Vehicle battery fault identification method and device and electronic equipment
By using the battery fault identification model of the filter layer and the long and short-term memory network layer in vehicle battery fault detection, combined with the multi-head attention layer and the full connection layer, the problem of insufficient battery fault detection accuracy in the existing technology is solved, and the more High recognition accuracy and reliability enhance the monitoring and early warning capabilities of battery status.
Patent Information
- Application Number
- CN202411213633.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-30
- Publication Date
- 2024-12-06
AI Technical Summary
The existing vehicle battery failure detection methods have shortcomings in terms of accuracy and reliability, especially under real-time data processing and hardware resource limitations, it is difficult to effectively identify battery failures, and false alarms or incomplete identification are prone to problems.
Using a battery fault identification model based on the filter layer and the long and short-term memory network layer, the current operating data of the vehicle battery is reduced and the timing feature extraction is extracted, and combined with the multi-head attention layer and the full connection layer, the accurate identification of battery faults is achieved. and prediction.
It improves the accuracy and reliability of vehicle battery fault identification, reduces the occurrence of false alarms, enhances real-time monitoring and early warning capabilities of battery status, and ensures the normal operation and safety of vehicle batteries.
Smart Images

Figure CN119089271A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of battery technology, and in particular to a vehicle battery fault identification method, device and electronic equipment. Background Art
[0002] Electric vehicles are of great significance in ensuring energy security and reducing environmental pollution. However, due to the complexity of the electrochemical system and the uncertainty of the operating environment, the power battery, as a core component, has become the main source of electric vehicle failure, which is also an important reason hindering the further popularization of vehicles. How to efficiently and accurately identify battery failures to avoid safety hazards has become an urgent problem to be solved.
[0003] Some battery fault detection methods in related technologies have been developed and applied, but there are still many diagnostic strategies that can only realize the analysis of experimental or offline data. There are two main reasons: on the one hand, some advanced fault detection methods require calculation of a large amount of high-quality data, which also puts higher requirements on hardware, but the current on-board battery management system cannot meet them; on the other hand, some fault detection methods have low fault identification accuracy, which may lead to problems such as failure to properly identify existing faults or false fault alarms.
[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention
[0005] The embodiments of the present invention provide a vehicle battery fault identification method, device and electronic device to at least solve the technical problem of inaccurate vehicle battery fault identification existing in the related art.
[0006] According to one aspect of an embodiment of the present invention, a vehicle battery fault identification method is provided, comprising: obtaining current operating data of a vehicle battery in a current sampling period; performing noise reduction processing on the current operating data using a filter layer in a battery fault identification model to obtain noise-reduced data, wherein the battery fault identification model comprises the filter layer and a long short-term memory network layer; based on the noise-reduced data, using the long short-term memory network layer to obtain a current fault identification result of the vehicle battery, wherein the long short-term memory network layer is used to extract temporal dependencies in the data; wherein the battery fault identification model is obtained through machine learning based on multiple groups of historical operating data of the vehicle battery in historical time periods, and corresponding historical fault identification results.
[0007] Through the noise reduction processing of the filter layer and the time series modeling capability of the long short-term memory network, the model can more accurately identify the fault conditions of the vehicle battery and improve the accuracy and reliability of fault identification. The use of long short-term memory networks can better capture the laws and changing trends in time series data, which helps to accurately predict battery failures and warn of possible problems in advance. Based on the current operating data of the vehicle battery, the battery fault identification model including the filter layer and the long short-term memory network layer is used to identify the fault of the vehicle battery, which can effectively improve the accuracy and reliability of fault identification, improve the fault prediction effect, and help ensure the normal operation and safety of the vehicle battery.
[0008] Optionally, based on the denoised data, the long short-term memory network layer is used to obtain the current fault identification result of the vehicle battery, including: based on the denoised data, the long short-term memory network layer is used to obtain time series data; based on the time series data, the multi-head attention layer is used to obtain the current fault identification result, wherein the multi-head attention layer is used to extract position dependencies in the input sequence.
[0009] Using the long short-term memory network layer to obtain time series data and then using the multi-head attention layer for fault identification can effectively extract the time series features and position dependencies in the data, help accurately identify faults and improve the prediction and generalization capabilities of the model. The above operations will help improve the accuracy and reliability of vehicle battery fault identification.
[0010] Optionally, when the battery fault identification model also includes a fully connected layer, the multi-head attention layer is used based on the time series data to obtain the current fault identification result, including: based on the time series data, the multi-head attention layer is used to obtain the attention result; based on the attention result, the fully connected layer is used to obtain the current fault identification result, wherein the fully connected layer is used to map the data to the fault category space.
[0011] Using multi-head attention layer and fully connected layer for fault identification based on time series data can make full use of the time series features and position dependencies in the data, improve the representation ability and classification accuracy of the battery fault identification model, help accurately identify faults and map data to fault category space. These operations will help improve the performance and accuracy of vehicle battery fault identification.
