Battery abnormality diagnosis method and device, computer device, readable storage medium and program product

Abstract

The application relates to a storage battery abnormality diagnosis method and device, computer equipment, a readable storage medium and a program product. The method comprises the following steps: inputting current multi-dimensional battery pack information of a storage battery pack and current multi-dimensional battery information of each storage battery in the storage battery pack into a first storage battery diagnosis model and a second storage battery diagnosis model respectively, obtaining a first abnormality probability and a second abnormality probability of each storage battery; performing fusion processing on the first abnormality probability and the second abnormality probability of each storage battery respectively, obtaining a target abnormality probability of each storage battery; screening a first abnormal storage battery with a target abnormality probability greater than a preset abnormality probability from the storage batteries, screening a second abnormal storage battery with a predicted degradation time less than a preset degradation time from the storage batteries; and obtaining a target abnormal storage battery in the storage battery pack according to the first abnormal storage battery and the second abnormal storage battery. The method can improve the accuracy of storage battery abnormality diagnosis.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of battery diagnosis, and in particular to a battery abnormality diagnosis method and device, a computer device, a computer readable storage medium, and a computer program product. BACKGROUND

[0002] A battery pack includes a plurality of batteries, and a single battery abnormality can cause the entire battery pack to be unable to be normally used. Therefore, it is very important to perform abnormality diagnosis on the batteries.

[0003] In the prior art, when performing abnormality diagnosis on the batteries in the battery pack, a single dimension information of the batteries, such as voltage, is measured by manual measurement at regular time intervals, so as to determine whether the batteries are abnormal. However, the single dimension information of the batteries is used to determine whether the batteries are abnormal, which can easily cause the accuracy of the battery abnormality diagnosis to be low. SUMMARY

[0004] Therefore, it is necessary to provide a battery abnormality diagnosis method, device, computer device, computer readable storage medium, and computer program product capable of improving the accuracy of battery abnormality diagnosis.

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

[0006] obtaining current multi-dimensional battery pack information of a battery pack to be diagnosed, and current multi-dimensional battery information of each battery in the battery pack;

[0007] inputting the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained first battery diagnosis model to obtain a first abnormality probability of each battery, and inputting the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained second battery diagnosis model to obtain a second abnormality probability of each battery;

[0008] respectively performing fusion processing on the first abnormality probability and the second abnormality probability of each battery to obtain a target abnormality probability of each battery;

[0009] selecting, from the each battery, a battery with a target abnormality probability greater than a preset abnormality probability as a first abnormal battery, and obtaining a predicted degradation time of each battery, and selecting, from the each battery, a battery with a corresponding predicted degradation time less than a preset degradation time as a second abnormal battery;

[0010] According to the first abnormal battery and the second abnormal battery, a target abnormal battery in the battery pack is obtained.

[0011] In one of the embodiments, the inputting of the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into the pre-trained first battery diagnosis model to obtain the first abnormal probability of each battery comprises:

[0012] The feature extraction processing is respectively performed on the current multi-dimensional battery pack information and the current multi-dimensional battery information of each battery to obtain a first feature vector corresponding to the current multi-dimensional battery pack information and a second feature vector corresponding to the current multi-dimensional battery information of each battery.

[0013] The first feature vector and the second feature vector corresponding to the current multi-dimensional battery information of each battery are respectively fused to obtain a fusion feature vector corresponding to the current multi-dimensional battery information of each battery.

[0014] The fusion feature vector corresponding to the current multi-dimensional battery information of each battery is again subjected to feature extraction processing to obtain a target feature vector corresponding to the current multi-dimensional battery information of each battery.

[0015] The target feature vector corresponding to the current multi-dimensional battery information of each battery is input into the pre-trained first battery diagnosis model to obtain the first abnormal probability of each battery.

[0016] In one of the embodiments, the pre-trained first battery diagnosis model is trained in the following manner:

[0017] The sample multi-dimensional battery pack information of a sample battery pack and the sample multi-dimensional battery information of each sample battery in the sample battery pack are obtained.

[0018] The feature extraction processing is respectively performed on the sample multi-dimensional battery pack information and the sample multi-dimensional battery information of each sample battery to obtain a first sample feature vector corresponding to the sample multi-dimensional battery pack information and a second sample feature vector corresponding to the sample multi-dimensional battery information of each sample battery.

[0019] The first sample feature vector and the second sample feature vector corresponding to the sample multi-dimensional battery information of each sample battery are respectively fused to obtain a sample fusion feature vector corresponding to the sample multi-dimensional battery information of each sample battery.

[0020] Respectively, the sample multi-dimensional battery information corresponding to the sample fusion feature vector of each sample battery is subjected to re-feature extraction processing to obtain a target feature vector corresponding to the sample multi-dimensional battery information of each sample battery;

[0021] Respectively, the target feature vector corresponding to the sample multi-dimensional battery information of each sample battery is input into the first battery diagnosis model to be trained to obtain a predicted abnormal probability of each sample battery;

[0022] The actual abnormal probability of each sample battery is obtained, and the first battery diagnosis model to be trained is iteratively trained according to the difference between the predicted abnormal probability and the actual abnormal probability of each sample battery, to obtain a trained first battery diagnosis model as the pre-trained first battery diagnosis model.

[0023] In one embodiment, the first abnormal probability and the second abnormal probability of each battery are fused to obtain a target abnormal probability of each battery, including:

[0024] The prediction accuracy of the first battery diagnosis model and the prediction accuracy of the second battery diagnosis model are obtained;

[0025] According to the prediction accuracy of the first battery diagnosis model and the prediction accuracy of the second battery diagnosis model, the corresponding relationship between the prediction accuracy and the model weight is queried to obtain a first model weight of the first battery diagnosis model and a second model weight of the second battery diagnosis model;

[0026] The first model weight is confirmed as a first weight corresponding to the first abnormal probability, and the second model weight is confirmed as a second weight corresponding to the second abnormal probability;

[0027] According to the first weight and the second weight, the first abnormal probability and the second abnormal probability of each battery are fused to obtain a target abnormal probability of each battery.

[0028] In one embodiment, the predicted degradation time of each battery is obtained, including:

[0029] The historical multi-dimensional battery pack information of the battery pack and the historical multi-dimensional battery information of each battery in the battery pack are obtained;

[0030] combine the current multi-dimensional battery information and the historical multi-dimensional battery information of each battery in the battery pack to obtain combined multi-dimensional battery information of each battery in the battery pack;

[0031] obtain a feature vector of the combined multi-dimensional battery information and a feature vector of the combined multi-dimensional battery information of each battery in the battery pack;

[0032] respectively fuse the feature vector of the combined multi-dimensional battery information and the feature vector of the combined multi-dimensional battery information of each battery in the battery pack to obtain a fused feature vector of the combined multi-dimensional battery information of each battery in the battery pack;

[0033] respectively input the fused feature vector of the combined multi-dimensional battery information of each battery in the battery pack into a pre-trained battery degradation prediction model to obtain a predicted degradation time of each battery.

[0034] In one of the embodiments, the pre-trained battery degradation prediction model is trained in the following manner:

[0035] obtain sample combined multi-dimensional battery information of a sample battery pack and sample combined multi-dimensional battery information of each sample battery in the sample battery pack;

[0036] obtain a feature vector of the sample combined multi-dimensional battery information and a feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack;

[0037] respectively fuse the feature vector of the sample combined multi-dimensional battery information and the feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack to obtain a fused feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack;

[0038] respectively input the fused feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack into a battery degradation prediction model to be trained to obtain a predicted degradation time of each sample battery;

[0039] obtain actual degradation times of the sample batteries, and iteratively train the battery degradation prediction model to be trained according to differences between the predicted degradation times and the actual degradation times of the sample batteries to obtain a trained battery degradation prediction model as the pre-trained battery degradation prediction model.

