the preferred order of the embodiments. Although the logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown in the drawings.

[0076] According to the description of the background art of the present invention, in the related art, there are many types of energy storage device failures, and the type of energy storage device failure cannot be determined by measuring the voltage and current at both ends of the energy storage device, resulting in a low failure recognition rate of the energy storage device.

[0077] To solve the above problems, the present invention discloses a fault handling method. Please refer to Figure 1, the specific process of this fault handling method can be as follows in steps S10 to S40, where:

[0078] Step S10, obtain the fault information to be processed;

[0079] In this embodiment, the main body executing the fault handling method can be a fault handling device, which is used to obtain the fault information to be processed of the target energy storage device with a fault. The fault information to be processed is obtained based on the energy storage device data of the target energy storage device with a fault risk. Based on the fault information to be processed, the fault type of the target energy storage device is detected, and a fault recognition result capable of identifying the energy storage device is obtained. The target energy storage device is the monitored energy storage device, and the energy storage device can be a parallel battery pack composed of multiple single cells connected in parallel. For a parallel battery pack, when a fault occurs, the changes in the total current and terminal voltage are not obvious. When a single cell fails, these faults of the single cell cannot be detected through the total current and terminal voltage. Moreover, the cost of installing current sensors on each branch of the single cell is too high, so generally the method of increasing current sensors is not selected to detect single cell faults, resulting in a decrease in the fault recognition rate of the energy storage device.

[0080] Optionally, the energy storage device can be a power source on an electric vehicle. Therefore, the fault handling device can be the control system of the electric vehicle or a server on a remote platform, etc.

[0081] During the operation of the target energy storage device, the energy storage device data of the target energy storage device can be collected as the energy storage device data to be detected at preset time intervals, or in response to an energy storage device detection instruction, or when other energy storage device detection conditions are met, through sensors, etc. When it is detected that the energy storage device data to be detected may have a fault, it is processed as the fault information to be processed. The collected energy storage device data can include the terminal voltage, total current, and remaining charge (SOC data) of the energy storage device, etc.

[0082] Optionally, the energy storage device data to be detected can be encoded to obtain the fault feature information. The encoding layer can be composed of bidirectional long short-term memory (Bi-LSTM) units. The encoding layer can be set with two layers. The first layer has 64 Bi-LSTM units, and the second layer has 32 Bi-LSTM units. The energy storage device data to be detected is encoded through the encoding layer to obtain the fault information to be processed.

[0083] Step S20, perform feature extraction on the fault information to be processed based on the first fault processing model to obtain fault feature information;

[0084] In this embodiment, after obtaining the energy storage device data to be detected, the energy storage device is input into the first fault processing model to extract fault feature information. The first fault processing model extracts fault feature information with stronger information based on model processing.

[0085] Step S30: Identify and process the fault feature information based on the second fault processing model to obtain a fault identification result.

[0086] In this embodiment, it is necessary to construct and train a second fault processing model in advance. The second fault processing model can be a classification model, and the classification model can adopt a machine learning model, etc. When training the second fault processing model, it is necessary to obtain training energy storage device data collected by several energy storage devices under different fault types and process them into training fault information. These training fault information can come from one or more energy storage devices, and each piece of training fault information also carries a label of the fault type of the corresponding energy storage device when the training fault information is collected. These training fault information also need to be input into the second fault processing model to extract the corresponding fault feature information. The fault feature information and the corresponding fault type labels corresponding to several pieces of training fault information will be divided into a training data set, a validation data set, and a test data set and input into the initial second fault processing model for the initial second fault processing model to learn and train. When the initial second fault processing model processes these training fault information, it continuously learns and optimizes the relationship between the fault feature information corresponding to the training fault information and the fault type label, so as to be able to map the fault feature information corresponding to the fault information to be processed and obtain the fault type label of the fault feature information corresponding to the fault information to be processed, thereby obtaining the fault type of the target energy storage device corresponding to the fault information to be processed, and further obtaining a fault identification result.

[0087] In the technical solution disclosed in this embodiment, obtain the fault information to be processed; extract features from the fault information to be processed based on the first fault processing model to obtain fault feature information; identify and process the fault feature information based on the second fault processing model to obtain a fault identification result. By processing the fault information to be identified through the first fault processing model, more prominent fault feature information can be obtained, and then the second fault processing model is used to identify and process the more prominent fault feature information, so that the fault types of energy storage devices that cannot be directly determined through energy storage device data can be distinguished, and a more accurate fault identification result can be obtained, thereby improving the accuracy of identifying energy storage device faults.

[0088] Further, step S20 includes:

[0089] Update the fault information to be processed based on the weight coefficients corresponding to the target data features in the fault information to be processed;

[0090] Process the updated fault information to be processed based on the first fault processing model to obtain the fault feature information.

[0091] In this embodiment, the fault information to be processed is composed of multiple data features. All or part of the data features are used as target data features, and weight values are set for them to indicate the importance of the target data features in the fault information to be processed. The fault information to be processed is updated based on the weight coefficients corresponding to the target features, so that the important target data features in the fault information to be processed are more prominent. Convolution processing is performed on the updated fault information to be processed in the first fault processing model. Convolution processing can change the dimension of the fault information to be processed, enhance the amount of information contained in each data feature, and improve the accuracy of the second fault model in identifying energy storage device faults.

[0092] Further, the weight coefficient is obtained by the extraction module based on the following steps:

[0093] Obtain multiple target data features of the fault information to be processed;

[0094] Calculate the similarity between the target data features and each data feature in the fault information to be processed;

[0095] Determine the weight coefficients corresponding to each target data feature according to the similarities corresponding to each target data feature.