[0012] Optionally, in the case where the battery fault identification model also includes a convolution layer and a pooling layer, the long short-term memory network layer is used to obtain time series data based on the denoised data, including: using the convolution layer to perform feature extraction processing on the denoised data to obtain feature data; using the pooling layer to perform dimensionality reduction processing on the feature data to obtain reduced dimensionality data; based on the reduced dimensionality data, using the long short-term memory network layer to obtain the time series data.
[0013] Using the convolution layer for feature extraction and the pooling layer for dimensionality reduction, and then obtaining the time series data through the long short-term memory network layer, can make full use of the spatial and temporal characteristics of the data, improve the model's ability to understand and represent the data, and reduce the computational complexity and memory consumption of the battery fault identification model.
[0014] Optionally, when the battery fault identification model also includes a normalization layer, the fully connected layer is used based on the attention result to obtain the current fault identification result, including: using the normalization layer to normalize the attention result to obtain a normalized result; using the fully connected layer to map the normalized result to the fault category space to obtain the current fault identification result.
[0015] By setting up data normalization processing on the alienation layer between the multi-head attention layer and the fully connected layer, the dimensional influence between the data can be eliminated, which helps to improve the stability and generalization ability of the battery fault identification model.
[0016] Optionally, the using of the filtering layer in the battery fault identification model to perform noise reduction processing on the current operating data to obtain the noise reduced data includes: using the sliding mean filtering method in the filtering layer to perform noise reduction processing on the current operating data to obtain the noise reduced data.
[0017] The sliding mean filter method can smooth the data by calculating the mean around the input data point, making the data more stable and continuous, which is conducive to subsequent data analysis and model building. At the same time, the sliding mean filter can eliminate high-frequency noise in the data, reduce the interference of noise on data analysis and prediction results, and improve the quality and reliability of the data.
[0018] Optionally, obtaining the current operating data of the vehicle battery in the current sampling period includes: obtaining initial operating data of the vehicle battery corresponding to multiple moments in the current sampling period, wherein the initial operating data includes at least the voltage, current, state of charge, insulation value, and temperature of the vehicle battery at the corresponding moments; and obtaining the current operating data based on the initial operating data corresponding to the multiple moments.
[0019] The initial operating data corresponding to multiple time moments can reflect the changes in the operating characteristics of the vehicle battery in the recent period. Subsequent vehicle battery fault identification based on the battery fault identification model based on the data can provide a more comprehensive understanding of the operating status of the vehicle battery, which helps to more accurately and timely predict possible vehicle battery failures.
[0020] Optionally, obtaining the current operating data based on the initial operating data corresponding to the multiple moments respectively includes: identifying abnormal data in the initial operating data corresponding to the multiple moments respectively, wherein the abnormal data includes at least one of the following: missing values, abnormal values, error codes; processing the abnormal data to obtain the current operating data, wherein the processing includes at least one of the following: missing value supplementation, abnormal value replacement, error code replacement.
[0021] By processing missing values of the initial operating data collected at each moment, the impact of missing data on the vehicle battery abnormality identification results can be reduced, thereby improving the accuracy and reliability of vehicle battery abnormality identification; by replacing or correcting abnormal values, the interference of abnormal values on vehicle battery abnormality identification can be reduced, thereby ensuring the stability and accuracy of vehicle battery abnormality identification; by correcting or replacing error codes, interference with vehicle battery abnormality identification due to other reasons can be eliminated.
[0022] According to another aspect of an embodiment of the present invention, a vehicle battery fault identification device is also provided, including: a data acquisition module, used to obtain current operating data of a vehicle battery in a current sampling period; a data denoising module, used to use a filtering layer in a battery fault identification model to perform denoising on the current operating data to obtain denoised data, wherein the battery fault identification model includes the filtering layer and a long short-term memory network layer; a fault identification module, used to obtain a current fault identification result of the vehicle battery based on the denoised data and using the long short-term memory network layer, wherein the long short-term memory network layer is used to extract temporal dependencies in the data; wherein the battery fault identification model is obtained through machine learning based on multiple groups of historical operating data of the vehicle battery in historical time periods, and corresponding historical fault identification results.
[0023] According to another aspect of an embodiment of the present invention, an electronic device is also provided, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any one of the vehicle battery fault identification methods.