[0040] In a second aspect, the present application further provides a battery abnormality diagnosis device, comprising:

[0041] an information acquisition module, configured to acquire current multi-dimensional battery pack information of a battery pack to be diagnosed, and current multi-dimensional battery information of each battery in the battery pack;

[0042] a probability determination module, configured to input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained first battery diagnosis model to obtain a first abnormality probability of the each battery, and input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained second battery diagnosis model to obtain a second abnormality probability of the each battery;

[0043] a probability fusion module, configured to respectively fuse the first abnormality probability and the second abnormality probability of the each battery to obtain a target abnormality probability of the each battery;

[0044] a battery screening module, configured to screen, from the each battery, a battery with a target abnormality probability greater than a preset abnormality probability as a first abnormal battery, and acquire a predicted degradation time of the each battery, and screen, from the each battery, a battery with a corresponding predicted degradation time less than a preset degradation time as a second abnormal battery;

[0045] a battery diagnosis module, configured to obtain a target abnormal battery in the battery pack according to the first abnormal battery and the second abnormal battery.

[0046] In a third aspect, the present application further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0047] acquiring current multi-dimensional battery pack information of a battery pack to be diagnosed, and current multi-dimensional battery information of each battery in the battery pack;

[0048] inputting the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained first battery diagnosis model to obtain a first abnormality probability of the each battery, and inputting the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained second battery diagnosis model to obtain a second abnormality probability of the each battery;

[0049] fuse the first abnormal probability and the second abnormal probability of each battery, respectively, to obtain a target abnormal probability of each battery;

[0050] select, from the batteries, a battery with a target abnormal probability greater than a preset abnormal probability as a first abnormal battery, and obtain a predicted degradation time of each battery, and select, from the batteries, a battery with a corresponding predicted degradation time less than a preset degradation time as a second abnormal battery;

[0051] obtain a target abnormal battery in the battery pack according to the first abnormal battery and the second abnormal battery.

[0052] In a fourth aspect, the present application also provides a computer readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the following steps:

[0053] obtain current multi-dimensional battery pack information of a battery pack to be diagnosed and current multi-dimensional battery information of each battery in the battery pack;

[0054] input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained first battery diagnosis model to obtain a first abnormal probability of each battery, and input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained second battery diagnosis model to obtain a second abnormal probability of each battery;

[0055] fuse the first abnormal probability and the second abnormal probability of each battery, respectively, to obtain a target abnormal probability of each battery;

[0056] select, from the batteries, a battery with a target abnormal probability greater than a preset abnormal probability as a first abnormal battery, and obtain a predicted degradation time of each battery, and select, from the batteries, a battery with a corresponding predicted degradation time less than a preset degradation time as a second abnormal battery;

[0057] obtain a target abnormal battery in the battery pack according to the first abnormal battery and the second abnormal battery.

[0058] In a fifth aspect, the present application also provides a computer program product comprising a computer program, the computer program being executed by a processor to implement the following steps:

[0059] obtain current multi-dimensional battery pack information of a battery pack to be diagnosed and current multi-dimensional battery information of each battery in the battery pack;

[0060] input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a first battery diagnosis model trained in advance to obtain a first abnormal probability of each battery, and input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a second battery diagnosis model trained in advance to obtain a second abnormal probability of each battery;

[0061] respectively fuse the first abnormal probability and the second abnormal probability of each battery to obtain a target abnormal probability of each battery;

[0062] select, from the batteries, a battery with a target abnormal probability greater than a preset abnormal probability as a first abnormal battery, and obtain a predicted degradation time of each battery, and select, from the batteries, a battery with a corresponding predicted degradation time less than a preset degradation time as a second abnormal battery;

[0063] obtain a target abnormal battery in the battery pack according to the first abnormal battery and the second abnormal battery.

[0064] The above battery abnormality diagnosis method, device, computer equipment, computer readable storage medium and computer program product, first acquire current multi-dimensional battery pack information of a battery pack to be diagnosed and current multi-dimensional battery information of each battery in the battery pack, input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained first battery diagnosis model to obtain a first abnormality probability of each battery, and input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained second battery diagnosis model to obtain a second abnormality probability of each battery, then fuse the first abnormality probability and the second abnormality probability of each battery respectively to obtain a target abnormality probability of each battery, then select, from each battery, a battery with a target abnormality probability greater than a preset abnormality probability as a first abnormal battery, and acquire a predicted degradation time of each battery, and select, from each battery, a battery with a corresponding predicted degradation time less than a preset degradation time as a second abnormal battery, and finally obtain a target abnormal battery in the battery pack according to the first abnormal battery and the second abnormal battery. In this way, when diagnosing the abnormality of the battery, the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack are comprehensively considered, and the pre-trained first battery diagnosis model and the second battery diagnosis model are combined, which is beneficial to improve the accuracy of the battery abnormality diagnosis. At the same time, the first abnormal battery with a target abnormality probability greater than a preset abnormality probability is selected from each battery, and the second abnormal battery with a corresponding predicted degradation time less than a preset degradation time is selected, and the target abnormal battery in the battery pack is obtained according to the first abnormal battery and the second abnormal battery, which is beneficial to further improve the accuracy of the battery abnormality diagnosis by comprehensively considering the target abnormality probability and the predicted degradation time of the battery. BRIEF DESCRIPTION OF DRAWINGS

[0065] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the present application or the related art. Obviously, the drawings in the following description only some embodiments of the present application, and for those skilled in the art, other related drawings can also be obtained without creative labor.

[0066] Figure 1 A flowchart of a battery abnormality diagnosis method in an embodiment;

[0067] Figure 2 A flowchart of a step of obtaining a first abnormality probability of each battery in an embodiment;

[0068] Figure 3 Flowchart of the battery abnormality diagnosis method in another embodiment;

[0069] Figure 4 Structural block diagram of the battery abnormality diagnosis device in an embodiment;

[0070] Figure 5 Internal structural diagram of the computer device in an embodiment. DETAILED DESCRIPTION

[0071] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application.

[0072] In the conventional technology, a single battery falling behind can cause the entire battery group to be unable to be normally used. The conventional maintenance method cannot comprehensively detect the battery and accurately find the falling behind battery, and can only replace the entire battery group, resulting in inaccurate battery diagnosis. Moreover, the conventional manual maintenance method is difficult to maintain according to the standard, and mostly only measures the voltage or internal resistance at a fixed time, which cannot accurately determine whether the battery is faulty. Based on this, the present application proposes a battery abnormality diagnosis method, which can improve the accuracy of battery abnormality diagnosis.

[0073] In an exemplary embodiment, as shown in FIG. 1, a battery abnormality diagnosis method is provided, and the present embodiment is exemplarily described by taking the method applied to a server; it can be understood that the method can also be applied to a terminal, and can also be applied to a system including a terminal and a server, and is realized through the interaction between the terminal and the server. The terminal can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, etc. The server can be a standalone physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. In the present embodiment, the following steps S101 to S105 are included. Among them:

[0074] In step S101, current multi-dimensional battery group information of a battery group to be diagnosed and current multi-dimensional battery information of each battery in the battery group are acquired.

[0075] The battery group to be diagnosed is composed of multiple batteries, which can be a lead-acid battery group. The battery can be a lead-acid battery.

[0076] The current multi-dimensional battery group information includes battery group information of multiple dimensions, such as group current, group voltage, ambient temperature, etc. The group current can be charging and discharging current, floating current, etc.

[0077] The current multi-dimension battery information includes battery information of multiple dimensions, such as liquid leakage, bulge, voltage, internal resistance, connection strip resistance, negative pole temperature, and the like.