[0096] In this embodiment, all data features in the fault information to be processed can be used as target data features, or data features can be selected from the data features as target data features according to other requirements. The first fault processing model can also have a self-attention channel, that is, the first fault processing model can be a convolutional model with a self-attention channel. For example, the convolutional model can be a SENet network with a self-attention channel. The number of convolutional kernels that can be set in the SENet network is 16, which is equivalent to being divided into 16 channels, and the stride is set to 2. Convolution processing is performed on the fault information to be processed based on the convolutional model. Convolution processing can change the dimension of the fault information to be processed, enhance the amount of information contained in each data feature, and improve the accuracy of the second fault model in identifying energy storage device faults.

[0097] Specifically, first, the to-be-processed fault information is input into the self-attention channel. The self-attention channel can process each data feature in the to-be-processed fault information as the target data feature, calculate the similarity between the target data feature and each data feature in the to-be-processed fault information. The higher the similarity, the greater the association between the target data feature and other feature data, and the more important the target data feature. Therefore, the similarity can represent the importance of the data feature, and the weight coefficient corresponding to the target data feature is determined according to the similarity. Based on the weight coefficient, the target data feature can be linearly updated, that is, the weight coefficient is multiplied by the target data feature to obtain the updated target data feature. After the updated target data features are combined or the target data features in the to-be-processed fault information are replaced, the to-be-processed fault information can be updated. In this way, the important data features in the to-be-processed fault information are enhanced by using the weight coefficient, and the data features with less effect on the current task are suppressed.

[0098] The updated to-be-processed fault information is subjected to convolution processing to further strengthen and reduce the dimension of the to-be-processed fault information, and then input into the second processing model to obtain the fault recognition result. In this way, by setting the weight coefficient through the self-attention channel, the important information is strengthened, and the data dimension is reduced through convolution processing, which can enable the second fault processing model to more accurately identify the energy storage device fault and improve the fault recognition rate.

[0099] Further, after step S30, it further includes:

[0100] Determine the fault type corresponding to the to-be-processed fault information according to the fault recognition result;

[0101] Execute the preset fault operation corresponding to the fault type, and the preset fault operation includes at least one of the following:

[0102] Output a fault prompt message;

[0103] Output the solution associated with the fault type.

[0104] In this embodiment, the energy storage device may be a battery pack composed of multiple single cells. Among the fault recognition results identified by the second fault processing model, it also includes various fault types of the energy storage device, specifically including: normal operation, open circuit of a single cell or the entire energy storage device, open circuit of the entire energy storage device, poor contact of a single cell, different aging degrees of multiple single cells, etc., which can be represented by fault type tags 4, 0, 1, 2, 3 respectively. Among them, normal operation, poor contact of a single cell, and different aging degrees of multiple single cells cannot be directly determined by detecting the data of the energy storage device. In this embodiment, by converting the fault information to be processed into fault feature information to enhance the data expressiveness, the corresponding fault recognition result can be determined through the second fault processing model, so that the number of fault types that can be detected based on the fault information to be processed increases, and the detected range can cover faults such as single cell faults that cannot be directly detected by the data of the energy storage device waiting for the fault information to be processed. Therefore, the recognition rate of energy storage device faults can be improved.

[0105] Corresponding fault operations can be taken on the target energy storage device. Since the fault types are different, their urgency levels are also different. Different abnormal handling operations can be taken for different fault types to ensure the normal operation of the target energy storage device. Among them, the corresponding fault operations that can be taken on the target energy storage device at least include: outputting fault prompt information according to the fault type to prompt the user about the working state of the target energy storage device, and outputting a solution associated with the fault type to inform the user how to handle the fault. After detecting the fault type, fault prompt information or a solution can be output according to the fault type and fed back to the customer, so that the user can understand the state of the target energy storage device and select different treatments for the target energy storage device according to the fault type, and replace the severely faulty energy storage device in time to ensure the use safety. It is also possible to execute the abnormal handling operation corresponding to the fault type to automatically handle the fault. For example, if the fault type is normal operation, a prompt message indicating that the energy storage device is operating normally can be output or the abnormal handling operation can be ignored. If the fault type is inconsistent aging of single cells, a prompt message for replacing the energy storage device can be output at preset intervals. If the fault type is a short circuit, the power supply function of the target energy storage device can be cut off, etc.

[0106] Optionally, referring to Figure 2 , based on any of the above embodiments, in the fault processing method of the present invention, the second fault processing model is obtained by the processing module based on the following steps:

[0107] Step S50: Obtain multiple training fault information corresponding to different fault recognition results, and determine the positive and negative categories corresponding to the training fault information based on the initial second fault processing model;

[0108] In this embodiment, according to the task objective of the second fault handling model, the fault identification results of the energy storage device that need to be identified are determined. For the energy storage device, the possible fault identification results it may obtain include: normal operation, open circuit of a single battery or the entire energy storage device, open circuit of the entire energy storage device, poor contact of a single battery, different aging degrees of multiple single batteries, etc. Multiple training energy storage device data corresponding to different fault identification results are respectively obtained and processed into multiple training fault information, that is, the training fault information corresponding to the energy storage device fault with different fault identification results is obtained. The training fault information also carries the label of its corresponding fault identification result, which is used to calculate the model loss, thereby optimizing the accuracy of the classification model.

[0109] First, the training fault information is processed and enhanced based on the first fault handling model to obtain the corresponding training fault feature information. Then, the training fault feature information is input into the initial second fault handling model. After the initial second fault handling model is trained, the second fault handling model is obtained. It has the same function as the second fault handling model and can process the training fault feature information to map out the corresponding fault identification result, thereby determining the classification result. However, the initial second fault handling model is not trained, and its model accuracy is not as good as that of the second fault handling model, and the accuracy of the classification result it determines needs to be improved.