[0024] In an embodiment of the present invention, current operating data of a vehicle battery in a current sampling period is obtained; a filter layer in a battery fault identification model is used to perform noise reduction processing on the current operating data to obtain noise reduction data, wherein the battery fault identification model includes the filter layer and a long short-term memory network layer; based on the noise reduction data, the long short-term memory network layer is used to obtain a current fault identification result of the vehicle battery, wherein the long short-term memory network layer is used to extract temporal dependencies in the data; wherein the battery fault identification model is based on multiple groups of historical operating data of the vehicle battery in historical time periods, and corresponding historical fault identification results, obtained through machine learning, thereby achieving the purpose of setting a battery fault identification model including a filter layer and a long short-term memory network layer to perform vehicle battery fault identification, thereby achieving the technical effect of improving the accuracy of vehicle battery fault identification, and further solving the technical problem of inaccurate vehicle battery fault identification existing in the related art. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:
[0026] Figure 1 is a flow chart of a vehicle battery fault identification method according to an embodiment of the present invention;
[0027] Figure 2 is a flow chart of an optional vehicle battery fault identification method according to an embodiment of the present invention;
[0028] Figure 3 is a schematic diagram of a network structure of an optional battery fault identification model according to an embodiment of the present invention;
[0029] Figure 4 is a schematic diagram of a vehicle battery fault identification device according to an embodiment of the present invention. DETAILED DESCRIPTION
[0030] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.
[0031] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0032] Electric vehicles are of great significance in ensuring energy security and reducing environmental pollution. However, due to the complexity of the electrochemical system and the uncertainty of the operating environment, the power battery, as a core component, has become the main source of electric vehicle failure, which is also an important reason hindering the further popularization of vehicles. For a long time, fires, explosions and other accidents caused by battery failures have been reported by the media. In order to alleviate people's safety anxiety about electric vehicles and solve industry problems, in addition to the continuous improvement of battery manufacturing and packaging technology, safety monitoring during vehicle use is equally important. Related technologies have been devoted to the research of fault diagnosis and prediction technology of power battery systems, and some battery fault detection methods have been developed and applied. However, there are still many diagnostic strategies that can only realize the analysis of experimental or offline data. There are two main reasons: on the one hand, some advanced fault detection methods require a large amount of high-quality data to be calculated, which also puts higher requirements on hardware, but the current on-board battery management system cannot meet them; on the other hand, some fault detection methods have low fault identification accuracy, which is prone to problems such as failure to identify existing faults normally or false fault alarms. As we all know, false alarms of vehicles are very bad, which can easily cause safety anxiety to drivers, so the reliability of the results of each diagnostic method should be worth considering.
[0033] In response to the above problems, an embodiment of the present invention provides a method embodiment for identifying vehicle battery faults. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.
[0034] Figure 1 is a flow chart of a vehicle battery fault identification method according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:
[0035] Step S102, obtaining current operating data of the vehicle battery during the current sampling period;
[0036] Optionally, the vehicle battery is a battery of an electric vehicle, and the current operating data of the vehicle battery can be obtained through, but not limited to, a big data cloud platform. Specifically, during the operation of the vehicle, the vehicle will send battery-related data to the big data cloud platform through its own battery management system. By interacting with the big data cloud platform, the current operating data of the vehicle battery can be obtained in a timely manner without adding additional detection equipment.
[0037] In an optional embodiment, current operating data of a vehicle battery in a current sampling period is obtained, including: obtaining initial operating data corresponding to each of multiple moments in the current sampling period, wherein the initial operating data includes at least the voltage, current, state of charge, insulation value, and temperature of the vehicle battery at the corresponding moment; and obtaining current operating data based on the initial operating data corresponding to each of the multiple moments.
[0038] Optionally, the current operating data is obtained based on the initial operating data of the vehicle battery collected at multiple sampling moments in the current time period, that is, the current operating data is time series data; the initial operating data corresponding to the multiple sampling moments can reflect the changes in operating characteristics of the vehicle battery in the recent period, and subsequent vehicle battery fault identification based on this data can provide a more comprehensive understanding of the operating status of the vehicle battery, which helps to more accurately and timely predict possible faults of the vehicle battery.
[0039] In an optional embodiment, current operating data is obtained based on initial operating data corresponding to multiple moments, including: identifying abnormal data in the initial operating data corresponding to multiple moments, wherein the abnormal data includes at least one of the following: missing values, abnormal values, error codes; processing the abnormal data to obtain current operating data, wherein the processing includes at least one of the following: missing value supplementation, abnormal value replacement, error code replacement.
[0040] It can be understood that missing values may lead to inaccurate identification of vehicle battery faults and deviations in identification results. By processing missing values of the initial operating data collected at each moment, the impact of missing data on the vehicle battery abnormality identification results can be reduced, thereby improving the accuracy and reliability of vehicle battery abnormality identification. Abnormal values are used to indicate larger values that deviate from normal values. The abnormal values may be caused by equipment failure, sensor error, etc. By replacing or correcting the abnormal values, the interference of abnormal values on vehicle battery abnormality identification can be reduced, thereby ensuring the stability and accuracy of vehicle battery abnormality identification. Error codes may be caused by problems with other devices or systems. By correcting or replacing the error codes, interference with vehicle battery abnormality identification due to other reasons can be eliminated.