[0078] Exemplarily, the server obtains, in response to an abnormality diagnosis request for the battery pack to be diagnosed, current multi-dimension battery pack information of the battery pack to be diagnosed and current multi-dimension battery information of each battery in the battery pack from the local database. Alternatively, the server obtains, in response to an abnormality diagnosis request for the battery pack to be diagnosed, current multi-dimension battery pack information of the battery pack to be diagnosed and current multi-dimension battery information of each battery in the battery pack collected by the battery intelligent monitoring and management system corresponding to the battery pack; wherein the battery intelligent monitoring and management system includes a battery pack module and a single battery module corresponding to each battery, the battery pack module is used to monitor multi-dimension battery pack information of the battery pack, and the single battery module is used to monitor multi-dimension battery information of the corresponding battery.

[0079] In step S102, the current multi-dimension battery pack information of the battery pack and the current multi-dimension battery information of each battery in the battery pack are input into a pre-trained first battery diagnosis model to obtain a first abnormality probability of each battery, and the current multi-dimension battery pack information of the battery pack and the current multi-dimension battery information of each battery in the battery pack are input into a pre-trained second battery diagnosis model to obtain a second abnormality probability of each battery.

[0080] The pre-trained first battery diagnosis model refers to a model used to output a first abnormality probability of a battery, such as a neural network model. The first abnormality probability refers to a probability of failure of the battery.

[0081] The pre-trained second battery diagnosis model refers to a model used to output a second abnormality probability of a battery, such as a deep learning model. The second abnormality probability refers to a probability of failure of the battery.

[0082] It should be noted that the model structures of the first battery diagnosis model and the second battery diagnosis model are different.

[0083] Exemplarily, the server inputs the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into the pre-trained first battery diagnosis model, performs a series of analysis and processing, such as feature extraction processing, probability calculation processing, etc., on the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack through the pre-trained first battery diagnosis model, and obtains the first abnormal probability of each battery. At the same time, the server inputs the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into the pre-trained second battery diagnosis model, performs a series of analysis and processing, such as feature extraction processing, probability calculation processing, etc., on the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack through the pre-trained second battery diagnosis model, and obtains the second abnormal probability of each battery.

[0084] Step S103, respectively, the first abnormal probability and the second abnormal probability of each battery are fused to obtain the target abnormal probability of each battery.

[0085] Among them, the target abnormal probability of the battery refers to the final probability of the battery failure, specifically refers to the fusion processing result of the first abnormal probability and the second abnormal probability of the battery, such as the weighted sum result of the first abnormal probability and the second abnormal probability of the battery.

[0086] Among them, the fusion processing can be weighted sum.

[0087] Exemplarily, the server respectively performs weighted sum processing on the first abnormal probability and the second abnormal probability of each battery to obtain the weighted sum processing result of each battery as the target abnormal probability of each battery.

[0088] Step S104, from each battery, the battery with a target abnormal probability greater than a preset abnormal probability is selected as a first abnormal battery, and the predicted degradation time of each battery is obtained, and the battery with a corresponding predicted degradation time less than a preset degradation time is selected from each battery as a second abnormal battery.

[0089] Among them, the preset abnormal probability refers to a pre-set probability value, such as 0.7, and the specific application is not limited.

[0090] Among them, the first abnormal battery refers to the battery in the battery pack with a corresponding target abnormal probability greater than the preset abnormal probability.

[0091] Among them, the predicted degradation time refers to the predicted time of the battery failure, specifically refers to the predicted time of the battery from good to bad, such as 3 days later, the battery from good to bad.

[0092] wherein the preset deterioration time refers to a preset deterioration time, such as 2 days, which is not limited in the present application.

[0093] wherein the second abnormal battery refers to a battery in the battery pack whose corresponding predicted deterioration time is less than the preset deterioration time.

[0094] Exemplarily, the server screens the batteries whose target abnormal probability is greater than the preset abnormal probability from the batteries, and takes them as the first abnormal batteries; then, the server obtains the historical multi-dimensional battery pack information of the battery pack and the historical multi-dimensional battery information of each battery in the battery pack from the local database, and then combines the current multi-dimensional battery pack information and the historical multi-dimensional battery pack information of the battery pack to obtain the combined multi-dimensional battery pack information of the battery pack, and combines the current multi-dimensional battery information and the historical multi-dimensional battery information of each battery in the battery pack to obtain the combined multi-dimensional battery information of each battery in the battery pack, and then inputs the combined multi-dimensional battery pack information of the battery pack and the combined multi-dimensional battery information of each battery in the battery pack into the pre-trained battery deterioration prediction model to obtain the predicted deterioration time of each battery, and finally screens the batteries whose corresponding predicted deterioration time is less than the preset deterioration time from the batteries, and takes them as the second abnormal batteries.

[0095] In step S105, the target abnormal battery in the battery pack is obtained according to the first abnormal battery and the second abnormal battery.

[0096] wherein the target abnormal battery refers to the finally determined abnormal battery, specifically, the union of the first abnormal battery and the second abnormal battery.

[0097] Exemplarily, the server takes both the first abnormal battery and the second abnormal battery as the target abnormal battery in the battery pack.

[0098] In the above battery abnormality diagnosis method, the current multi-dimensional battery pack information of the battery pack to be diagnosed and the current multi-dimensional battery information of each battery in the battery pack are first obtained, and the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack are input into a pre-trained first battery diagnosis model to obtain a first abnormality probability of each battery, and the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack are input into a pre-trained second battery diagnosis model to obtain a second abnormality probability of each battery. Then, the first abnormality probability and the second abnormality probability of each battery are fused to obtain a target abnormality probability of each battery. Then, from each battery, a battery with a target abnormality probability greater than a preset abnormality probability is selected as a first abnormal battery. The predicted degradation time of each battery is obtained, and from each battery, a battery with a corresponding predicted degradation time less than a preset degradation time is selected as a second abnormal battery. Finally, according to the first abnormal battery and the second abnormal battery, a target abnormal battery in the battery pack is obtained. In this way, when diagnosing the abnormality of the battery, the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack are comprehensively considered, and the pre-trained first battery diagnosis model and the second battery diagnosis model are combined, which is beneficial to improve the accuracy of the battery abnormality diagnosis. At the same time, from each battery, a first abnormal battery with a target abnormality probability greater than a preset abnormality probability is selected, and a second abnormal battery with a corresponding predicted degradation time less than a preset degradation time is selected, and according to the first abnormal battery and the second abnormal battery, a target abnormal battery in the battery pack is obtained. By comprehensively considering the target abnormality probability and the predicted degradation time of the battery, the accuracy of the battery abnormality diagnosis is further improved.

[0099] In one exemplary embodiment, as shown in Figure 2 The step S102 of inputting the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into the pre-trained first battery diagnosis model to obtain the first abnormality probability of each battery specifically includes the following steps S201 to S204. Wherein:

[0100] In step S201, the current multi-dimensional battery pack information and the current multi-dimensional battery information of each battery are respectively subjected to feature extraction processing to obtain a first feature vector corresponding to the current multi-dimensional battery pack information and a second feature vector corresponding to the current multi-dimensional battery information of each battery.

[0101] Step S202, respectively, the first feature vector and the second feature vector corresponding to the current multi-dimensional battery information of each battery are fused to obtain the fusion feature vector corresponding to the current multi-dimensional battery information of each battery.

[0102] Step S203, the fusion feature vector corresponding to the current multi-dimensional battery information of each battery is processed for feature extraction again to obtain the target feature vector corresponding to the current multi-dimensional battery information of each battery.

[0103] Step S204, the target feature vector corresponding to the current multi-dimensional battery information of each battery is input into the first battery diagnosis model trained in advance to obtain the first abnormal probability of each battery.

[0104] Among them, the first feature vector refers to the feature vector corresponding to the current multi-dimensional battery information; the second feature vector refers to the feature vector corresponding to the current multi-dimensional battery information.