[0110] In this embodiment, there are multiple fault identification results, so the second fault handling model needs to perform a multi-classification task. The initial second fault handling model can use the fuzzy support vector machine FSVM-CIL to classify the training fault information to obtain the corresponding fault identification result. The initial second fault handling model will classify the training fault information into positive and negative categories based on the fault identification result corresponding to the training fault information and its corresponding correct fault identification result label, as the positive and negative categories corresponding to the training data.

[0111] Step S50: For the target training fault information of each positive and negative category, determine the importance of the target training fault information according to the difference between the target training fault information and the central estimation point of all training fault information belonging to the same positive and negative category;

[0112] For each positive and negative category, determine the target training fault information corresponding to the positive and negative category. All the training fault information in this positive and negative category can be used as the target training fault information. Calculate the central estimation point of all training fault information belonging to the same positive and negative category, calculate the difference between each target training fault information in this positive and negative category and the central estimation point, and determine the importance of the corresponding target training fault information in its own positive and negative category according to this difference. The specific formula is as follows:

[0113]

[0114]

[0115] Among them, β ∈ [0, 1] is a hyperparameter, is the center of the spherical region and also the center estimation point of the set of all target training fault information corresponding to the same positive and negative categories, x i represents the i-th target training fault information under the same positive and negative categories, f(x i ) represents the importance of the i-th target training fault information under its belonging positive and negative categories.

[0116] Step S60: Determine an adjustment variable based on the initial second fault processing model according to the importance, the weight values of the positive and negative categories, and the optimization model.

[0117] In this embodiment, for the target training fault information belonging to the same positive and negative categories, after determining the importance corresponding to the target training fault information, according to the importance of all target training fault information, the weight values of the positive and negative categories to which the target training fault information belongs, and the optimization model, determine the adjustment variable of the initial second fault processing model.

[0118] Further, step S60 includes:

[0119] For each target training fault data, determine the membership value of the target fault data according to the importance of the target training fault data and the weight values of the positive and negative categories to which the target training fault data belongs;

[0120] Substitute the membership values corresponding to each target training data into the optimization model to obtain the solution result of the optimization model, and determine the adjustment variable according to the solution result.

[0121] For each target training fault data, the importance of the target training fault data under its belonging category and the weight of its belonging positive and negative categories can be determined. Based on the above two parameters, the membership of the target training fault information in its belonging positive and negative categories can be calculated. The calculation formula of the membership is as follows:

[0122]

[0123]

[0124] Among them, represents the membership value of a training fault information with a positive and negative category of positive in the positive class in the positive class, represents a training fault information with a positive and negative category of negative in the membership value in the negative class. r + and r- They are the weight values of the positive and negative classes respectively. Through the weight values, the class imbalance can be reflected, and r can be set + >r - 。According to the membership degree, the fault recognition result of the initial second fault handling model for the training fault information can also be determined.

[0125] The optimized model obtains a preset optimized model. The optimized model is set based on the following optimization formula, and the membership degrees corresponding to each target training fault information are substituted into this optimization formula:

[0126]

[0127]

[0128] Among them, w is the classification interface vector, ξi is the relaxation factor of the i-th training fault information, C is the penalty factor of the i-th training fault information, and the value of the optimization formula represents the model error. In order to make the value corresponding to the optimization formula reach the minimum or be lower than the error threshold, the relaxation factor and the penalty factor need to be adjusted. Therefore, by solving the above formula, the target relaxation factor and the target penalty factor that make the value corresponding to the optimization formula reach the minimum or be lower than the error threshold can be obtained. According to the target relaxation factor and the target penalty factor, the adjustment variables of the initial second fault handling model can be determined.

[0129] Step S70: Optimize the initial second fault handling model according to the adjustment variables to obtain the second fault handling model.

[0130] In this embodiment, in order to make the model error smaller, that is, to make the value of the optimized model reach the minimum or be lower than the error threshold, the target relaxation factor and the target penalty factor can be determined according to the optimized model, and the adjustment variables of the initial second fault handling model can be determined according to the target relaxation factor and the target penalty factor, so as to optimize the initial second fault handling model, improve its model accuracy, and thus obtain the second fault handling model.

[0131] In the technical solution disclosed in this embodiment, the importance of the target training fault information is determined by the difference between the target training fault information and the central estimation point of all target training fault information, and the importance is adjusted for the weight value of the target training fault information, so that in the process of training the second fault handling model, more attention is paid to the important training fault information, which can improve the efficiency of model training, improve the accuracy of the second fault handling model after training, and thus improve the fault recognition rate of the energy storage device.

[0132] Optionally, step S50 further includes:

[0133] Obtain training energy storage device data of multiple normally operating energy storage devices from the cloud platform; and / or

[0134] Adjust the resistance values of the resistors on the current branches where each single battery in the simulated energy storage device is located to simulate different fault identification results of the energy storage device, and collect the training energy storage device data corresponding to the fault identification results;

[0135] Among them, the fault identification results include normal operation and fault identification results of multiple single batteries.

[0136] In this embodiment, for the energy storage device, the fault identification results that need to be identified include: normal operation, open circuit of a single battery or the entire energy storage device, open circuit of the entire energy storage device, poor contact of a single battery, and different aging degrees of multiple single batteries. Among them, when the energy storage device data of the energy storage device in normal operation is used as the training energy storage device data, it can be obtained from the cloud platform. For example, when the energy storage device is applied to an electric vehicle, a large number of electric vehicles upload the energy storage device data of the energy storage device in normal operation to the cloud platform as the training energy storage device data with the identification result of normal operation for training use, which reduces the difficulty of obtaining training data.

[0137] In an application scenario, the target energy storage device powers an electric vehicle. The training energy storage device data in normal operation required for training the fault processing model for fault detection of the target energy storage device can be historical data from the big data cloud platform of the remote service and management system of the electric vehicle. And this platform mainly serves new energy electric vehicles, can monitor and analyze the operation data of the vehicles, obtain the energy storage device data of the target energy storage device to be detected of the target electric vehicle, process it into the fault information to be processed, input it into the first fault processing model and the second fault processing model in sequence, and output the fault processing result, so as to provide real-time diagnosis, prediction, statistical analysis and safety warning.

