Battery monitoring method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of battery monitoring, and discloses a battery monitoring method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining battery state signals, including voltage, current, temperature and internal resistance, of a power battery, generating a battery signal set based on the battery state signals, performing embedding representation on the battery signal set to obtain spatial embedding representation and time embedding representation, performing spatial feature extraction on the spatial embedding representation to obtain spatial dependency features, and performing time feature extraction according to the spatial dependency features and the time embedding representation to obtain time dependency features; performing feature fusion on the spatial dependency features and the time dependency features to obtain fusion features, and performing battery monitoring based on the fusion features; according to the method, the change of the battery state signal of the power battery in time and space dimensions is comprehensively considered, and the accuracy and reliability of battery monitoring are improved.

Description

Technical Field

[0001] The present application relates to the technical field of battery monitoring, and in particular to a battery monitoring method, device, electronic device, and storage medium. Background Art

[0002] In new energy vehicles, the power battery, as one of its core components, directly impacts the vehicle's range and overall performance. Power batteries not only fulfill the critical function of storing and delivering electrical energy but also contribute to the safety and economic efficiency of the entire vehicle. Therefore, strengthening the monitoring and research of power battery anomalies in new energy vehicles is particularly important. Effective anomaly monitoring can not only proactively prevent potential safety risks, but also effectively extend battery life and reduce maintenance costs, thereby further promoting the popularization and development of new energy vehicles.

[0003] However, power battery systems are complex, involving multiple battery cells connected in series and parallel. Their performance and safety are affected by numerous factors, such as temperature, current, and charge / discharge status. Existing anomaly monitoring methods cannot accurately detect power battery anomalies and are insufficient to comprehensively and quickly identify potential problems. Clearly, a new battery monitoring method is urgently needed to address at least one of these issues.

[0004] It should be noted that the above content only provides background technical information related to this application and does not necessarily constitute prior art. Summary of the Invention

[0005] In view of the shortcomings of the prior art described above, the present application provides a battery monitoring method, device, electronic device and storage medium, which comprehensively consider the changes in the battery status signal of the power battery in the time and space dimensions, and improve the accuracy and reliability of battery monitoring.

[0006] Other features and advantages of the present application will become apparent from the following detailed description, or may be learned in part by practice of the present application.

[0007] According to one aspect of an embodiment of the present application, a battery monitoring method is provided, including: obtaining a battery status signal of a power battery, wherein the battery status signal includes voltage, current, temperature, and internal resistance; generating a battery signal set based on the battery status signal, embedding the battery signal set to obtain a spatial embedding representation and a temporal embedding representation; performing spatial feature extraction on the spatial embedding representation to obtain a spatially dependent feature, performing temporal feature extraction based on the spatially dependent feature and the temporal embedding representation to obtain a temporally dependent feature; performing feature fusion on the spatially dependent feature and the temporally dependent feature to obtain a fused feature, and performing battery monitoring based on the fused feature.

[0008] In one embodiment of the present application, based on the aforementioned scheme, after obtaining the battery status signal of the power battery, the method further includes: classifying the battery status signal to obtain a classification result of the battery status signal; if the classification result is a high-frequency signal, denoising the battery status signal through Kalman filtering; if the classification result is a low-frequency signal, interpolating the battery status signal.

[0009] In one embodiment of the present application, based on the aforementioned scheme, the battery status signal is denoised by Kalman filtering, including: defining the state variables, state transfer equations and observation equations of the battery status signal, constructing a state space model, the preset parameters of the state space model including state variables, error covariance, process noise covariance and observation noise covariance; initializing the state variables, the error covariance, the process noise covariance and the observation noise covariance of the state space model; predicting the state variables at the current moment based on the state variables at the previous moment through the state space model, and predicting the error covariance at the current moment based on the error covariance at the previous moment and the process noise covariance; calculating based on the error covariance at the current moment and the observation noise covariance to obtain the Kalman gain of the state space model, and updating the state variables at the current moment and the error covariance at the current moment based on the Kalman gain to complete the denoising of the battery status signal.

[0010] In one embodiment of the present application, based on the aforementioned scheme, a battery signal set is generated based on the battery status signal, and the battery signal set is embedded to obtain a spatial embedding representation and a temporal embedding representation, including: aligning the battery status signal according to a time series and generating the battery signal set; slicing the battery signal set according to a preset time length to obtain a plurality of signal slices; performing node embedding on each of the signal slices to obtain an embedding vector of the node; based on the embedding vector of the node, calculating the spatial dependent edge embedding between nodes of different channels at the same time step, thereby obtaining the spatial embedding representation; based on the embedding vector of the node, calculating the temporal dependent edge embedding between nodes of different channels at adjacent time steps, thereby obtaining the temporal embedding representation.