[0105] Among them, the fusion feature vector corresponding to the current multi-dimensional battery information of the battery refers to the fusion processing result of the first feature vector and the second feature vector corresponding to the current multi-dimensional battery information of the battery.

[0106] Among them, the fusion feature vector is processed for feature extraction again, mainly to extract deeper features of the fusion feature vector, so that the final first abnormal probability is more accurate.

[0107] Among them, the target feature vector corresponding to the current multi-dimensional battery information of the battery refers to the feature extraction processing result of the fusion feature vector corresponding to the current multi-dimensional battery information of the battery.

[0108] Exemplarily, the server extracts features from the current multi-dimensional battery pack information and the current multi-dimensional battery information of each battery cell by a first feature extraction model (such as a first convolutional network model), respectively, to obtain a first feature vector corresponding to the current multi-dimensional battery pack information and a second feature vector corresponding to the current multi-dimensional battery information of each battery cell; then, the server fuses the first feature vector and the second feature vector corresponding to the current multi-dimensional battery information of each battery cell by a fusion processing instruction to obtain a fusion feature vector corresponding to the current multi-dimensional battery information of each battery cell; then, the server extracts features from the fusion feature vector corresponding to the current multi-dimensional battery information of each battery cell by a second feature extraction model (such as a second convolutional network model) to obtain a target feature vector corresponding to the current multi-dimensional battery information of each battery cell; finally, the server inputs the target feature vector corresponding to the current multi-dimensional battery information of each battery cell into a first battery diagnosis model trained in advance, and processes the target feature vector corresponding to the current multi-dimensional battery information of each battery cell by the first battery diagnosis model to obtain a first abnormal probability of each battery cell.

[0109] In addition, after obtaining the target feature vector corresponding to the current multi-dimensional battery information of each battery cell, the server can also input the target feature vector corresponding to the current multi-dimensional battery information of each battery cell into a second battery diagnosis model trained in advance, and process the target feature vector corresponding to the current multi-dimensional battery information of each battery cell by the second battery diagnosis model to obtain a second abnormal probability of each battery cell. It should be noted that the internal processing processes of the first battery diagnosis model and the second battery diagnosis model are different.

[0110] In this embodiment, the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery cell in the battery pack are comprehensively considered, and the first battery diagnosis model trained in advance is combined, which is beneficial to improve the determination accuracy of the first abnormal probability of the battery.

[0111] In an example embodiment, the pre-trained first battery diagnosis model is trained by: obtaining sample multi-dimensional battery pack information of a sample battery pack and sample multi-dimensional battery information of each sample battery in the sample battery pack; performing feature extraction processing on the sample multi-dimensional battery pack information and the sample multi-dimensional battery information of each sample battery, respectively, to obtain a first sample feature vector corresponding to the sample multi-dimensional battery pack information and a second sample feature vector corresponding to the sample multi-dimensional battery information of each sample battery; performing fusion processing on the first sample feature vector and the second sample feature vector corresponding to the sample multi-dimensional battery information of each sample battery, respectively, to obtain a sample fusion feature vector corresponding to the sample multi-dimensional battery information of each sample battery; performing re-feature extraction processing on the sample fusion feature vector corresponding to the sample multi-dimensional battery information of each sample battery, to obtain a target feature vector corresponding to the sample multi-dimensional battery information of each sample battery; inputting the target feature vector corresponding to the sample multi-dimensional battery information of each sample battery into the first battery diagnosis model to be trained, to obtain a predicted abnormal probability of each sample battery; obtaining an actual abnormal probability of each sample battery, and performing iterative training on the first battery diagnosis model to be trained according to a difference between the predicted abnormal probability and the actual abnormal probability of each sample battery, to obtain a trained first battery diagnosis model as the pre-trained first battery diagnosis model.

[0112] The first battery diagnosis model to be trained can be a neural network model.

[0113] The predicted abnormal probability of the battery refers to a predicted probability of failure of the battery; and the actual abnormal probability of the battery refers to an actual probability of failure of the battery.

[0114] Exemplarily, the server first obtains sample multi-dimensional battery pack information of a sample battery pack and sample multi-dimensional battery information of each sample battery in the sample battery pack from the local database, then performs feature extraction processing on the sample multi-dimensional battery pack information and the sample multi-dimensional battery information of each sample battery by using the first feature extraction model to obtain a first sample feature vector corresponding to the sample multi-dimensional battery pack information and a second sample feature vector corresponding to the sample multi-dimensional battery information of each sample battery, respectively; then performs fusion processing on the first sample feature vector and the second sample feature vector corresponding to the sample multi-dimensional battery information of each sample battery to obtain a sample fusion feature vector corresponding to the sample multi-dimensional battery information of each sample battery; then performs re-feature extraction processing on the sample fusion feature vector corresponding to the sample multi-dimensional battery information of each sample battery by using the second feature extraction model to obtain a target feature vector corresponding to the sample multi-dimensional battery information of each sample battery; inputs the target feature vector corresponding to the sample multi-dimensional battery information of each sample battery into the first battery diagnosis model to be trained, and performs a series of processing on the target feature vector corresponding to the sample multi-dimensional battery information of each sample battery by using the first battery diagnosis model to be trained to obtain a predicted abnormal probability of each sample battery; finally, the server obtains actual abnormal probabilities of each sample battery from the local database, and calculates a first loss value according to a difference between the predicted abnormal probability and the actual abnormal probability of each sample battery, iteratively trains the first battery diagnosis model to be trained according to the first loss value, and stops training until the first loss value based on the predicted abnormal probability output by the trained first battery diagnosis model is less than a first preset loss value, and the trained first battery diagnosis model is taken as the trained first battery diagnosis model, thereby obtaining the pre-trained first battery diagnosis model.

[0115] In addition, the training process of the pre-trained second battery diagnosis model can refer to the training process of the pre-trained first battery diagnosis model, which will not be described herein.

[0116] In this embodiment, the sample multi-dimensional battery pack information of the sample battery pack and the sample multi-dimensional battery information of each sample battery in the sample battery pack are used to iteratively train the first battery diagnosis model to be trained, which is beneficial to improve the accuracy of the first abnormal probability output by the trained first battery diagnosis model.

[0117] In an example embodiment, the step S103 of fusing the first abnormal probability and the second abnormal probability of each battery respectively to obtain the target abnormal probability of each battery specifically includes the following contents: obtaining the prediction accuracy of the first battery diagnosis model and the prediction accuracy of the second battery diagnosis model; querying the corresponding relationship between the prediction accuracy and the model weight according to the prediction accuracy of the first battery diagnosis model and the prediction accuracy of the second battery diagnosis model to obtain the first model weight of the first battery diagnosis model and the second model weight of the second battery diagnosis model; confirming the first model weight as the first weight corresponding to the first abnormal probability and confirming the second model weight as the second weight corresponding to the second abnormal probability; and fusing the first abnormal probability and the second abnormal probability of each battery respectively according to the first weight and the second weight to obtain the target abnormal probability of each battery.

[0118] wherein the prediction accuracy and the model weight have a one-to-one corresponding relationship.

[0119] wherein the target abnormal probability of the battery = the first abnormal probability of the battery x the first weight + the second abnormal probability of the battery x the second weight.

[0120] For example, the server obtains the prediction accuracy of the first battery diagnosis model and the prediction accuracy of the second battery diagnosis model from the local database; then queries the corresponding relationship between the prediction accuracy and the model weight according to the prediction accuracy of the first battery diagnosis model and the prediction accuracy of the second battery diagnosis model to obtain the first model weight corresponding to the prediction accuracy of the first battery diagnosis model as the first model weight of the first battery diagnosis model and obtain the second model weight corresponding to the prediction accuracy of the second battery diagnosis model as the second model weight of the second battery diagnosis model; then confirms the first model weight as the first weight corresponding to the first abnormal probability and confirms the second model weight as the second weight corresponding to the second abnormal probability; and finally sums the first abnormal probability and the second abnormal probability of each battery according to the first weight and the second weight to obtain the target abnormal probability of each battery, i.e., the target abnormal probability of the battery = the first abnormal probability of the battery x the first weight + the second abnormal probability of the battery x the second weight.