[0138] In this embodiment, multiple training energy storage device data corresponding to different fault identification results can also be obtained through simulation experiments. Set up a simulated energy storage device. The simulated energy storage device includes multiple parallel single batteries, and an adjustable resistor is set on the branch where each single battery is located. By adjusting the resistance value of the resistor, the current magnitude on the branch where the corresponding single battery is located can be changed, so as to realize the fault identification results corresponding to different fault types in the simulated energy storage device, and thus collect the training energy storage device data corresponding to different fault categories.

[0139] Exemplarily, referring to Figure 3 the shown simulated energy storage device, each single battery is connected with a resistor, and its resistance value is changed to simulate different degrees of connection faults. In addition, the simulated energy storage device is also provided with a 16-bit high-precision AD converter for measuring the terminal voltage and total current of the energy storage device. CAN communication is used to transmit the collected data to the fault processing device for subsequent operations and analysis.

[0140] During the discharge process of the energy storage device, different types of energy storage device failures are simulated and added to the circuit. The specific operations include: (1) Disconnecting the first branch cell01 to simulate poor contact of a single battery; (2) Randomly adding contact resistors with different resistance values, such as 5, 10, 15, 20, 25, and 30 mΩ, to each single battery to simulate inconsistent aging of single batteries. Generally speaking, energy storage device failures rarely occur during the use of the energy storage device. In this way, by simulating the energy storage device to obtain training energy storage device data with failures, the cost and time for obtaining training failure information with failures are greatly reduced, and the training efficiency is improved.

[0141] Optionally, referring to Figure 4 , based on any of the above embodiments, in another embodiment of the fault handling method of the present invention, before step S30, it further includes:

[0142] S80. Sequentially input the energy storage device data to be detected into the encoding layer and the decoding layer corresponding to the encoding layer to obtain reconstructed energy storage device data;

[0143] In this embodiment, before processing the energy storage device data to be detected of the target energy storage device into fault information, it is also necessary to pre-evaluate whether a fault has occurred: input the energy storage device data to be detected into a preset encoding layer, and according to the encoding features obtained by the encoding layer, then input it into the decoding layer corresponding to the encoding layer to realize reconstructing the energy storage device data to be detected close to the input preset encoder, and obtain reconstructed energy storage device data.

[0144] Optionally, before inputting the energy storage device data (including training energy storage device data and energy storage device data to be detected) into the decoding layer, perform normalization processing on it. Taking the training energy storage device data as an example, before inputting multiple training energy storage device data into the preset encoding layer, it is also necessary to perform normalization processing on each training energy storage device data. The data normalization method is linear function normalization, and after normalization, it is more universal.

[0145]

[0146] Among them, X is the data before normalization, X norm is the data after normalization, X max and X min correspond to the maximum value and the minimum value in multiple training energy storage device data respectively, and the range of the normalized data can be [0, 1].

[0147] S90. Based on the reconstruction error between the energy storage device data and the reconstructed energy storage device data, determine the abnormal risk coefficient of the energy storage device data;

[0148] In this embodiment, the decoding layer corresponds to the encoding layer. The two have the same structure, but the input and output are opposite. Specifically, the number of layers of the decoding layer and the encoding layer is the same, and the encoding units in each layer are the same. Referring to Figure 5 , the encoding layer has two layers. The first layer has 64 Bi-LSTM units, and the second layer has 32 Bi-LSTM units. Correspondingly, the first layer of the decoder layer has 32 Bi-LSTM units, and the second layer of the decoder has 64 Bi-LSTM units. The final output of the decoder is the reconstruction of the input. In theory, based on the decoding layer, the encoded features can be reconstructed into the corresponding energy storage device data. However, if there is a risk of failure in the energy storage device data, resulting in a weakened association between data structures, it will lead to a reconstruction error between the reconstructed energy storage device data and the energy storage device data to be detected. The greater the reconstruction error, the greater the abnormal risk coefficient. The reconstruction error can be used as the abnormal risk coefficient of the energy storage device data, which can be used as a judgment benchmark for whether there is a risk of failure in the target energy storage device corresponding to the energy storage device data.

[0149] S100. When the abnormal risk coefficient is greater than the error threshold, use the output result of the encoding layer as the to-be-processed fault information.

[0150] In this embodiment, if the abnormal risk coefficient is greater than the error threshold, it is considered that there is a relatively high risk of failure in the target energy storage device corresponding to the energy storage device data to be detected. It is necessary to use the output result of the encoding layer, that is, the encoded features, as the to-be-processed fault information and further input it into the first fault processing model and the second fault processing model to obtain a fault recognition result. If the abnormal risk coefficient is less than or equal to the error threshold, it is considered that there is no relatively high risk of failure in the target energy storage device corresponding to the energy storage device data to be detected, and there is no need for a further fault recognition result, and no other operations are performed.

[0151] Exemplarily, a specific application scenario is provided below. Please refer to Figure 6 :

[0152] After obtaining the energy storage device data to be detected of the target energy storage device, the data is normalized and output to the Bi-LSTM encoding layer to obtain encoding features. The encoding features are input into the Bi-LSTM decoding layer to obtain the reconstructed energy storage device data. The two are compared to determine the reconstruction error. If the abnormal risk coefficient corresponding to the reconstruction error is less than the error threshold, the target energy storage device corresponding to the energy storage device data to be detected is normal and no operation is performed. If the abnormal risk coefficient corresponding to the reconstruction error is greater than the error threshold, it is determined that there is a fault risk in the target energy storage device corresponding to the energy storage device data to be detected. Then, the encoding features output by the Bi-LSTM encoding layer are used as the fault information to be processed and input into the first fault processing model, that is, the SENet network with an attention mechanism, for convolution processing. The fault feature information after convolution is extracted and input into the second fault processing model. The second fault processing model can be a classification model of FSVM-CIL to determine the fault recognition result, and corresponding fault operations or normal operation are selected according to the fault recognition result, and no operation is performed.