[0011] In one embodiment of the present application, based on the aforementioned scheme, spatial features are extracted from the spatial embedding representation to obtain spatial dependency features, including: using binomial distribution as the prior spatial distribution, and determining the overall edge distribution between nodes based on the embedding vector of the node; deriving the prior distribution of the first signal slice from the historical signal through a linear transformation on the time series; eliminating the target parameter in the binomial distribution through a preset Gaussian distribution, and performing a posterior derivation on the prior distribution to obtain the posterior distribution of the second signal slice; extracting samples from the spatial dependency distribution of the second signal slice according to the posterior distribution to obtain the spatial dependency features.

[0012] In one embodiment of the present application, based on the aforementioned scheme, time feature extraction is performed according to the spatial dependency feature and the time embedding representation to obtain the time dependency feature, including: defining a Gaussian variable as the time dependency of the first signal slice and the second signal slice, predicting the distribution of the time embedding representation based on the Gaussian variable through the spatial dependency feature to obtain a conditional Gaussian distribution; sampling from the conditional Gaussian distribution to generate the time dependency feature of the first signal slice and the second signal slice.

[0013] In one embodiment of the present application, based on the aforementioned scheme, the spatially dependent features and the time-dependent features are fused to obtain fused features, and battery monitoring is performed based on the fused features, including: obtaining comprehensive features based on the spatially dependent features and the time-dependent features; inputting the comprehensive features into dual-graph convolution for fusion to obtain the fused features; and performing battery monitoring based on the fused features to obtain battery monitoring results.

[0014] According to one aspect of an embodiment of the present application, a battery monitoring device is provided, including: a data acquisition module for acquiring a battery status signal of a power battery, wherein the battery status signal includes voltage, current, temperature and internal resistance; an embedding representation module for generating a battery signal set based on the battery status signal, and embedding the battery signal set to obtain a spatial embedding representation and a temporal embedding representation; a feature extraction module for performing spatial feature extraction on the spatial embedding representation to obtain a spatially dependent feature, and performing temporal feature extraction based on the spatially dependent feature and the temporal embedding representation to obtain a temporally dependent feature; and a battery monitoring module for performing feature fusion on the spatially dependent feature and the temporally dependent feature to obtain a fused feature, and performing battery monitoring based on the fused feature.

[0015] According to one aspect of an embodiment of the present application, an electronic device is provided, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the electronic device implements a battery monitoring method as described in any one of the above embodiments.

[0016] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor of a computer, the computer is caused to execute the battery monitoring method as described in any one of the above embodiments.

[0017] Beneficial effects of the present application: The present application obtains a battery status signal of a power battery, wherein the battery status signal includes voltage, current, temperature and internal resistance, generates a battery signal set based on the battery status signal, embeds the battery signal set to obtain a spatial embedding representation and a temporal embedding representation, extracts spatial features from the spatial embedding representation to obtain spatially dependent features, extracts temporal features based on the spatially dependent features and the temporal embedding representation to obtain temporally dependent features, fuses the spatially dependent features and the time-dependent features to obtain fused features, and performs battery monitoring based on the fused features; the above method comprehensively considers the changes in the battery status signal of the power battery in the temporal and spatial dimensions, thereby improving the accuracy and reliability of battery monitoring.

[0018] In addition, by performing classification processing on the battery status signal of the power battery and data preprocessing including denoising processing or interpolation processing, the consistency and accuracy of the signal are ensured.

[0019] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings are incorporated into and constitute a part of the specification, illustrating embodiments consistent with the present application and, together with the specification, serving to explain the principles of the present application. It is obvious that the drawings described below are merely some embodiments of the present application, and a person of ordinary skill in the art can derive other drawings based on these drawings without inventive effort. In the drawings: Figure 1 is a schematic diagram of an exemplary system architecture shown in an exemplary embodiment of the present application; Figure 2 is a flowchart of a battery monitoring method shown in an exemplary embodiment of the present application; Figure 3 is a flow chart of a battery monitoring method according to another exemplary embodiment of the present application; Figure 4is a schematic diagram of a battery monitoring method according to an exemplary embodiment of the present application; Figure 5 1 is a schematic diagram of a feature extraction module framework of a battery monitoring method according to an exemplary embodiment of the present application; Figure 6 is a block diagram of a battery monitoring device shown in an exemplary embodiment of the present application; Figure 7 A schematic diagram of the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0021] The following will describe the embodiments of the present application with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand the other advantages and effects of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be understood that the preferred embodiments are only for the purpose of illustrating the present application and are not intended to limit the scope of protection of the present application.

[0022] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present application. Therefore, the illustrations only show components related to the present application and are not drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component can be changed at will, and the component layout type may also be more complicated.

[0023] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application. However, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present application difficult to understand.

[0024] As global environmental awareness continues to grow, new energy vehicles (NEVs) are becoming a key direction for future transportation development due to their exceptional performance in reducing greenhouse gas emissions and lowering reliance on fossil fuels. By using electric power, NEVs can effectively reduce air pollutant emissions, thereby contributing positively to improving urban air quality and addressing climate change. Furthermore, the promotion of NEVs will help promote energy transition and boost the utilization of clean energy.