[0041] Optionally, battery-related data can be obtained from the battery big data platform as the initial operating data of the vehicle battery, including the original data such as the voltage, current, remaining power state (State of Charge, SOC), insulation value and temperature of the vehicle battery at the corresponding sampling time. For example, in order to solve the problem of data packet loss caused by factors such as unstable transmission signals, if a small part of the data is missing, the gap is filled by interpolation. If a large amount of data is missing, the corresponding problem data is removed and the duplicate data is directly deleted; secondly, outlier replacement operations are required for bit errors and invalid data; finally, the data is segmented and reconstructed from different dimensions according to actual needs.
[0042] Step S104, using the filter layer in the battery fault identification model to perform noise reduction processing on the current operation data to obtain noise reduction data, wherein the battery fault identification model includes a filter layer and a long short-term memory network layer;
[0043] It can be understood that the battery fault identification model includes at least a filtering layer and a long short-term memory network layer, wherein the filtering layer can perform feature extraction and noise reduction on the input current operating data to obtain noise-reduced data to improve data quality and reduce the impact of noise on subsequent analysis.
[0044] In an optional embodiment, the filtering layer in the battery fault identification model is used to perform noise reduction processing on the current operating data to obtain noise-reduced data, including: using a sliding mean filtering method in the filtering layer to perform noise reduction processing on the current operating data to obtain noise-reduced data.
[0045] It can be understood that the sliding mean filtering method can smooth the fluctuations and noise in the data, eliminate mutations and outliers in the data, and thus achieve data noise reduction. Specifically, the sliding mean filtering method can smooth the data by calculating the mean around the input data point, making the data more stable and continuous, which is conducive to subsequent data analysis and model building. At the same time, the sliding mean filtering can eliminate high-frequency noise in the data, reduce the interference of noise on data analysis and prediction results, and improve the quality and reliability of the data.
[0046] Optionally, a sliding mean filtering method is introduced to reduce the interference of noise signals, and its equation can be expressed as:
[0047]
[0048] Among them, y MAF[τ] represents the output result of the moving average filter, that is, the filter output value at time τ; τ represents the current time ψ represents the window size of the moving average filter, that is, the number of data points used to calculate the moving average x[τ-i] represents the input data value at time τ-i, that is, the original data value of the filter input; i represents the offset in the moving average window, from 0 to ψ-1; Represents the normalization coefficient used to calculate the moving average, which is the inverse of the window size and is used to average the data values within the window.
[0049] Step S106, based on the noise reduction data, using the long short-term memory network layer to obtain the current fault identification result of the vehicle battery, wherein the long short-term memory network layer is used to extract the temporal dependency in the data;
[0050] Optionally, the battery fault identification model is obtained through machine learning based on multiple sets of historical operation data of the vehicle battery in historical periods and corresponding historical fault identification results. During the training process, the parameters of the model can be adjusted according to the evaluation indicators.
[0051] Optionally, since the current operating data of the input vehicle battery is time series data, and the Long Short-Term Memory (LSTM) layer is the core of processing time series data and can capture long-term dependencies in the sequence, the network layer can control the flow of information through the forget gate, input gate, and output gate.
[0052] It should be noted that the long short-term memory network layer has strong memory and sequence modeling capabilities, and can make full use of the temporal characteristics of time series data to improve the accuracy and reliability of fault identification. In addition, the long short-term memory network layer can process continuous data streams in real time, monitor the operating status of the vehicle battery in a timely manner, discover potential fault problems, and promptly issue warnings and handle them, thereby improving the safety and stability of the vehicle.
[0053] In an optional embodiment, when the battery fault identification model also includes a multi-head attention layer, a long short-term memory network layer is used based on the denoised data to obtain the current fault identification result of the vehicle battery, including: based on the denoised data, a long short-term memory network layer is used to obtain time series data; based on the time series data, a multi-head attention layer is used to obtain the current fault identification result, wherein the multi-head attention layer is used to extract position dependencies in the input sequence.
[0054] Optionally, a multi-head attention layer can also be set in the battery fault identification model, and the multi-head attention layer is set after the long short-term memory network to process the output results of the long short-term memory network. Specifically, the denoised data is used as the input of the long short-term memory network layer, and the denoised data is processed by the long short-term memory network layer to capture the long-term dependency features in the corresponding input sequence (ie, denoised data); the time series data is further processed by the multi-head attention layer to capture the position dependency in the corresponding input sequence (ie, time series data) to predict the current fault identification results of the vehicle battery. Among them, the multi-head attention layer calculates multiple attention weights in parallel, so that the model can simultaneously pay attention to information at different positions in the input sequence. This is particularly useful for capturing complex dependencies.
[0055] It should be noted that the long short-term memory network layer can model time series data and capture long-term dependencies in the data, which is suitable for processing data with time series characteristics; the long short-term memory network layer can extract important features in the time series data, help identify the laws and patterns in the data, and improve the battery fault identification model's ability to understand and predict the data. The multi-head attention layer can help the model better focus on the dependencies between different positions in the input sequence when processing sequence data, and improve the representation and generalization capabilities of the battery fault identification model; the multi-head attention layer can effectively extract the position dependencies in the input sequence, help the battery fault identification model to more accurately understand the inherent structure and laws of the data, and improve the accuracy and effect of fault identification. In summary, the long short-term memory network layer is used to obtain time series data, and then the multi-head attention layer is used for fault identification, which can effectively extract the time series features and position dependencies in the data, help accurately identify faults and improve the model's prediction and generalization capabilities. The above operations will help improve the accuracy and reliability of vehicle battery fault identification.