[0121] In the embodiment, when fusing the first abnormal probability and the second abnormal probability of each battery, the first weight corresponding to the first abnormal probability and the second weight corresponding to the second abnormal probability are comprehensively considered to make the finally determined target abnormal probability more accurate, thereby improving the determination accuracy of the target abnormal probability of each battery.

[0122] In an exemplary embodiment, the step S104 of obtaining the predicted degradation time of each battery cell specifically includes the following steps: obtaining historical multi-dimensional battery pack information of the battery pack and historical multi-dimensional battery information of each battery cell in the battery pack; combining the current multi-dimensional battery pack information and the historical multi-dimensional battery pack information to obtain combined multi-dimensional battery pack information of the battery pack, and combining the current multi-dimensional battery information and the historical multi-dimensional battery information of each battery cell in the battery pack to obtain combined multi-dimensional battery information of each battery cell in the battery pack; obtaining a feature vector of the combined multi-dimensional battery pack information and a feature vector of the combined multi-dimensional battery information of each battery cell in the battery pack; respectively performing fusion processing on the feature vector of the combined multi-dimensional battery pack information and the feature vector of the combined multi-dimensional battery information of each battery cell in the battery pack to obtain a fusion feature vector of the combined multi-dimensional battery information of each battery cell in the battery pack; and respectively inputting the fusion feature vector of the combined multi-dimensional battery information of each battery cell in the battery pack into a pre-trained battery degradation prediction model to obtain the predicted degradation time of each battery cell.

[0123] The historical multi-dimensional battery pack information includes historical battery pack information of multiple dimensions, and the historical multi-dimensional battery information includes historical battery information of multiple dimensions.

[0124] The pre-trained battery degradation prediction model is a model for outputting the predicted degradation time of the battery cell, such as a neural network model, a deep learning model, etc.

[0125] The fusion feature vector of the combined multi-dimensional battery information of each battery cell in the battery pack is the fusion processing result of the feature vector of the combined multi-dimensional battery pack information and the feature vector of the combined multi-dimensional battery information of the battery cell.

[0126] Exemplarily, the server obtains historical multi-dimensional battery pack information of the battery pack and historical multi-dimensional battery information of each battery in the battery pack from the local database; then combines the current multi-dimensional battery pack information and the historical multi-dimensional battery pack information to obtain combined multi-dimensional battery pack information of the battery pack, and combines the current multi-dimensional battery information and the historical multi-dimensional battery information of each battery in the battery pack to obtain combined multi-dimensional battery information of each battery in the battery pack; then performs feature extraction processing on the combined multi-dimensional battery pack information and the combined multi-dimensional battery information of each battery in the battery pack through a third feature extraction model (such as a third convolutional network model) to obtain a feature vector of the combined multi-dimensional battery pack information and a feature vector of the combined multi-dimensional battery information of each battery in the battery pack; then respectively fuses the feature vector of the combined multi-dimensional battery pack information and the feature vector of the combined multi-dimensional battery information of each battery in the battery pack to obtain a fusion feature vector of the combined multi-dimensional battery information of each battery in the battery pack; and finally respectively inputs the fusion feature vector of the combined multi-dimensional battery information of each battery in the battery pack into the pre-trained battery degradation prediction model, and performs a series of processing on the fusion feature vector of the combined multi-dimensional battery information of each battery through the pre-trained battery degradation prediction model to obtain a predicted degradation time of each battery.

[0127] In this embodiment, the historical multi-dimensional battery pack information and the current multi-dimensional battery pack information of the battery pack, and the historical multi-dimensional battery information and the current multi-dimensional battery information of each battery in the battery pack are comprehensively considered, and the pre-trained battery degradation prediction model is combined, which is beneficial to improve the determination accuracy of the predicted degradation time of the battery.

[0128] In an exemplary embodiment, the pre-trained battery degradation prediction model is trained by: obtaining sample combined multi-dimensional battery pack information of a sample battery pack and sample combined multi-dimensional battery information of each sample battery in the sample battery pack; obtaining a feature vector of the sample combined multi-dimensional battery pack information and a feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack; respectively performing fusion processing on the feature vector of the sample combined multi-dimensional battery pack information and the feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack to obtain a fusion feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack; inputting the fusion feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack into the battery degradation prediction model to be trained to obtain a predicted degradation time of each sample battery; obtaining an actual degradation time of each sample battery, and iteratively training the battery degradation prediction model to be trained according to a difference between the predicted degradation time and the actual degradation time of each sample battery to obtain a trained battery degradation prediction model as the pre-trained battery degradation prediction model.

[0129] The sample combined multi-dimensional battery pack information of the sample battery pack is obtained by combining sample current multi-dimensional battery pack information and sample historical multi-dimensional battery pack information of the sample battery pack.

[0130] The sample combined multi-dimensional battery information of the sample battery is obtained by combining sample current multi-dimensional battery information and sample historical multi-dimensional battery information of the sample battery.

[0131] The battery degradation prediction model to be trained refers to a neural network model, a deep learning model, etc.

[0132] Exemplarily, the server first acquires sample combined multi-dimensional battery pack information of a sample battery pack and sample combined multi-dimensional battery information of each sample battery in the sample battery pack from the local database; then, the server performs feature extraction processing on the sample combined multi-dimensional battery pack information and the sample combined multi-dimensional battery information of each sample battery in the sample battery pack by using the third feature extraction model to obtain a feature vector of the sample combined multi-dimensional battery pack information and a feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack; then, the server respectively fuses the feature vector of the sample combined multi-dimensional battery pack information and the feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack by using a fusion processing instruction to obtain a fusion feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack; then, the server respectively inputs the fusion feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack into the battery degradation prediction model to be trained, and performs a series of processing, such as time prediction processing, on the fusion feature vector of the sample combined multi-dimensional battery information of each sample battery by using the battery degradation prediction model to be trained to obtain a predicted degradation time of each sample battery; finally, the server acquires actual degradation times of each sample battery from the local database, and calculates a second loss value according to a difference between the predicted degradation time and the actual degradation time of each sample battery, iteratively trains the battery degradation prediction model to be trained according to the second loss value, and stops the training when the second loss value obtained based on the predicted degradation time output by the trained battery degradation prediction model is less than a second preset loss value, so as to obtain the trained battery degradation prediction model, thereby obtaining the pre-trained battery degradation prediction model.

[0133] In this embodiment, the sample combined multi-dimensional battery pack information of the sample battery pack and the sample combined multi-dimensional battery information of each sample battery in the sample battery pack are used to iteratively train the battery degradation prediction model to be trained, which is beneficial to improving the accuracy of the predicted degradation time output by the trained battery degradation prediction model.

[0134] In one exemplary embodiment, as shown in Figure 3 Another battery abnormality diagnosis method is provided, which is applied to a server as an example and includes the following steps S301 to S312. Wherein:

[0135] In step S301, current multi-dimensional battery pack information of a battery pack to be diagnosed and current multi-dimensional battery information of each battery in the battery pack are acquired.

[0136] Step S302, input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into the pre-trained first battery diagnosis model to obtain the first abnormal probability of each battery.

[0137] Step S303, input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into the pre-trained second battery diagnosis model to obtain the second abnormal probability of each battery.

[0138] Step S304, obtain the prediction accuracy of the first battery diagnosis model and the prediction accuracy of the second battery diagnosis model.