[0025] Prior knowledge refers to knowledge that comes before experience. While the use of prior knowledge has been proven effective in multiple fields, there is currently little research on its application in abnormality monitoring of power batteries for new energy vehicles.

[0026] First of all, it should be noted that embedding representation is a technique that maps discrete or high-dimensional data into low-dimensional continuous vectors. Its core goal is to capture the semantic relationships and intrinsic characteristics between data. Reflecting semantic similarity through distance in vector space allows machines to process complex data more efficiently.

[0027] A time step is the time interval between each data point and its previous data point in time series data.

[0028] Figure 1 It is a schematic diagram of an exemplary system architecture shown in an exemplary embodiment of the present application.

[0029] Reference Figure 1 As shown, the system architecture may include a data acquisition device 101 and a computer device 102. The computer device 102 may be at least one of a desktop graphics processing unit (GPU) computer, a GPU computing cluster, a neural network computer, and the like. The data acquisition device 101 is used to collect battery status signals of power batteries, wherein the battery status signals include but are not limited to voltage, current, temperature, and internal resistance. In this embodiment, the data acquisition device 101 obtains the above data and provides it to the computer device 102 for processing. Relevant technicians can use the computer device 102 to generate a battery signal set based on the battery status signal, embed the battery signal set to obtain a spatial embedding representation and a temporal embedding representation, extract spatial features from the spatial embedding representation to obtain spatially dependent features, extract temporal features based on the spatially dependent features and the temporal embedding representation to obtain time-dependent features, perform feature fusion on the spatially dependent features and the time-dependent features to obtain fused features, and perform battery monitoring based on the fused features. It should be noted that the data acquisition device 101 and the computer device 102 provided in this embodiment are only examples and should not limit the functions and scope of use of the embodiments of the present application.

[0030] It should be noted that the battery monitoring method provided in the embodiment of the present application is generally executed by the computer device 102 , and accordingly, the battery monitoring device is generally provided in the computer device 102 .

[0031] Figure 2 The flowchart of the battery monitoring method shown in an exemplary embodiment of the present application is as follows. The battery monitoring method can be executed by a computing and processing device. The computing and processing device can be Figure 1The computer device 102 shown in FIG. Figure 2 As shown, the battery monitoring method includes at least steps S210 to S240, which are described in detail as follows: In step S210 , a battery status signal of the power battery is acquired.

[0032] The battery status signal includes but is not limited to voltage, current, temperature and internal resistance.

[0033] In this embodiment, the signal data that can be collected in the power battery include voltage, current, temperature, vibration and pressure signal lights. Among them, the key signals related to the abnormal monitoring and analysis of the power battery include voltage, current, temperature and internal resistance. The voltage and current signals can reveal the charge and discharge status and efficiency of the battery. The temperature signal helps to detect abnormal thermal activity to prevent thermal runaway. The change of internal resistance may indicate the aging or damage of the battery. Comprehensive analysis of these signals can help identify abnormal conditions of the power battery, thereby ensuring the safety and performance of the battery. Therefore, signals such as voltage, current, temperature and internal resistance can be composed of a power battery status signal library.

[0034] In one embodiment of the present application, the process after obtaining the battery status signal of the power battery also includes: classifying the battery status signal to obtain a classification result of the battery status signal; if the classification result is a high-frequency signal, denoising the battery status signal through Kalman filtering; if the classification result is a low-frequency signal, interpolating the battery status signal.

[0035] In this embodiment, the battery status signal is classified, and a preprocessing method for the battery status signal is determined based on the classification result. If the classification result is a high-frequency signal, the battery status signal is denoised using a Kalman filter; if the classification result is a low-frequency signal, the battery status signal is interpolated. High-frequency signals include but are not limited to voltage and current, and low-frequency signals include but are not limited to temperature. For example, the voltage and current are denoised, while the temperature is interpolated.

[0036] In one embodiment of the present application, the process of denoising a battery status signal through Kalman filtering includes the following steps: defining the state variables, state transfer equations and observation equations of the battery status signal, constructing a state space model, and the preset parameters of the state space model include state variables, error covariance, process noise covariance and observation noise covariance; initializing the state variables, error covariance, process noise covariance and observation noise covariance of the state space model; predicting the state variables at the current moment based on the state variables at the previous moment through the state space model, and predicting the error covariance at the current moment based on the error covariance and process noise covariance at the previous moment; calculating based on the error covariance and observation noise covariance at the current moment to obtain the Kalman gain of the state space model, and updating the state variables at the current moment and the error covariance at the current moment based on the Kalman gain to complete the denoising of the battery status signal.