[0056] In an optional embodiment, when the battery fault identification model also includes a fully connected layer, a multi-head attention layer is used based on the time series data to obtain the current fault identification result, including: based on the time series data, a multi-head attention layer is used to obtain the attention result; based on the attention result, a fully connected layer is used to obtain the current fault identification result, wherein the fully connected layer is used to map the data to the fault category space.
[0057] Optionally, a fully connected layer can also be set in the battery fault identification model. The fully connected layer can be set after the multi-head attention layer to map the output result of the multi-head attention layer to the fault category space, wherein the fault category space is used to indicate the fault state and / or fault type of the vehicle battery, and the fault state is faulty or not faulty.
[0058] It should be noted that the multi-head attention layer can help the battery fault recognition model better understand the dependencies between different positions in the input sequence, focus on the positions with important information, and improve the representation ability of the model; the multi-head attention layer can extract key information and features in the input data, help the battery fault recognition model better capture important patterns and laws in the data, and improve the accuracy and robustness of vehicle battery fault recognition. The fully connected layer is used to map the feature representation of the multi-head attention layer to the fault category space, associate the extracted features with the fault category, and make the final classification decision; the fully connected layer can map and classify the high-order features extracted by the battery fault recognition model with the fault category, and finally obtain the recognition result of the current fault. In summary, the use of multi-head attention layer and fully connected layer for fault recognition based on time series data can make full use of the time series features and position dependencies in the data, improve the representation ability and classification accuracy of the battery fault recognition model, help accurately identify faults and map data to the fault category space. These operations will help improve the performance and accuracy of vehicle battery fault recognition.
[0059] In an optional embodiment, when the battery fault identification model also includes a convolution layer and a pooling layer, a long short-term memory network layer is used based on the denoised data to obtain time series data, including: using a convolution layer to perform feature extraction processing on the denoised data to obtain feature data; using a pooling layer to perform dimensionality reduction processing on the feature data to obtain reduced dimensionality data; based on the reduced dimensionality data, using a long short-term memory network layer to obtain time series data.
[0060] Optionally, a convolution layer and a pooling layer may be set in the battery fault identification model, wherein the convolution layer is set after the filtering layer to extract local features in the noise reduction data (such as the fluctuation pattern of the voltage or current of the vehicle battery, etc.), which are crucial for the fault identification of the vehicle battery; the pooling layer takes the feature data output by the convolution layer as input, and is used to perform dimensionality reduction processing on the input feature data. The dimension of the input feature data is reduced by downsampling, and the obtained dimensionality reduction data retains important information and reduces redundancy.
[0061] It should be noted that the convolution layer can effectively extract local features and spatial structures in the data, helping the battery fault identification model to capture important feature information in the data; by performing feature extraction processing through the convolution layer, the original data can be converted into a higher-level feature representation, which helps to improve the battery fault identification model's ability to understand and characterize the data. The pooling layer can perform dimensionality reduction processing on the input feature data, reduce the dimension and parameter amount of the feature data, while retaining important feature information; by performing dimensionality reduction processing on the input feature data through the pooling layer, the computational complexity and memory consumption of the battery fault identification model can be reduced. Using the convolution layer for feature extraction, the pooling layer for dimensionality reduction processing, and then obtaining the time series data through the long short-term memory network layer, the spatial and time series features of the data can be fully utilized, improving the model's ability to understand and characterize the data while reducing the computational complexity and memory consumption of the battery fault identification model.
[0062] In an optional embodiment, when the battery fault identification model also includes a normalization layer, a fully connected layer is used based on the attention result to obtain the current fault identification result, including: using the normalization layer to normalize the attention result to obtain a normalized result; using the fully connected layer to map the normalized result to the fault category space to obtain the current fault identification result.
[0063] Optionally, a normalization layer can also be set in the battery fault identification model. The normalization layer is set between the multi-head attention layer and the fully connected layer. The normalization layer uses the attention result output by the multi-head attention layer as input, normalizes the attention result, and uses the normalized result as input to the fully connected layer. It should be noted that the normalization layer can standardize the data so that the data has a similar scale and range. By setting the normalization of the data in the alienation layer between the multi-head attention layer and the fully connected layer, the dimensional influence between the data can be eliminated, which helps to improve the stability and generalization ability of the battery fault identification model, and ultimately achieve accurate identification of battery faults.
[0064] Optionally, the battery fault identification model may also include an input layer and an output layer, wherein the input layer serves as the first layer of the battery fault identification model and is used for data input; the output layer serves as the last layer of the battery fault identification model and is used for outputting vehicle battery fault identification results, wherein the output layer may apply an activation function (such as softmax for classification problems) to generate a probability distribution or other form of battery fault identification results as output.