[0139] Step S305, according to the prediction accuracy of the first battery diagnosis model and the prediction accuracy of the second battery diagnosis model, query the corresponding relationship between the prediction accuracy and the model weight to obtain the first model weight of the first battery diagnosis model and the second model weight of the second battery diagnosis model.

[0140] Step S306, confirm the first model weight as the first weight corresponding to the first abnormal probability, and confirm the second model weight as the second weight corresponding to the second abnormal probability; according to the first weight and the second weight, respectively, fuse the first abnormal probability and the second abnormal probability of each battery to obtain the target abnormal probability of each battery.

[0141] Step S307, from each battery, select the battery with a target abnormal probability greater than a preset abnormal probability as a first abnormal battery.

[0142] Step S308, obtain the historical multi-dimensional battery pack information of the battery pack and the historical multi-dimensional battery information of each battery in the battery pack.

[0143] Step S309, combine the current multi-dimensional battery pack information and the historical multi-dimensional battery pack information of the battery pack to obtain the combined multi-dimensional battery pack information of the battery pack, and combine the current multi-dimensional battery information and the historical multi-dimensional battery information of each battery in the battery pack to obtain the combined multi-dimensional battery information of each battery in the battery pack.

[0144] Step S310, input the combined multi-dimensional battery pack information of the battery pack and the combined multi-dimensional battery information of each battery in the battery pack into the pre-trained battery degradation prediction model to obtain the predicted degradation time of each battery.

[0145] Step S311, from each battery, select the battery with a corresponding predicted degradation time less than a preset degradation time as a second abnormal battery.

[0146] Step S312, according to the first abnormal battery and the second abnormal battery, obtaining the target abnormal battery in the battery pack.

[0147] In the above battery abnormality diagnosis method, when diagnosing the abnormality of the battery, the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack are comprehensively considered, and the first battery diagnosis model and the second battery diagnosis model trained in advance are combined, which is beneficial to improve the accuracy of the battery abnormality diagnosis. At the same time, from each battery, the first abnormal battery with a target abnormal probability greater than a preset abnormal probability is screened out, and the second abnormal battery with a corresponding predicted degradation time less than a preset degradation time is screened out, and according to the first abnormal battery and the second abnormal battery, the target abnormal battery in the battery pack is obtained. By comprehensively considering the target abnormal probability and the predicted degradation time of the battery, it is beneficial to further improve the accuracy of the battery abnormality diagnosis.

[0148] In order to more clearly illustrate the battery abnormality diagnosis method provided by the embodiments of the present application, the battery abnormality diagnosis method will be specifically described below with one specific embodiment. In an exemplary embodiment, another battery abnormality diagnosis method is provided. By analyzing the current battery pack information (such as group voltage, current (such as charging and discharging current, floating current), environmental temperature) of the battery pack and the current multi-dimensional information (such as single battery leakage, bulge, voltage, internal resistance, connection strip resistance, negative pole temperature, etc.) of each single battery in the battery pack, the abnormal single battery in the battery pack can be accurately found out. The specific content is as follows:

[0149] (1) Obtain the current battery pack information (such as group voltage, current (such as charging and discharging current, floating current), environmental temperature) of the battery pack and the current multi-dimensional information (such as single battery leakage, bulge, voltage, internal resistance, connection strip resistance, negative pole temperature, etc.) of each single battery in the battery pack.

[0150] (2) Use the first battery diagnosis model to analyze the current battery pack information of the battery pack and the current multi-dimensional information of each single battery in the battery pack, and obtain the first failure probability of each single battery; use the second battery diagnosis model to analyze the current battery pack information of the battery pack and the current multi-dimensional information of each single battery in the battery pack, and obtain the second failure probability of each single battery; combine the first failure probability and the second failure probability of each single battery to obtain the target failure probability of each single battery.

[0151] (3) From each single battery, screen out the battery with a target failure probability greater than a preset threshold as a first failure battery.

[0152] (4) Obtain historical battery pack information (such as pack voltage, current (such as charging and discharging current, floating current), ambient temperature) of the battery pack, and historical multi-dimensional information (such as single cell leakage, bulge, voltage, internal resistance, connection strip resistance, negative pole temperature, etc.) of each single battery in the battery pack, and use a model or algorithm to analyze the historical battery pack information and the current battery pack information, the historical multi-dimensional information and the current multi-dimensional information of each single battery, to obtain the degradation trend of each single battery, and finally determine the second fault battery based on the degradation trend of each single battery.

[0153] (5) Obtain the final fault battery according to the first fault battery and the second fault battery, and perform an alarm.

[0154] The above embodiments, when diagnosing the battery in the battery pack, comprehensively consider the current battery pack information of the battery pack and the current multi-dimensional information of each single battery in the battery pack, which is beneficial to improve the accuracy of the battery abnormality diagnosis.

[0155] It should be understood that, although each step in the flowchart involved in each embodiment as described above is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other sequences. Moreover, at least part of the steps in the flowchart involved in each embodiment as described above can include multiple steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed alternately or alternately with at least part of other steps or steps or stages in other steps.

[0156] Based on the same inventive concept, the embodiments of the present application also provide a battery abnormality diagnosis device for implementing the above-mentioned battery abnormality diagnosis method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more battery abnormality diagnosis device embodiments provided below can refer to the limitations of the battery abnormality diagnosis method in the above text, and will not be repeated here.

[0157] In one exemplary embodiment, as shown in Figure 4 a battery abnormality diagnosis device is provided, which includes an information acquisition module 410, a probability determination module 420, a probability fusion module 430, a battery screening module 440, and a battery diagnosis module 450, wherein:

[0158] The information acquisition module 410 is configured to acquire current multi-dimensional battery pack information of the battery pack to be diagnosed and current multi-dimensional battery information of each battery in the battery pack.

[0159] The probability determination module 420 is configured to input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained first battery diagnosis model to obtain a first abnormal probability of each battery, and input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained second battery diagnosis model to obtain a second abnormal probability of each battery.

[0160] The probability fusion module 430 is configured to fuse the first abnormal probability and the second abnormal probability of each battery, respectively, to obtain a target abnormal probability of each battery.

[0161] The battery screening module 440 is configured to screen, from the batteries, a battery with a target abnormal probability greater than a preset abnormal probability as a first abnormal battery, and acquire a predicted degradation time of each battery, and screen, from the batteries, a second abnormal battery with a corresponding predicted degradation time less than a preset degradation time.

[0162] The battery diagnosis module 450 is configured to obtain a target abnormal battery in the battery pack according to the first abnormal battery and the second abnormal battery.

[0163] In an exemplary embodiment, the probability determination module 420 is further configured to perform feature extraction processing on the current multi-dimensional battery pack information and the current multi-dimensional battery information of each battery, respectively, to obtain a first feature vector corresponding to the current multi-dimensional battery pack information and a second feature vector corresponding to the current multi-dimensional battery information of each battery; perform fusion processing on the first feature vector and the second feature vector corresponding to the current multi-dimensional battery information of each battery, respectively, to obtain a fusion feature vector corresponding to the current multi-dimensional battery information of each battery; perform re-feature extraction processing on the fusion feature vector corresponding to the current multi-dimensional battery information of each battery, respectively, to obtain a target feature vector corresponding to the current multi-dimensional battery information of each battery; and input the target feature vector corresponding to the current multi-dimensional battery information of each battery into the pre-trained first battery diagnosis model to obtain the first abnormal probability of each battery.