[0037] In some embodiments, the battery status signal of the power battery is first denoised to reduce noise interference in subsequent feature extraction. This denoising process primarily targets high-frequency signals, such as voltage and current. In this embodiment, Kalman filtering is primarily used for noise removal. The specific steps are described below: Set the state variable x k To represent the actual state of the battery (such as voltage or current), the dynamic model of the power battery system can be expressed as: Formula (1) Among them, x k is the state variable at time k, representing the real state of the battery at time k, F k is the state transfer matrix, x k-1 is the state variable at time k-1, B k is the control input matrix at time k, u k is the control input at time k, w k is the process noise at time k. Time k can be understood as the current time, and time k-1 as the previous time.

[0038] The observation model of the power battery system can be expressed as: Formula (2) Among them, Z k is the observed value of the state at time k, representing the battery observation state at time k, is the observation matrix, x k is the state variable at time k, is the observation noise at time k.

[0039] The dynamic model can also be understood as the state transfer equation of the state space model, and the observation model can be understood as the observation equation of the state space model, and the state space model can be established based on the defined state variables, state transfer equations and observation equations.

[0040] Then you need to initialize the state variables and the error covariance P0. Initialized state variables are typically set based on prior knowledge, while the error covariance represents the uncertainty about the initial state and can be preset to a value, such as a large constant. This can also be understood as setting the battery's initial state (including initial voltage and initial current) based on prior knowledge, and setting the initial covariance matrix P0. Furthermore, the noise covariance matrix is ​​initialized. This includes the process noise covariance Q, which reflects the uncertainty of the system model, and the observation noise covariance R, which reflects the uncertainty of the measurement equipment.

[0041] Then we can enter the denoising process of Kalman filtering. The state variables at the previous moment are used to predict the state variables at the current moment. The expression is as follows: Formula (3) in, is the predicted value of the current state, i.e., the state variable at the current moment, i.e., the predicted battery state at moment k. k is the state transition matrix, is the state variable of the previous moment, that is, the battery observation state at time k-1, B k is the control input matrix, u k is the control input at time k.

[0042] At the same time, the error covariance is updated by taking the error covariance P of the previous moment as k-1 Perform the transfer and add the process noise covariance Q k The error covariance at the current moment can be predicted , its specific expression is as follows: = Formula (4) in, is the error covariance of the prediction at the current moment, F k is the state transfer matrix, P k-1 is the error covariance at the previous moment, is the transposed state transfer matrix, Q k is the process noise covariance.

[0043] After completing the prediction for a moment, the state needs to be updated to use the measured data to correct the prediction result and obtain the optimal estimate. First, the Kalman gain needs to be calculated to measure the relationship between the predicted value and the observed value: Formula (5) in, Represents the Kalman gain, which mainly determines the relative weight of the predicted value and the observed value during the update process. is the error covariance of the current moment of the prediction, that is, the error covariance at moment k, is the observation matrix, is the transposed observation matrix, represents the observation noise covariance.

[0044] Based on the Kalman gain, the current state variable, that is, the current battery predicted state, can be updated: Formula (6) in, It represents the filtered denoised signal, i.e. the state variable at the current moment, i.e. the predicted battery state at the k moment after denoising. is the predicted battery state at time k, is the Kalman gain, Z k is the observed state of the battery at time k, is the observation matrix.

[0045] The error covariance at the current moment is then updated to reflect the new uncertainty: Formula (7) Among them, P k is the updated error covariance at the current moment, that is, the updated error covariance at the k moment. I represents the unit matrix, which is used to correct the prediction covariance matrix. is the Kalman gain, is the observation matrix, is the error covariance of the prediction at time k.

[0046] In some embodiments, the temperature can be interpolated by one or more of linear interpolation, polynomial interpolation, spline interpolation, Lagrange interpolation, Newton interpolation, and Kriging interpolation. This application does not limit this and should not impose any restrictions on the functions and scope of use of the embodiments of this application.

[0047] In one embodiment, interpolation can also be achieved through the following methods: The known temperature data is cleaned and organized to remove outliers and noise, ensuring an accurate correspondence between timestamps and temperature values. Depending on the selected interpolation method, a corresponding algorithm is developed or an existing mathematical library is used to implement the interpolation process. The known temperature data is used as input, and time points are generated at equal or uneven intervals (if the time intervals are uneven) based on the target number. The interpolated temperature values ​​corresponding to these time points are then calculated. The interpolation results are verified and evaluated to ensure that the interpolated temperature values ​​align with the actual power battery temperature variation trend. Methods such as cross-validation and residual analysis can be used to evaluate the accuracy and reliability of the interpolation results.

[0048] In step S220 , a battery signal set is generated based on the battery status signal, and an embedding representation is performed on the battery signal set to obtain a spatial embedding representation and a temporal embedding representation.