[0065] Through the above steps S102 to S108, it is possible to set up a battery fault identification model including multiple network layers such as a filtering layer, a long short-term memory network layer, and a multi-head attention layer to perform vehicle battery fault identification, thereby achieving the technical effect of improving the accuracy of vehicle battery fault identification, and further solving the technical problem of inaccurate vehicle battery fault identification existing in related technologies.
[0066] Based on the above embodiments and optional embodiments, the present invention proposes an optional implementation mode: Figure 2 is a flow chart of an optional vehicle battery fault identification method according to an embodiment of the present invention, such as Figure 2 As shown, the method includes:
[0067] The method is implemented by a battery fault recognition model including at least a filter layer, a long short-term memory network LSTM layer, a multi-head attention layer and a fully connected layer, and the method includes:
[0068] S1: Obtain battery-related data during the current sampling period through the big data cloud platform, and perform data preprocessing to obtain the current operating data of the vehicle battery, including:
[0069] S11: Obtain the voltage, current, SOC, insulation value, temperature and other data uploaded by the vehicle battery management system (BMS) from the battery big data platform. The entire data acquisition time range is all data seven days before the fault time point;
[0070] S12: Perform data cleaning. For the data packet loss problem caused by factors such as unstable transmission signals, if a small part of the data is missing, the gaps are filled by interpolation. If a large amount of data is missing, the corresponding problem data is removed, and the duplicate data is directly deleted. Secondly, outlier replacement operations are required for bit errors and invalid data. Finally, the data is segmented and reconstructed from different dimensions according to actual needs, and the pre-processed data obtained is used as the current operating data x1, x2,…, x n .
[0071] S2: Perform noise reduction processing on the current operating data of the vehicle battery to obtain noise reduction data b1, b2, …, b n ;
[0072] Among them, a sliding mean filtering method is used to reduce the interference of noise signals, and the window size can be selected as 4.
[0073] S3: Input the denoised data into the long short-term memory network layer to obtain time series data.
[0074] S4: Input the time series data obtained in step S3 into the head attention layer for multi-head attention mechanism processing to obtain the attention result.
[0075] S5: Input the attention result into the fully connected layer, map it to the fault category space through the fully connected layer, and obtain the current fault recognition result;
[0076] Among them, the fault category space is used to indicate whether the vehicle battery has a fault. Specifically, the fault probability value of the current operating data of the vehicle battery can be obtained through the fully connected layer. By setting a threshold, if the fault probability value is higher than or equal to the threshold, the vehicle battery is considered to be faulty; if the fault probability value is lower than the threshold, the vehicle battery is considered to be normal.
[0077] Optional, Figure 3 is a schematic diagram of a network structure of an optional battery fault identification model according to an embodiment of the present invention, such as Figure 3 As shown, the battery fault identification model includes not only a filter layer, a long short-term memory network LSTM layer, a multi-head attention layer, and a fully connected layer, but also an input layer and an output layer, wherein the input layer is used as the first layer of the battery fault identification model for data input; the output layer is used as the last layer of the battery fault identification model for outputting the vehicle battery fault identification result, wherein the output layer can apply an activation function (such as a softmax function for classification problems) to generate a probability distribution or other forms of battery fault identification results as output.
[0078] In this embodiment, first, pre-processing is performed on the voltage, current, SOC, temperature, insulation resistance and other data in the original data of vehicle battery operation; secondly, noise reduction is performed on the processed data, and characteristic variables are extracted from the noise-reduced data; finally, the obtained data is processed using a long short-term memory network model and a multi-head attention mechanism to achieve the effect of battery fault identification.
[0079] In this embodiment, a vehicle battery fault identification device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the terms "module" and "device" can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.
[0080] According to an embodiment of the present invention, a device embodiment for implementing the above-mentioned vehicle battery fault identification method is also provided. Figure 4 is a schematic diagram of the structure of a vehicle battery fault identification device according to an embodiment of the present invention. Figure 4As shown, the above-mentioned vehicle battery fault identification device includes: a data acquisition module 400, a data noise reduction module 402, and a fault identification module 404, wherein:
[0081] The data acquisition module 400 is used to acquire the current operation data of the vehicle battery during the current sampling period;
[0082] The data denoising module 402 is connected to the data acquisition module 400 and is used to perform denoising processing on the current operation data by using the filter layer in the battery fault identification model to obtain denoised data, wherein the battery fault identification model includes the filter layer and the long short-term memory network layer;
[0083] A fault identification module 404, connected to the data denoising module 402, is used to obtain a current fault identification result of the vehicle battery based on the denoised data by using a long short-term memory network layer, wherein the long short-term memory network layer is used to extract the temporal dependency in the data;
[0084] Among them, the battery fault identification model is obtained through machine learning based on multiple sets of historical operation data of the vehicle battery in historical periods and the corresponding historical fault identification results.