[0153] In the technical solution disclosed in the embodiment, the energy storage device data to be detected of the target energy storage device is reconstructed through the encoding layer and the corresponding decoding layer to obtain the actually reconstructed reconstructed energy storage device data. Based on the comparison between the reconstructed energy storage device data and the energy storage device data to be detected, the reconstruction error is obtained, so as to pre-determine the abnormal risk coefficient corresponding to the energy storage device data to be detected, and then determine whether there is a fault risk in the target energy storage device corresponding to the energy storage device to be detected, and then determine whether to perform fault recognition. On the one hand, unnecessary computational consumption is reduced, and on the other hand, the fault recognition rate can be further improved.

[0154] Further, before step S100, it further includes:

[0155] Obtain multiple sets of verified energy storage device data;

[0156] Input the verified energy storage device data into the preset encoding layer and the decoding layer in sequence to obtain the reconstructed verified energy storage device data corresponding to the verified energy storage device data;

[0157] Based on all the verified energy storage device data and the reconstructed verified energy storage device data, determine the error threshold.

[0158] In this embodiment, it is necessary to obtain the energy storage device data collected during the normal operation of multiple energy storage devices as the verified energy storage device data. Based on the verified energy storage device data during the normal operation of the energy storage device, the reconstruction error that should exist when there is no abnormal risk in the energy storage device can be determined. Then, the verified energy storage device data is first input into the above-mentioned encoding layer to obtain the encoding features corresponding to the verified energy storage device data. Then, the encoding features corresponding to the verified energy storage device data are input into the decoding layer corresponding to the encoding layer to obtain the reconstructed verified energy storage device data. The reconstruction error between each verified energy storage device data and the corresponding reconstructed verified energy storage device data is determined, and the error threshold of the abnormal risk coefficient can be determined based on these reconstruction errors.

[0159] Optionally, based on all the verified energy storage device data and the reconstructed verified energy storage device data, determining the error threshold includes:

[0160] For each of the verified energy storage device data, determine the verification reconstruction error between each verified energy storage device data and the corresponding reconstructed verified energy storage device data;

[0161] Based on the verification reconstruction error, count the number of verification reconstruction errors in each error interval;

[0162] Determine the target error interval according to the number of verification reconstruction errors in each error interval, and determine the error threshold according to the target error interval.

[0163] In this embodiment, the reconstruction error between each verified energy storage device data and the corresponding reconstructed verified energy storage device data can find out the most likely reconstruction error when multiple energy storage devices are operating normally. Specifically, for each verified energy storage device data, determine the verification reconstruction error between each verified energy storage device data and the corresponding reconstructed verified energy storage device data. The verification reconstruction error is the different reconstruction errors that will occur when the energy storage device is operating normally. Based on the verification reconstruction error, count the number of verification reconstruction errors in each error interval, and fit a reconstruction error distribution image based on the number of verification reconstruction errors in each error interval. It will be found that the image is a quasi-normal distribution image N(μ,σ 2 ). Therefore, the average value μ and variance σ of each verification reconstruction error can also be calculated. Then, based on the average value μ and variance σ, select the target error interval as the error interval where the reconstruction error is most likely to occur during normal operation. For example, take (0, μ + 2σ) as the target error interval, and r = μ + 2σ as the error threshold. In this way, when the abnormal risk coefficient is greater than μ + 2σ, it is considered that the target energy storage device that has collected the energy storage device data corresponding to the abnormal risk coefficient has a fault risk.

[0164] By verifying the energy storage device data during the normal operation of the energy storage device in this way, calculating the reconstructed verification energy storage device data obtained during the normal operation of the energy storage device, and calculating the internal situation of the corresponding verification reconstruction error variance to set the error threshold of the abnormal risk coefficient, the accuracy of the abnormal risk assessment of the energy storage device can be improved, thereby increasing the fault recognition rate.

[0165] Optionally, based on any of the above embodiments, in another embodiment of the fault handling method of the present invention, before step S80, it further includes:

[0166] Step S81, obtaining a plurality of historical energy storage device data and current energy storage device data of the target energy storage device within a preset time period;

[0167] In this embodiment, the energy storage device data of the target energy storage device can be collected at regular intervals, and a preset number of the latest collected energy storage device data can be saved. In this way, when it is necessary to identify the fault of the target energy storage device, a plurality of historical energy storage device data and current energy storage device data of the target energy storage device within a preset time period can be obtained.

[0168] Step S82, predicting the change trend of the energy storage device data of the energy storage device within a future time period according to the change curve fitted by the plurality of historical energy storage device data and the current energy storage device data;

[0169] In this embodiment, according to a plurality of historical energy storage device data and current energy storage device data, the change curve of the energy storage device data of the target energy storage device within a preset time period can be fitted, and the energy storage device data change curve can be extended according to the nth derivative result, so as to obtain the change trend of the energy storage device data of the target energy storage device within a future time period.

[0170] Step S83, determining predicted energy storage device data according to the current energy storage device data and the change trend, and using at least one of the predicted energy storage device data and the historical energy storage device data and the current energy storage device data as the energy storage device data to be detected of the target energy storage device.

[0171] In this embodiment, according to the current energy storage device data and the change trend, the predicted energy storage device data corresponding to one or more times within a future time period can be determined. Furthermore, the predicted energy storage device data, the historical energy storage device data, and the current energy storage device data can be used as the energy storage device data to be detected of the target energy storage device to identify the fault of the target energy storage device corresponding to the historical time point, the current time point, and the future time point. Thus, the probability of identifying the fault of the energy storage device is increased, the fault recognition rate of the energy storage device can be improved, and when there is no obvious damage to the energy storage device in the early stage of the fault, the possible faults detected can be identified, reminding the user to repair in time to avoid accidents.