[0049] In one embodiment of the present application, a battery signal set is generated based on a battery status signal, and an embedding representation is performed on the battery signal set to obtain a spatial embedding representation and a temporal embedding representation. The process includes the following steps: aligning the battery status signals according to a time series and generating a battery signal set; slicing the battery signal set according to a preset time length to obtain a plurality of signal slices; performing node embedding on each signal slice to obtain an embedding vector of the node; based on the embedding vector of the node, calculating the spatial dependent edge embedding between nodes of different channels at the same time step, thereby obtaining a spatial embedding representation; based on the embedding vector of the node, calculating the temporal dependent edge embedding between nodes of different channels at adjacent time steps, thereby obtaining a temporal embedding representation.

[0050] In this embodiment, the current and voltage denoised signals after Kalman filtering, the interpolated temperature signal, and the power battery internal resistance signal are aligned in time series to form a battery signal set of the power battery. , where the number of channels C is the number of signals, corresponding to current, voltage, temperature, and internal resistance. In this embodiment, the number of channels is equal to 4, and the number of samples is N. For the battery signal set of the power battery Specifically, the battery signal set is sliced ​​according to the preset time length to obtain multiple signal slices. , embed the nodes inside it, and get , , all node embeddings constitute a node set ,Then the nodes are connected to generate edge embeddings, and spatially dependent edge embeddings and time-dependent edge embeddings are obtained. The spatially dependent edge embeddings are used as spatial embedding representations, and the time-dependent edge embeddings are used as temporal embedding representations.

[0051] Among them, spatially dependent edge embedding It can be expressed as: Formula (8) Time-dependent edge embedding It can be expressed as: Formula (9) in, is a linear layer, is the spatially dependent edge embedding, is the time-dependent edge embedding, is the node of channel i at time step t, is the node of channel j at time step t, is the node of channel j at time step t-1, i∈{1,2,3,4}, j∈{1,2,3,4}, and the channel corresponds to current, voltage, temperature and internal resistance.

[0052] In some embodiments, the processed current, voltage, temperature and internal resistance signals are aligned in time series to construct a battery signal set of the power battery. According to application requirements (such as time window), the battery signal set X is divided into multiple signal slices , each slice contains signal data within a fixed time length. For each slice Perform formatting to ensure that all signals have consistent data structures at the same time point.

[0053] In some embodiments, a battery signal set is obtained Afterwards, the battery signal set X is divided into multiple time windows, i.e., multiple signal slices, each of which has a length of T. Assuming that the sliding window step is S, the index range of the kth signal slice is: [k·S, k·S+T−1], where each signal slice contains C×T data. The signal value at each time step t is Corresponds to a node, where i represents the signal channel (i∈{1,2,3,4}, corresponding to current, voltage, temperature and internal resistance, where the battery status signal corresponding to the value of i can be set according to requirements, and this application does not limit this), and t represents the time step. Each node The feature of is the signal value of the channel at this time step. For each node in the signal slice , generate node embeddings through a fully connected layer (or linear layer) : =Linear( ), all nodes are embedded into a node set H={ }. Edge embedding is then performed by the following steps: spatial dependencies are defined based on the connections between different channels i and j at the same time step t. Spatially dependent edge embedding is calculated: , where f() is a linear layer with two node embeddings as input and , the output is the edge embedding. The temporal dependency is defined based on the connections between different channels i and j at adjacent time steps t and t−1. Calculate the temporal dependent edge embedding: .

[0054] In another embodiment, edge embeddings can be generated by applying a linear layer to node pairs (i, j) at the same time step t, and accumulating spatially dependent edge embeddings: Apply to the node pair (i, j) at adjacent time steps t and t−1 and accumulate to obtain the time-dependent edge embedding: .

[0055] In step S230, spatial feature extraction is performed on the spatial embedding representation to obtain spatial dependent features, and temporal feature extraction is performed based on the spatial dependent features and the temporal embedding representation to obtain temporal dependent features.

[0056] In one embodiment of the present application, spatial feature extraction is performed on the spatial embedding representation to obtain spatial dependency features, which includes the following steps: using the binomial distribution as the prior spatial distribution, and determining the overall edge distribution between nodes based on the embedding vector of the node; deriving the prior distribution of the first signal slice from the historical signal through a linear transformation on the time series; eliminating the target parameter in the binomial distribution through a preset Gaussian distribution, and performing a posterior derivation on the prior distribution to obtain the posterior distribution of the second signal slice; extracting samples from the spatial dependency distribution of the second signal slice according to the posterior distribution to obtain the spatial dependency features.

[0057] In some embodiments, after completing the embedding representation, the spatially dependent edge embedding The input is sent to the variational space encoder for spatial dependency feature extraction. Since the power battery pack is composed of hundreds of battery cells, if each cell is abstracted as a node, the spatial dependency relationship between nodes can be represented by edges. In order to model the overall edge distribution between nodes, the binomial distribution is used as the prior spatial distribution. The overall edge distribution is defined as: Formula (10) in, Represents node v i and v j The overall edge distribution between Represents node v i and v j The side relationship between represents the probability of each edge existing, B(·) is the probability mass function of the binomial distribution, v i and v jrepresents the node embedding, n is the number of trials of the binomial distribution (which can be understood as the number of potential edges). The parameters of the prior distribution are and n are usually embedded from nodes v through a series of linear transformations i and v j Derived in .