[0085] In an embodiment of the present invention, a data acquisition module 400 is provided to obtain the current operating data of the vehicle battery in the current sampling period; a data denoising module 402 is connected to the data acquisition module 400, and is used to perform denoising processing on the current operating data using the filter layer in the battery fault identification model to obtain denoised data, wherein the battery fault identification model includes a filter layer and a long short-term memory network layer; a fault identification module 404 is connected to the data denoising module 402, and is used to obtain the current fault identification result of the vehicle battery based on the denoised data using the long short-term memory network layer, wherein the long short-term memory network layer is used to extract the temporal dependency in the data; wherein the battery fault identification model is based on multiple groups of historical operating data of the vehicle battery in the historical period, and the corresponding historical fault identification results, obtained through machine learning, thereby achieving the purpose of setting a battery fault identification model including a filter layer and a long short-term memory network layer to perform fault identification of the vehicle battery, thereby achieving the technical effect of improving the accuracy of vehicle battery fault identification, and thus solving the technical problem of inaccurate vehicle battery fault identification existing in the related art.
[0086] It should be noted that the above modules can be implemented by software or hardware. For example, for the latter, it can be implemented in the following ways: the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.
[0087] It should be noted that the above data acquisition module 400, data noise reduction module 402, and fault identification module 404 correspond to steps S102 to S106 in the embodiment, and the examples and application scenarios implemented by the above modules and the corresponding steps are the same, but are not limited to the contents disclosed in the above embodiments. It should be noted that the above modules can be run in a computer terminal as part of the device.
[0088] It should be noted that the optional or preferred implementation of this embodiment can refer to the relevant description in the embodiment, which will not be repeated here.
[0089] The above-mentioned vehicle battery fault identification device may also include a processor and a memory. The above-mentioned data acquisition module 400, data noise reduction module 402, fault identification module 404, etc. are all stored in the memory as program modules, and the processor executes the above-mentioned program modules stored in the memory to realize corresponding functions.
[0090] The processor includes a kernel, which retrieves the corresponding program module from the memory. The kernel may be one or more. The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. The memory includes at least one memory chip.
[0091] According to an embodiment of the present application, an embodiment of a non-volatile storage medium is also provided. Optionally, in this embodiment, the non-volatile storage medium includes a stored program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute any of the vehicle battery fault identification methods.
[0092] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group, and the non-volatile storage medium includes a stored program.
[0093] Optionally, when the program is running, the device where the non-volatile storage medium is located is controlled to perform the following functions: obtain the current operating data of the vehicle battery in the current sampling period; use the filtering layer in the battery fault identification model to perform noise reduction processing on the current operating data to obtain noise-reduced data, wherein the battery fault identification model includes a filtering layer and a long short-term memory network layer; based on the noise-reduced data, a long short-term memory network layer is used to obtain the current fault identification result of the vehicle battery, wherein the long short-term memory network layer is used to extract the temporal dependencies in the data; wherein the battery fault identification model is based on multiple groups of historical operating data of the vehicle battery in historical time periods, and the corresponding historical fault identification results, and is obtained through machine learning.
[0094] According to an embodiment of the present application, an embodiment of a processor is also provided. Optionally, in this embodiment, the processor is used to run a program, wherein when the program is run, any one of the vehicle battery fault identification methods is executed.
[0095] According to an embodiment of the present application, an embodiment of a computer program product is also provided. When executed on a data processing device, it is suitable for executing a program that initializes any one of the above-mentioned vehicle battery fault identification method steps.
[0096] Optionally, the above-mentioned computer program product, when executed on a data processing device, is suitable for executing a program that initializes the following method steps: obtaining current operating data of a vehicle battery in a current sampling period; using a filter layer in a battery fault identification model to perform noise reduction processing on the current operating data to obtain noise-reduced data, wherein the battery fault identification model includes a filter layer and a long short-term memory network layer; based on the noise-reduced data, using a long short-term memory network layer to obtain a current fault identification result of the vehicle battery, wherein the long short-term memory network layer is used to extract temporal dependencies in the data; wherein the battery fault identification model is obtained through machine learning based on multiple sets of historical operating data of the vehicle battery in historical time periods, and the corresponding historical fault identification results.
[0097] An embodiment of the present invention provides an electronic device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, the following steps are implemented: obtaining current operating data of a vehicle battery in a current sampling period; using a filter layer in a battery fault identification model to perform noise reduction processing on the current operating data to obtain noise-reduced data, wherein the battery fault identification model includes a filter layer and a long short-term memory network layer; based on the noise reduction data, using a long short-term memory network layer to obtain a current fault identification result of the vehicle battery, wherein the long short-term memory network layer is used to extract temporal dependencies in the data; wherein the battery fault identification model is obtained through machine learning based on multiple groups of historical operating data of the vehicle battery in historical time periods, and corresponding historical fault identification results.