[0172] This embodiment also provides a fault processing device, which can be specifically integrated in a fault processing device. For example, as Figure 7 shown, the fault processing device may include:

[0173] An acquisition module 1001, configured to acquire module for acquiring fault information to be processed;

[0174] An extraction module 1002, configured to perform feature extraction on the fault information to be processed based on a first fault processing model to obtain fault feature information;

[0175] A processing module 1003, configured to perform identification processing on the fault feature information based on a second fault processing model to obtain a fault identification result.

[0176] Preferably, the extraction module 1002 performs feature extraction on the fault information to be processed based on a first fault processing model to obtain fault feature information, including:

[0177] Updating the fault information to be processed based on the weight coefficients corresponding to each target data feature in the fault information to be processed;

[0178] Based on the first fault processing model, processing the updated fault information to be processed to obtain the fault feature information;

[0179] Preferably, the weight coefficient is obtained by the extraction module 1002 based on the following steps, including:

[0180] Obtaining a plurality of target data features in the fault information to be processed;

[0181] Calculating the similarity between the target data feature and each data feature in the fault information to be processed;

[0182] Determining the weight coefficient corresponding to each target data feature according to the similarity corresponding to each target data feature;

[0183] Preferably, before the acquisition module 1001 acquires the fault information to be processed, it further includes:

[0184] Sequentially inputting the energy storage device data to be detected into an encoding layer and a decoding layer corresponding to the encoding layer to obtain reconstructed energy storage device data;

[0185] Based on the reconstruction error between the energy storage device data and the reconstructed energy storage device data, determining the abnormal risk coefficient of the energy storage device data;

[0186] When the abnormal risk coefficient is greater than the error threshold, using the output result of the encoding layer as the fault information to be processed;

[0187] Preferably, before the obtaining module 1001 uses the output result of the encoding layer as the to-be-processed fault information when the abnormal risk coefficient is greater than the error threshold, the method further includes:

[0188] Obtain data of multiple verification energy storage devices;

[0189] Input the data of the verification energy storage devices into the encoding layer and the decoding layer in sequence to obtain reconstructed verification energy storage device data corresponding to the data of the verification energy storage devices;

[0190] Determine the error threshold based on all the data of the verification energy storage devices and the reconstructed verification energy storage device data;

[0191] Preferably, when the obtaining module 1001 determines the error threshold based on all the data of the verification energy storage devices and the reconstructed verification energy storage device data, it includes:

[0192] For each data of the verification energy storage devices, determine the verification reconstruction error between each data of the verification energy storage devices and the corresponding reconstructed verification energy storage device data;

[0193] Based on the verification reconstruction error, count the number of verification reconstruction errors in each error interval;

[0194] Determine a target error interval according to the number of verification reconstruction errors in each error interval, and determine the error threshold according to the target error interval;

[0195] Preferably, the second fault processing model is obtained by the processing module 1003 based on the following steps:

[0196] Obtain multiple training fault information corresponding to different fault recognition results, and determine the positive and negative categories corresponding to the training fault information based on the initial second fault processing model;

[0197] For the target training fault information of each positive and negative category, determine the importance of the target training fault information according to the difference between the target training fault information and the central estimation point of all the training fault information belonging to the same positive and negative category;

[0198] Determine an adjustment variable based on the initial second fault processing model according to the importance, the weight value of the positive and negative categories, and the optimization model;

[0199] Optimize the initial second fault processing model according to the adjustment variable to obtain the second fault processing model.

[0200] In this embodiment, the to-be-processed fault information is obtained; feature extraction is performed on the to-be-processed fault information based on a first fault processing model to obtain fault feature information; and identification processing is performed on the fault feature information based on a second fault processing model to obtain a fault identification result. By processing the to-be-processed fault information through the first fault processing model, more prominent fault feature information is obtained. Furthermore, by using the second fault processing model to perform identification processing on the more prominent fault feature information, it is possible to distinguish the fault types of the energy storage device that cannot be directly determined through the energy storage device data, and a more accurate fault identification result is obtained, thereby improving the accuracy of identifying the faults of the energy storage device.

[0201] As Figure 8 shown, Figure 8 FIG. is a schematic structural diagram of a fault processing device provided by an embodiment of the present invention. The fault processing device 1100 includes a processor 1101 having one or more processing cores, a memory 1102 having one or more computer-readable storage media, and a computer program stored on the memory 1102 and executable on the processor. Among them, the processor 1101 is electrically connected to the memory 1102. Those skilled in the art can understand that the structural diagram of the fault processing device shown in the figure does not constitute a limitation on the fault processing device, and may include more or fewer components than those shown, or combine certain components, or have different component arrangements.

[0202] The processor 1101 is the control center of the fault processing device 1100, connects various parts of the entire fault processing device 1100 through various interfaces and lines, and executes various functions and processes data of the fault processing device 1100 by running or loading software programs and / or units stored in the memory 1102, and calling data stored in the memory 1102, so as to perform overall monitoring on the fault processing device 1100. The processor 1101 may be a central processing unit (CPU), a graphics processing unit (GPU), a network processor (NP), etc., and can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention.

[0203] In the embodiment of the present invention, the processor 1101 in the fault processing device 1100 will load the instructions corresponding to the processes of one or more application programs into the memory 1102 according to the following steps, and the processor 1101 will run the application programs stored in the memory 1102 to implement various functions, such as:

[0204] Obtain the to-be-processed fault information;

[0205] Perform feature extraction on the to-be-processed fault information based on a first fault processing model to obtain fault feature information;

[0206] Identify and process the fault feature information based on the second fault processing model to obtain a fault identification result.