[0164] In an example embodiment, the battery anomaly diagnosis apparatus further comprises a first training module configured to obtain sample multi-dimensional battery pack information of a sample battery pack and sample multi-dimensional battery information of each sample battery in the sample battery pack; perform feature extraction processing on the sample multi-dimensional battery pack information and the sample multi-dimensional battery information of each sample battery, respectively, to obtain a first sample feature vector corresponding to the sample multi-dimensional battery pack information and a second sample feature vector corresponding to the sample multi-dimensional battery information of each sample battery; perform fusion processing on the first sample feature vector and the second sample feature vector corresponding to the sample multi-dimensional battery information of each sample battery, respectively, to obtain a sample fusion feature vector corresponding to the sample multi-dimensional battery information of each sample battery; perform re-feature extraction processing on the sample fusion feature vector corresponding to the sample multi-dimensional battery information of each sample battery, respectively, to obtain a target feature vector corresponding to the sample multi-dimensional battery information of each sample battery; input the target feature vector corresponding to the sample multi-dimensional battery information of each sample battery into the first battery diagnosis model to be trained, to obtain a predicted abnormal probability of each sample battery; obtain an actual abnormal probability of each sample battery, and perform iterative training on the first battery diagnosis model to be trained according to a difference between the predicted abnormal probability and the actual abnormal probability of each sample battery, to obtain a trained first battery diagnosis model as the pre-trained first battery diagnosis model.

[0165] In an example embodiment, the probability fusion module 430 is further configured to obtain a prediction accuracy of the first battery diagnosis model and a prediction accuracy of the second battery diagnosis model; query a corresponding relationship between prediction accuracy and model weight according to the prediction accuracy of the first battery diagnosis model and the prediction accuracy of the second battery diagnosis model, to obtain a first model weight of the first battery diagnosis model and a second model weight of the second battery diagnosis model; confirm the first model weight as a first weight corresponding to the first abnormal probability and the second model weight as a second weight corresponding to the second abnormal probability; and perform fusion processing on the first abnormal probability and the second abnormal probability of each battery according to the first weight and the second weight, respectively, to obtain a target abnormal probability of each battery.

[0166] In an example embodiment, the battery screening module 440 is further configured to acquire historical multi-dimensional battery pack information of the battery pack and historical multi-dimensional battery information of each battery in the battery pack; combine the current multi-dimensional battery pack information and the historical multi-dimensional battery pack information to obtain combined multi-dimensional battery pack information of the battery pack, and combine the current multi-dimensional battery information and the historical multi-dimensional battery information of each battery in the battery pack to obtain combined multi-dimensional battery information of each battery in the battery pack; acquire a feature vector of the combined multi-dimensional battery pack information and a feature vector of the combined multi-dimensional battery information of each battery in the battery pack; respectively fuse the feature vector of the combined multi-dimensional battery pack information and the feature vector of the combined multi-dimensional battery information of each battery in the battery pack to obtain a fused feature vector of the combined multi-dimensional battery information of each battery in the battery pack; and input the fused feature vector of the combined multi-dimensional battery information of each battery in the battery pack into the pre-trained battery degradation prediction model to obtain a predicted degradation time of each battery.

[0167] In an example embodiment, the battery abnormality diagnosis apparatus further comprises a second training module configured to acquire sample combined multi-dimensional battery pack information of a sample battery pack and sample combined multi-dimensional battery information of each sample battery in the sample battery pack; acquire a feature vector of the sample combined multi-dimensional battery pack information and a feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack; respectively fuse the feature vector of the sample combined multi-dimensional battery pack information and the feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack to obtain a fused feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack; input the fused feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack into the battery degradation prediction model to be trained to obtain a predicted degradation time of each sample battery; acquire an actual degradation time of each sample battery, and iteratively train the battery degradation prediction model to be trained according to a difference between the predicted degradation time and the actual degradation time of each sample battery to obtain a trained battery degradation prediction model as the pre-trained battery degradation prediction model.

[0168] Each module in the above battery abnormality diagnosis apparatus can be realized by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in hardware form, or stored in a memory in the computer device in software form to be called and executed by the processor.

[0169] In an exemplary embodiment, a computer device, which can be a server, is provided, and an internal structure diagram thereof can be as shown in FIG. Figure 5 The computer device includes a processor, a memory, an input / output interface, and a communication interface. The processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The database of the computer device is configured to store data such as a target abnormal probability of a storage battery and a predicted degradation time. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with terminals outside through a network connection. The computer program is executed by the processor to implement a storage battery abnormality diagnosis method.

[0170] Those skilled in the art can understand that Figure 5 The structure shown in FIG.

[0171] In an embodiment, a computer device is also provided, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0172] In an embodiment, a computer readable storage medium is provided, which stores a computer program. The computer program is executed by a processor to implement the steps in the above method embodiments.

[0173] In an embodiment, a computer program product is provided, which includes a computer program. The computer program is executed by a processor to implement the steps in the above method embodiments.

[0174] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by users or authorized by all parties, and the collection, use, and processing of related data need to comply with relevant regulations.

[0175] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. The non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. The volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., without being limited thereto.

[0176] The technical features of the above embodiments can be combined in any manner. To make the description concise, all possible combinations of the technical features in the above embodiments are not described, but as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present application.

[0177] The above-described embodiments are merely illustrative of several embodiments of the present application, which are described in more detail and in a specific manner, but should not be construed as limiting the scope of the patent of the present application. It should be noted that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A storage battery abnormality diagnosing method characterized by comprising: The method comprises: obtaining current multi-dimensional battery pack information of a battery pack to be diagnosed and current multi-dimensional battery information of each battery in the battery pack; inputting the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a first battery diagnosis model trained in advance to obtain a first abnormal probability of each battery, and inputting the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a second battery diagnosis model trained in advance to obtain a second abnormal probability of each battery; obtaining a prediction accuracy of the first battery diagnosis model and a prediction accuracy of the second battery diagnosis model; querying a corresponding relationship between prediction accuracies and model weights according to the prediction accuracy of the first battery diagnosis model and the prediction accuracy of the second battery diagnosis model to obtain a first model weight of the first battery diagnosis model and a second model weight of the second battery diagnosis model; confirming the first model weight as a first weight corresponding to the first abnormal probability and confirming the second model weight as a second weight corresponding to the second abnormal probability; and fusing the first abnormal probability and the second abnormal probability of each battery according to the first weight and the second weight to obtain a target abnormal probability of each battery; from the batteries, selecting a battery with a target abnormal probability greater than a preset abnormal probability as a first abnormal battery, obtaining historical multi-dimensional battery pack information of the battery pack and historical multi-dimensional battery information of each battery in the battery pack; combining the current multi-dimensional battery pack information and the historical multi-dimensional battery pack information of the battery pack to obtain combined multi-dimensional battery pack information of the battery pack, combining the current multi-dimensional battery information and the historical multi-dimensional battery information of each battery in the battery pack to obtain combined multi-dimensional battery information of each battery in the battery pack; obtaining a feature vector of the combined multi-dimensional battery pack information and a feature vector of the combined multi-dimensional battery information of each battery in the battery pack; fusing the feature vector of the combined multi-dimensional battery pack information and the feature vector of the combined multi-dimensional battery information of each battery in the battery pack to obtain a fused feature vector of the combined multi-dimensional battery information of each battery in the battery pack; and inputting the fused feature vector of the combined multi-dimensional battery information of each battery in the battery pack into a battery degradation prediction model trained in advance to obtain a predicted degradation time of each battery, and selecting a battery with a corresponding predicted degradation time less than a preset degradation time from the batteries as a second abnormal battery; obtaining a target abnormal battery in the battery pack according to the first abnormal battery and the second abnormal battery.