[0058] Assuming that the prior distribution is time-dependent (i.e., the spatial dependence distribution under different signal slices may be different), the prior distribution of the t-1th first signal slice can be expressed as: Formula (11) in, yes After a series of linear transformations, It is derived from historical signals through linear transformation on the time series.

[0059] After completing the signal slice node embedding and prior distribution definition, in order to further optimize the model, it is necessary to derive and verify the posterior distribution. is discrete, and direct optimization is difficult. Therefore, this embodiment uses Gaussian distribution to approximate the binomial distribution, thereby eliminating the parameters n and The discreteness of . In this embodiment, the Gaussian distribution is used to eliminate the parameters n and , where the parameters n and is the target parameter.

[0060] For a given Gaussian distribution ,in, is the mean of the Gaussian distribution, is the variance of the Gaussian distribution. The posterior distribution of the second signal slice can be defined as: Formula (12) in, is the posterior distribution, m represents the mean of the Gaussian approximation of the binomial distribution, and its value can be calculated by the following formula: Formula (13) Where m represents the mean of the Gaussian approximation of the binomial distribution, is the variance of the Gaussian distribution, To prevent computational explosion of approximate weights, their values ​​can be defined as: Formula (14) in, To prevent computational explosion of approximate weights, is the mean of the Gaussian distribution, is a hyperparameter, is a soft-add function.

[0061] Draw samples from the spatially dependent distribution of the current signal slice, the second signal slice: Formula (15) in, Indicates the spatial dependency of the current signal slice, i.e., the spatial dependency feature. is a random variable drawn from a standard normal distribution, and m represents the mean of the Gaussian approximation to the binomial distribution.

[0062] In one embodiment of the present application, time feature extraction is performed based on spatial dependency features and time embedding representation, and the process of obtaining the time dependency features includes the following steps: defining a Gaussian variable as the time dependency of the first signal slice and the second signal slice, predicting the distribution of the time embedding representation based on the Gaussian variable through the spatial dependency features to obtain a conditional Gaussian distribution; sampling from the conditional Gaussian distribution to generate the time dependency features of the first signal slice and the second signal slice.

[0063] In this embodiment, after completing the spatial dependency feature extraction, based on the spatial dependency feature and time-dependent edge embedding , temporal dependency feature extraction is performed in the Gaussian temporal graph encoder. The spatiotemporal dependency of a signal slice can be viewed as weighting each edge of the spatial dependency with a Gaussian variable. The Gaussian variable is defined as the temporal dependency between the first signal slice t and the second signal slice t-1. Assume that such a Gaussian variable is conditioned on the spatial dependency feature: Formula (16) in, is the time-dependent edge embedding of the t-th signal slice, is a spatially dependent feature, and are the mean and variance of the time dependence, which are related to the spatial dependence features related.

[0064] Then the conditional Gaussian transformation is performed. The signal slice is defined as a sample of the following conditional distribution: Formula (17) in, represents the time dependency between the t-th first signal slice and the t-1-th second signal slice, i.e., the time dependency feature. is a spatially dependent feature, and are the mean and variance of the time dependence, ℓ is a random sample from a standard normal distribution.

[0065] In step S240 , feature fusion is performed on the space-dependent features and the time-dependent features to obtain fused features, and battery monitoring is performed based on the fused features.

[0066] In one embodiment of the present application, a comprehensive feature is obtained based on the spatial dependence feature and the time dependence feature; the comprehensive feature is input into the dual-graph convolution for fusion to obtain a fusion feature; and battery monitoring is performed based on the fusion feature to obtain a battery monitoring result.

[0067] In this embodiment, a multilayer perceptron (MLP) acts as a classifier, performing nonlinear transformation and learning on the fused features, mapping them to a space capable of distinguishing between normal and abnormal conditions, thereby deriving the final anomaly detection classification result. The MLP utilizes nonlinear activation functions (such as ReLU, Sigmoid, and Tanh) to break the constraints of linear relationships and learn complex patterns in the data through backpropagation, thereby accurately distinguishing between normal and abnormal conditions. The MLP (Multi-Layer Perceptron) is a typical feedforward neural network. In anomaly detection tasks, its core role is to act as a classifier, performing deep learning and transformation on the data obtained through the feature extraction and fusion stages to achieve accurate classification between normal and abnormal conditions.

[0068] In some embodiments, when implementing the battery signal collection After extracting the spatial and temporal dependencies, we use dual-graph convolution to further extract features of the dependencies of each time slice to support the final abnormality detection. and time-dependent features , perform matrix multiplication on it to obtain a comprehensive feature representation Secondly, the comprehensive feature representation can be input into the dual-graph convolution as the initial feature matrix for deep feature mining and fusion. For a given graph convolution, its basic form can be expressed as: Formula (18) in, It is l +1 layer feature matrix, is the adjacency matrix The degree matrix of is the weight matrix of the graph convolutional layer, is the ReLU activation function.