[0207] For the specific implementation of each of the above operations, reference may be made to the previous embodiments, which will not be elaborated herein.

[0208] Optionally, as Figure 8 shown, the fault processing device 1100 further includes: a touch display screen 1103, a radio frequency circuit 1104, an audio circuit 1105, an input unit 1106, and a power supply 1107. Among them, the processor 1101 is electrically connected to the touch display screen 1103, the radio frequency circuit 1104, the audio circuit 1105, the input unit 1106, and the power supply 1107 respectively. Those skilled in the art can understand that Figure 8 the structure of the fault processing device shown in

[0209] The touch display screen 1103 can be used to display a graphical user interface and receive operation instructions generated by a user's interaction with the graphical user interface. The touch display screen 1103 may include a display panel and a touch panel. Among them, the display panel can be used to display information input by the user or information provided to the user, as well as various graphical user interfaces of the fault handling device. These graphical user interfaces can be composed of graphics, text, icons, videos, and any combination thereof. Optionally, the display panel can be configured in the form of a liquid crystal display (LCD), an organic light-emitting diode (OLED), etc. The touch panel can be used to collect touch operations of the user on or near it (such as operations of the user using any suitable object or accessory such as a finger or a stylus on or near the touch panel), and generate corresponding operation instructions, and the operation instructions execute the corresponding program. Optionally, the touch panel can include two parts: a touch detection device and a touch controller. Among them, the touch detection device detects the touch position of the user and detects the signal brought by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into contact coordinates, and then sends it to the processor 1101, and can receive and execute the commands sent by the processor 1101. The touch panel can cover the display panel. When the touch panel detects a touch operation on or near it, it is transmitted to the processor 1101 to determine the type of touch event. Subsequently, the processor 1101 provides a corresponding visual output on the display panel according to the type of touch event. In the embodiment of the present invention, the touch panel and the display panel can be integrated into the touch display screen 1103 to implement input and output functions. However, in some embodiments, the touch panel and the touch panel can be implemented as two independent components to implement input and output functions. That is, the touch display screen 1103 can also be used as a part of the input unit 1106 to implement the input function.

[0210] The radio frequency circuit 1104 can be used to transmit and receive radio frequency signals to establish wireless communication with a network device or other fault handling devices through wireless communication, and transmit and receive signals with the network device or other fault handling devices.

[0211] The audio circuit 1105 can be used to provide an audio interface between the user and the fault handling device through a speaker and a microphone. The audio circuit 1105 can transmit the electrical signal converted from the received audio data to the speaker, which converts it into a sound signal for output; on the other hand, the microphone converts the collected sound signal into an electrical signal, which is received by the audio circuit 1105 and then converted into audio data. After the audio data is output and processed by the processor 1101, it is sent through the radio frequency circuit 1104 to, for example, another fault handling device, or the audio data is output to the memory 1102 for further processing. The audio circuit 1105 may also include an earphone jack to provide communication between the peripheral earphone and the fault handling device.

[0212] The input unit 1106 can be used to receive input digital, character information or user feature information (such as fingerprint, iris, face information, etc.), and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function controls.

[0213] The power supply 1107 is used to supply power to each component of the fault handling device 1100. Optionally, the power supply 1107 can be logically connected to the processor 1101 through a power management system, so as to implement functions such as management of charging, discharging, and power consumption management through the power management system. The power supply 1107 may also include any components such as one or more DC or AC power supplies, a recharge system, a power failure detection circuit, a power converter or inverter, and a power status indicator.

[0214] Although Figure 8 not shown in the figure, the fault handling device 1100 may also include a camera, a sensor, a Wi-Fi module, a Bluetooth module, etc., which will not be elaborated here.

[0215] In the above embodiments, the descriptions of the various embodiments have their own focuses. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0216] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructions or by controlling relevant hardware through instructions. The instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0217] For this reason, an embodiment of the present invention provides a computer-readable storage medium, in which multiple computer programs are stored. The computer programs can be loaded by a processor to execute any one of the fault handling methods provided by the embodiments of the present invention. The computer programs can execute the steps of the following fault handling method:

[0218] Obtain the fault information to be processed;

[0219] Extract features from the to-be-processed fault information based on the first fault processing model to obtain fault feature information;

[0220] Perform identification processing on the fault feature information based on the second fault processing model to obtain a fault identification result.

[0221] For the specific implementation of each of the above operations, reference can be made to the previous embodiments, which will not be elaborated here.

[0222] Among them, the computer-readable storage medium may include: read-only memory (ROM, Read Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disc, etc.

[0223] Since the computer program stored in the computer-readable storage medium can execute any fault processing method provided by the embodiments of the present invention, the beneficial effects achievable by any fault processing method provided by the embodiments of the present invention can be realized. For details, refer to the previous embodiments, which will not be elaborated here.

[0224] In the above embodiments of the fault processing device, computer-readable storage medium, fault processing equipment, and computer program product, the descriptions of each embodiment have their own focuses. For the parts not elaborated in a certain embodiment, reference can be made to the relevant descriptions of other embodiments. Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes and the beneficial effects that can be brought by the above-described fault processing device, computer-readable storage medium, computer program product, fault processing equipment, and their corresponding units can refer to the description of the fault processing method in the above embodiments, which will not be elaborated here specifically.

[0225] The above has introduced in detail a fault processing method, device, fault processing equipment, computer-readable storage medium, and computer program product provided by the embodiments of the present invention. Specific examples are used in this article to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation on the present invention.

Claims

1. A fault handling method, characterized in that, The method includes: Obtaining the fault information to be processed; Performing feature extraction on the fault information to be processed based on a first fault processing model to obtain fault feature information; Performing identification processing on the fault feature information based on a second fault processing model to obtain a fault identification result.