[0098] The above sequence of the embodiments of the present invention is for description only and does not represent the superiority or inferiority of the embodiments.
[0099] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0100] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the above modules can be a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, modules or indirect coupling or communication connection of modules, which can be electrical or other forms.
[0101] The modules described above as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment.
[0102] In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of software functional modules.
[0103] If the above-mentioned integrated module is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable non-volatile storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a non-volatile storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the various embodiments of the present invention. The aforementioned non-volatile storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.
[0104] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A vehicle battery fault identification method, characterized in that: include: Obtain the current operating data of the vehicle battery during the current sampling period; Using a filter layer in a battery fault identification model to perform noise reduction processing on the current operation data to obtain noise reduction data, wherein the battery fault identification model includes the filter layer and a long short-term memory network layer; Based on the denoised data, using the long short-term memory network layer, obtaining a current fault identification result of the vehicle battery, wherein the long short-term memory network layer is used to extract temporal dependencies in the data; Among them, the battery fault identification model is obtained through machine learning based on multiple groups of historical operation data of the vehicle battery in historical periods and corresponding historical fault identification results.
2. The method according to claim 1, characterized in that In the case where the battery fault identification model further includes a multi-head attention layer, the current fault identification result of the vehicle battery is obtained based on the noise reduction data using the long short-term memory network layer, including: Based on the denoised data, using the long short-term memory network layer to obtain time series data; Based on the time series data, the multi-head attention layer is used to obtain the current fault identification result, wherein the multi-head attention layer is used to extract position dependencies in the input sequence.
3. The method according to claim 2, characterized in that In the case where the battery fault identification model also includes a fully connected layer, the multi-head attention layer is used based on the time series data to obtain the current fault identification result, including: Based on the time series data, using the multi-head attention layer to obtain an attention result; Based on the attention result, the fully connected layer is used to obtain the current fault identification result, wherein the fully connected layer is used to map the data to the fault category space.
4. The method according to claim 2, characterized in that: In the case where the battery fault identification model further includes a convolution layer and a pooling layer, the denoised data is based on the long short-term memory network layer to obtain time series data, including: Using the convolutional layer to perform feature extraction processing on the denoised data to obtain feature data; Using a pooling layer to perform dimensionality reduction processing on the feature data to obtain dimensionality reduction data; Based on the dimension reduction data, the long short-term memory network layer is used to obtain the time series data.
5. The method according to claim 3, characterized in that: In the case where the battery fault identification model further includes a normalization layer, the current fault identification result is obtained by using the fully connected layer based on the attention result, including: The normalization layer is used to normalize the attention result to obtain a normalized result; The normalized result is mapped to the fault category space using the fully connected layer to obtain the current fault identification result.
6. The method according to claim 1, characterized in that The adopting the filter layer in the battery fault identification model to perform noise reduction processing on the current operation data to obtain noise reduction data includes: The sliding mean filtering method in the filtering layer is used to perform noise reduction processing on the current running data to obtain the noise reduction data.
7. The method according to any one of claims 1 to 6, characterized in that: The obtaining of current operating data of the vehicle battery during the current sampling period includes: Acquire initial operation data of the vehicle battery corresponding to a plurality of moments in the current sampling period, wherein the initial operation data at least includes the voltage, current, state of charge, insulation value, and temperature of the vehicle battery at the corresponding moments; The current operation data is obtained based on the initial operation data respectively corresponding to the multiple moments.
8. The method according to claim 7, characterized in that The obtaining the current operation data based on the initial operation data respectively corresponding to the multiple moments includes: Identifying abnormal data in the initial operation data corresponding to the multiple moments, wherein the abnormal data includes at least one of the following: a missing value, an abnormal value, and an error code; The abnormal data is processed to obtain the current running data, wherein the processing includes at least one of the following: missing value supplementation, abnormal value replacement, and error code replacement.
9. A vehicle battery fault identification device, characterized in that: include: A data acquisition module, used to acquire current operating data of the vehicle battery during the current sampling period; A data denoising module, configured to perform denoising processing on the current operating data by using a filter layer in a battery fault identification model to obtain denoised data, wherein the battery fault identification model includes the filter layer and a long short-term memory network layer; A fault identification module, configured to obtain a current fault identification result of the vehicle battery based on the noise reduction data and using the long short-term memory network layer, wherein the long short-term memory network layer is used to extract temporal dependencies in the data; Among them, the battery fault identification model is obtained through machine learning based on multiple groups of historical operation data of the vehicle battery in historical periods and corresponding historical fault identification results.
10. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the vehicle battery fault identification method described in any one of claims 1 to 8.
Citation Information
Patent Citations
New energy vehicle power battery fault diagnosis method and system based on big data
CN118182151A
Multi-fault diagnosis method and device for power battery system, medium and product
CN118393364A
Method and apparatus for predicting voltage inconsistency fault of battery cells, and server
WO2022174679A1
Cited By
Fault identification method and device for battery cluster and nonvolatile storage medium
CN121208691A