2. The method of claim 1, wherein, The current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack are input into a pre-trained first battery diagnosis model to obtain a first abnormal probability of each battery, including: The current multi-dimensional battery pack information and the current multi-dimensional battery information of each battery are respectively subjected to feature extraction processing to obtain a first feature vector corresponding to the current multi-dimensional battery pack information and a second feature vector corresponding to the current multi-dimensional battery information of each battery; The first feature vector and the second feature vector corresponding to the current multi-dimensional battery information of each battery are respectively subjected to fusion processing to obtain a fusion feature vector corresponding to the current multi-dimensional battery information of each battery; The fusion feature vector corresponding to the current multi-dimensional battery information of each battery is subjected to re-feature extraction processing to obtain a target feature vector corresponding to the current multi-dimensional battery information of each battery; The target feature vector corresponding to the current multi-dimensional battery information of each battery is input into the pre-trained first battery diagnosis model to obtain a first abnormal probability of each battery.

3. The method of claim 1, wherein, The pre-trained first battery diagnosis model is obtained by training in the following manner: Sample multi-dimensional battery pack information of a sample battery pack and sample multi-dimensional battery information of each sample battery in the sample battery pack are obtained; The sample multi-dimensional battery pack information and the sample multi-dimensional battery information of each sample battery are respectively subjected to feature extraction processing to obtain a first sample feature vector corresponding to the sample multi-dimensional battery pack information and a second sample feature vector corresponding to the sample multi-dimensional battery information of each sample battery; The first sample feature vector and the second sample feature vector corresponding to the sample multi-dimensional battery information of each sample battery are respectively subjected to fusion processing to obtain a sample fusion feature vector corresponding to the sample multi-dimensional battery information of each sample battery; The sample fusion feature vector corresponding to the sample multi-dimensional battery information of each sample battery is subjected to re-feature extraction processing to obtain a target feature vector corresponding to the sample multi-dimensional battery information of each sample battery; The target feature vector corresponding to the sample multi-dimensional battery information of each sample battery is input into a first battery diagnosis model to be trained to obtain a predicted abnormal probability of each sample battery; Actual abnormal probabilities of the sample batteries are obtained, and the first battery diagnosis model to be trained is iteratively trained according to differences between the predicted abnormal probabilities and the actual abnormal probabilities of the sample batteries, to obtain a trained first battery diagnosis model as the pre-trained first battery diagnosis model.

4. The method of claim 1, wherein, The pre-trained battery degradation prediction model is obtained by training in the following manner: Sample combined multi-dimensional battery pack information of a sample battery pack and sample combined multi-dimensional battery information of each sample battery in the sample battery pack are obtained; Obtain the feature vector of the sample combined multi-dimensional battery pack information and the feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack; Respectively fuse the feature vector of the sample combined multi-dimensional battery pack information and the feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack, to obtain the fusion feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack; Respectively input the fusion feature vector of the sample combined multi-dimensional battery information of each sample battery in the sample battery pack into the battery degradation prediction model to be trained, to obtain the predicted degradation time of each sample battery; Obtain the actual degradation time of each sample battery, and according to the difference between the predicted degradation time and the actual degradation time of each sample battery, iteratively train the battery degradation prediction model to be trained, to obtain the trained battery degradation prediction model as the pre-trained battery degradation prediction model.

5. A storage battery abnormality diagnosing device characterized by comprising: The device comprises: An information acquisition module configured to acquire current multi-dimensional battery pack information of a battery pack to be diagnosed and current multi-dimensional battery information of each battery in the battery pack; A probability determination module configured to input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained first battery diagnosis model, to obtain a first abnormal probability of each battery, and input the current multi-dimensional battery pack information of the battery pack and the current multi-dimensional battery information of each battery in the battery pack into a pre-trained second battery diagnosis model, to obtain a second abnormal probability of each battery; A probability fusion module configured to acquire a prediction accuracy of the first battery diagnosis model and a prediction accuracy of the second battery diagnosis model, query a corresponding relationship between prediction accuracy and model weight according to the prediction accuracy of the first battery diagnosis model and the prediction accuracy of the second battery diagnosis model, to obtain a first model weight of the first battery diagnosis model and a second model weight of the second battery diagnosis model, confirm the first model weight as a first weight corresponding to the first abnormal probability, and confirm the second model weight as a second weight corresponding to the second abnormal probability, and fuse the first abnormal probability and the second abnormal probability of each battery according to the first weight and the second weight, to obtain a target abnormal probability of each battery. The battery screening module is configured to screen, from the batteries, a battery with an abnormal probability greater than a preset abnormal probability as a first abnormal battery, and obtain historical multi-dimensional battery pack information of the battery pack and historical multi-dimensional battery information of each battery in the battery pack; combine the current multi-dimensional battery pack information and the historical multi-dimensional battery pack information to obtain combined multi-dimensional battery pack information of the battery pack, and combine the current multi-dimensional battery information and the historical multi-dimensional battery information of each battery in the battery pack to obtain combined multi-dimensional battery information of each battery in the battery pack; obtain a feature vector of the combined multi-dimensional battery pack information and a feature vector of the combined multi-dimensional battery information of each battery in the battery pack; respectively fuse the feature vector of the combined multi-dimensional battery pack information and the feature vector of the combined multi-dimensional battery information of each battery in the battery pack to obtain a fused feature vector of the combined multi-dimensional battery information of each battery in the battery pack; and respectively input the fused feature vector of the combined multi-dimensional battery information of each battery in the battery pack into a pre-trained battery degradation prediction model to obtain a predicted degradation time of each battery, and screen, from the batteries, a battery with a predicted degradation time less than a preset degradation time as a second abnormal battery. The battery diagnosis module is configured to obtain a target abnormal battery in the battery pack according to the first abnormal battery and the second abnormal battery.

6. The apparatus of claim 5, wherein, The probability determination module is further configured to respectively perform feature extraction processing on the current multi-dimensional battery pack information and the current multi-dimensional battery information of each battery to obtain a first feature vector corresponding to the current multi-dimensional battery pack information and a second feature vector corresponding to the current multi-dimensional battery information of each battery; respectively perform fusion processing on the first feature vector and the second feature vector corresponding to the current multi-dimensional battery information of each battery to obtain a fused feature vector corresponding to the current multi-dimensional battery information of each battery; respectively perform re-feature extraction processing on the fused feature vector corresponding to the current multi-dimensional battery information of each battery to obtain a target feature vector corresponding to the current multi-dimensional battery information of each battery; and respectively input the target feature vector corresponding to the current multi-dimensional battery information of each battery into a pre-trained first battery diagnosis model to obtain a first abnormal probability of each battery.

7. The apparatus of claim 5, wherein, The device further comprises a first training module configured to acquire sample multi-dimensional battery pack information of a sample battery pack and sample multi-dimensional battery information of each sample battery in the sample battery pack; perform feature extraction processing on the sample multi-dimensional battery pack information and the sample multi-dimensional battery information of each sample battery, respectively, to obtain a first sample feature vector corresponding to the sample multi-dimensional battery pack information and a second sample feature vector corresponding to the sample multi-dimensional battery information of each sample battery; perform fusion processing on the first sample feature vector and the second sample feature vector corresponding to the sample multi-dimensional battery information of each sample battery, respectively, to obtain a sample fusion feature vector corresponding to the sample multi-dimensional battery information of each sample battery; perform re-feature extraction processing on the sample fusion feature vector corresponding to the sample multi-dimensional battery information of each sample battery, respectively, to obtain a target feature vector corresponding to the sample multi-dimensional battery information of each sample battery; input the target feature vector corresponding to the sample multi-dimensional battery information of each sample battery into a first battery diagnosis model to be trained, to obtain a predicted abnormal probability of each sample battery; acquire an actual abnormal probability of each sample battery, and perform iterative training on the first battery diagnosis model to be trained according to a difference between the predicted abnormal probability and the actual abnormal probability of each sample battery, to obtain a trained first battery diagnosis model as the pre-trained first battery diagnosis model. 8.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-7. The computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 4.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 4.

10. A computer program product comprising a computer program, characterized in that, The computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 4. The computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 4.

Images