2. The fault handling method according to claim 1, characterized in that, The performing feature extraction on the fault information to be processed based on a first fault processing model to obtain fault feature information includes: Updating the fault information to be processed based on the weight coefficients corresponding to the target data features in the fault information to be processed; Performing processing on the updated fault information to be processed based on the first fault processing model to obtain the fault feature information.

3. The fault handling method according to claim 2, wherein, The weight coefficients are obtained based on the following steps: Obtaining multiple target data features in the fault information to be processed; Calculating the similarity between the target data features and each data feature in the fault information to be processed; Determining the weight coefficients corresponding to the target data features according to the similarities corresponding to the target data features.

4. The fault handling method according to claim 1, wherein Before obtaining the fault information to be processed, it further includes: Sequentially inputting the data of the energy storage device to be detected into an encoding layer and a decoding layer corresponding to the encoding layer to obtain reconstructed energy storage device data; Determining the abnormal risk coefficient of the energy storage device data based on the reconstruction error between the energy storage device data and the reconstructed energy storage device data; When the abnormal risk coefficient is greater than the error threshold, using the output result of the encoding layer as the fault information to be processed.

5. The fault handling method according to claim 4, wherein Before the step of using the output result of the encoding layer as the fault information to be processed when the abnormal risk coefficient is greater than the error threshold, it further includes: Obtaining multiple verified energy storage device data; Sequentially inputting the verified energy storage device data into the encoding layer and the decoding layer to obtain reconstructed verified energy storage device data corresponding to the verified energy storage device data; Determining the error threshold based on all the verified energy storage device data and the reconstructed verified energy storage device data.

6. The fault handling method according to claim 5, wherein The determining the error threshold based on all the verified energy storage device data and the reconstructed verified energy storage device data includes: For each of the verified energy storage device data, determining the verification reconstruction error between the verified energy storage device data and the corresponding reconstructed verified energy storage device data; Based on the verification reconstruction error, counting the number of verification reconstruction errors in each error interval; Determining a target error interval according to the number of verification reconstruction errors in each error interval, and determining the error threshold according to the target error interval.

7. The fault handling method according to claim 1, characterized in that The second fault processing model is obtained based on the following steps: Obtaining multiple training fault information corresponding to different fault identification results, and determining the positive and negative categories corresponding to the training fault information based on an initial second fault processing model; For the target training fault information of each positive and negative category, determining the importance of the target training fault information according to the difference between the target training fault information and the central estimation point of all the training fault information belonging to the same positive and negative category; Determining an adjustment variable based on the initial second fault processing model according to the importance, the weight values of the positive and negative categories, and an optimization model; Optimize the initial second fault handling model according to the adjustment variable to obtain the second fault handling model.

8. A fault handling device, characterized in that, The fault handling device includes: An acquisition module, configured to acquire fault information to be processed; An extraction module, configured to perform feature extraction on the fault information to be processed based on a first fault handling model to obtain fault feature information; A processing module, configured to perform identification processing on the fault feature information based on a second fault handling model to obtain a fault identification result; Preferably, the extraction module performs feature extraction on the fault information to be processed based on a first fault handling model to obtain fault feature information, including: Updating the fault information to be processed based on the weight coefficients corresponding to each target data feature in the fault information to be processed; Processing the updated fault information to be processed based on the first fault handling model to obtain the fault feature information; Preferably, the weight coefficient is obtained by the extraction module based on the following steps: Obtain multiple target data features in the fault information to be processed; Calculate the similarity between the target data feature and each data feature in the fault information to be processed; Determine the weight coefficient corresponding to each target data feature according to the similarity corresponding to each target data feature; Preferably, before the acquisition module acquires the fault information to be processed, it further includes: Sequentially inputting the energy storage device data to be detected into an encoding layer and a decoding layer corresponding to the encoding layer to obtain reconstructed energy storage device data; Determine the abnormal risk coefficient of the energy storage device data based on the reconstruction error between the energy storage device data and the reconstructed energy storage device data; When the abnormal risk coefficient is greater than the error threshold, use the output result of the encoding layer as the fault information to be processed; Preferably, before the acquisition module uses the output result of the encoding layer as the fault information to be processed when the abnormal risk coefficient is greater than the error threshold, it further includes: Obtain multiple verified energy storage device data; Sequentially input the verified energy storage device data into the encoding layer and the decoding layer to obtain reconstructed verified energy storage device data corresponding to the verified energy storage device data; Determine the error threshold based on all the verified energy storage device data and the reconstructed verified energy storage device data; Preferably, the acquisition module determines the error threshold based on all the verified energy storage device data and the reconstructed verified energy storage device data, including: For each verified energy storage device data, determine the verification reconstruction error between each verified energy storage device data and the corresponding reconstructed verified energy storage device data; Based on the verification reconstruction error, count the number of verification reconstruction errors in each error interval; Determine a target error interval according to the number of verification reconstruction errors in each error interval, and determine the error threshold according to the target error interval; Preferably, the second fault handling model is obtained by the processing module based on the following steps: Obtain multiple training fault information corresponding to different fault identification results, and determine the positive and negative categories corresponding to the training fault information based on an initial second fault handling model; For the target training fault information of each positive and negative category, determine the importance of the target training fault information according to the difference between the target training fault information and the central estimation point of all training fault information belonging to the same positive and negative category; According to the importance, the weight value of the positive and negative categories and the optimization model, determine the adjustment variable based on the initial second fault handling model; Optimize the initial second fault handling model according to the adjustment variable to obtain the second fault handling model.

9. A fault handling device, characterized in that, It includes a processor and a memory, and the memory stores multiple instructions; the processor loads the instructions from the memory to execute the steps of the fault handling method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores multiple instructions, and the instructions are suitable for being loaded by a processor to execute the steps of the fault handling method according to any one of claims 1 to 7.

Images