[0069] After the convolution process of the two images is completed, the convolution results of the two images are fused. In this invention, additive fusion is used to fuse the features. can be expressed as: Formula (19) in, Indicates the l +1 layer of fusion features, Indicates the l +1 layer's first comprehensive feature, Indicates the l Second comprehensive feature of the +1 layer.

[0070] Finally, a multi-layer perceptron is used to map the features to the anomaly monitoring space to obtain the final classification results.

[0071] This application ensures the consistency and accuracy of the signal by classifying and preprocessing the power battery status signal. After using signal denoising and interpolation technology, the signal is reorganized into a standardized signal set, and then the spatial dependent features are extracted by the variational space encoder, and then the time dependent features are extracted in combination with the Gaussian time series encoder. This feature extraction method not only effectively extracts the spatial information from multiple signals, but also fully considers the information in the signal time series intersection, and finally uses dual-graph convolution for deep feature extraction and fusion, effectively combining spatial and temporal information. Finally, accurate abnormal situation judgment is performed through a multi-layer perceptron, which significantly improves the accuracy and reliability of power battery abnormality monitoring.

[0072] Figure 3 is a flow chart of a battery monitoring method according to another exemplary embodiment of the present application. In an exemplary embodiment, referring to Figure 3 As shown, the battery monitoring method includes at least the following steps: obtaining the battery status signal of the power battery to form a power battery status signal library; using the Kalman filter model to perform data preprocessing operations such as denoising on the battery status signal; introducing neural field theory to model the spatially dependent prior distribution, and then using an encoder to derive the posterior distribution from the prior distribution; using a Gaussian time series encoder to calculate the comprehensive time dependency between different battery status signals; using dual-graph convolution to obtain fused features from the spatial and temporal dependencies, and finally outputting the monitoring results. The specific execution of each step has been described in detail in the previous embodiment and will not be repeated here.

[0073] In one embodiment of the present application, the present application proposes a battery monitoring framework for new energy vehicle power batteries based on variational space and Gaussian time graph model. The framework mainly includes three parts: signal preprocessing, feature extraction module (FE Module) and classifier. The specific structure is as follows Figure 4 As shown, Figure 4This is a schematic diagram of the battery monitoring method framework shown in an exemplary embodiment of the present application. The feature extraction module includes modules such as node embedding (SE-space, SE-time), variational space encoder (VSE), Gaussian time graph encoder (GTGE) and dual graph fusion (DGCN). The specific structure is as follows Figure 5 As shown, Figure 5 This is a schematic diagram of the feature extraction module framework of the battery monitoring method shown in an exemplary embodiment of the present application. The main process is described as follows: the different signals in the power battery status signal library are classified into categories, and different data preprocessing, such as denoising and interpolation, is performed on the signals of different categories. The signal denoising algorithm is used to remove noise from high-frequency signals and interpolate low-frequency signals, so that each signal has the same number of sample points in the same slice. The signal is then reorganized into a battery signal set and embedded; the signal embedded representation is then input into the variational space encoder to extract spatial dependent features and derive a posterior for the prior distribution; then, based on the spatial dependent features and the time embedded representation, the time dependent features are obtained in the Gaussian time series encoder; then, the spatial dependent features and the time dependent features are fused using dual graph convolution to obtain fused features, which are then input into the multi-layer perceptron for the final abnormality monitoring results. The specific manner in which each module and unit performs operations has been described in detail in the aforementioned embodiments and will not be repeated here.

[0074] This application significantly improves the accuracy and reliability of abnormality monitoring by comprehensively considering the spatial and temporal dependencies of power battery signals and utilizing prior knowledge and cross-temporal relationships between signals. It can more comprehensively capture changes in battery status, overcome the limitations of traditional technologies, and provide a more effective solution for power battery abnormality monitoring.

[0075] In addition, this application combines the variational space and Gaussian time series graph models to simultaneously capture the spatial and temporal dependencies of the signal. This application ensures the consistency and accuracy of the data by classifying and preprocessing the battery status signal, and uses signal denoising and interpolation techniques to create a standardized signal set. The variational space encoder and Gaussian time series graph encoder are used to extract spatial dependency features and temporal dependency features respectively, and then deep feature fusion is achieved through dual-graph convolution. Finally, anomaly monitoring is performed through a multi-layer perceptron, which significantly improves the accuracy and reliability of anomaly monitoring and breaks through the limitations of traditional technologies.

[0076] Figure 6 This is a block diagram of a battery monitoring device shown in an exemplary embodiment of the present application. The device can be applied to Figure 1The implementation environment shown is specifically configured in the computer device 102. The apparatus may also be applicable to other exemplary implementation environments and specifically configured in other devices. This embodiment does not limit the implementation environment to which the apparatus is applicable.

[0077] like Figure 6 As shown, the exemplary battery monitoring device includes: a data acquisition module 610 , an embedding representation module 620 , a feature extraction module 630 and a battery monitoring module 640 .

[0078] Among them, the data acquisition module 610 is used to obtain the battery status signal of the power battery, wherein the battery status signal includes voltage, current, temperature and internal resistance; the embedding representation module 620 is used to generate a battery signal set based on the battery status signal, embed the battery signal set to obtain a spatial embedding representation and a temporal embedding representation; the feature extraction module 630 is used to extract spatial features from the spatial embedding representation to obtain spatially dependent features, and extract temporal features based on the spatially dependent features and the temporal embedding representation to obtain time-dependent features; the battery monitoring module 640 is used to fuse the spatially dependent features and the time-dependent features to obtain fused features, and perform battery monitoring based on the fused features.

[0079] It should be noted that the battery monitoring device provided in the above-described embodiments and the battery monitoring method provided in the above-described embodiments share the same concept. The specific manner in which the various modules and units perform their operations has been described in detail in the method embodiments and will not be repeated here. In actual applications, the battery monitoring device provided in the above-described embodiments may, as needed, allocate the aforementioned functions to different functional modules. This means, the internal structure of the device may be divided into different functional modules to perform all or part of the aforementioned functions, and this is not intended to be limiting herein.

[0080] An embodiment of the present application also provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs, which, when executed by one or more processors, enables the electronic device to implement the battery monitoring method provided in the above-mentioned embodiments.

[0081] Figure 7 The following is a schematic diagram showing the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application. Figure 7 The computer system 700 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0082] like Figure 7As shown, computer system 700 includes a central processing unit (CPU) 701, which can perform various appropriate actions and processes, such as executing the methods provided in the various embodiments described above, based on programs stored in read-only memory (ROM) 702 or programs loaded from storage 708 into random access memory (RAM) 703. RAM 703 also stores various programs and data required for system operation. CPU 701, ROM 702, and RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to bus 704.

[0083] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, mouse, and the like; an output section 707 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and speakers; a storage section 708 including devices such as a hard disk; and a communication section 709 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. Removable media 711, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 710 as needed, so that computer programs read from the media can be installed in the storage section 708 as needed.

[0084] In particular, according to embodiments of the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for executing the methods illustrated in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section 709 and / or installed from removable media 711. When executed by the central processing unit (CPU) 701, the computer program performs the various functions defined in the system of the present application.

[0085] It should be noted that the computer-readable medium described in the embodiments of this application may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may, for example, be an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable computer program. This propagated data signal may take a variety of forms, including, but not limited to, an electromagnetic signal, an optical signal, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. A computer program embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wired, or any suitable combination thereof.

[0086] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. Among them, each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0087] The units involved in the embodiments described in this application may be implemented by software or hardware, and the units described may also be set in a processor. In some cases, the names of these units do not constitute limitations on the units themselves.

[0088] Another aspect of the present application provides a computer-readable storage medium having a computer program stored thereon. When executed by a computer processor, the computer program causes the computer to perform the battery monitoring methods provided in the above embodiments. The computer-readable storage medium may be included in the electronic device described in the above embodiments, or may exist independently and not be incorporated into the electronic device.

[0089] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiment of the application, the features and functions of two or more modules or units described above can be concretized in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be concretized.

[0090] Another aspect of the present application provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the battery monitoring method provided in each of the above embodiments.

[0091] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the embodiments of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the embodiments of the present application.

[0092] Those skilled in the art will readily conceive of other embodiments of the present application after considering the specification and practicing the embodiments disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of this application and include common knowledge or customary techniques in the art that are not disclosed herein.

[0093] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Anyone skilled in the art may modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, any equivalent modifications or alterations accomplished by a person of ordinary skill in the art without departing from the spirit and technical concepts disclosed in this application shall be covered by the claims of this application.

Claims

1. A battery monitoring method, characterized in that: include: Acquire a battery status signal of the power battery, wherein the battery status signal includes voltage, current, temperature, and internal resistance; generating a battery signal set based on the battery status signal, and performing embedding representation on the battery signal set to obtain a spatial embedding representation and a temporal embedding representation; Performing spatial feature extraction on the spatial embedding representation to obtain spatial dependency features, and performing temporal feature extraction based on the spatial dependency features and the temporal embedding representation to obtain temporal dependency features; The spatially dependent features and the time-dependent features are fused to obtain fused features, and battery monitoring is performed based on the fused features.

2. The battery monitoring method according to claim 1, characterized in that: After obtaining the battery status signal of the power battery, the method further includes: classifying the battery status signal to obtain a classification result of the battery status signal; If the classification result is a high-frequency signal, the battery status signal is subjected to denoising processing by using Kalman filtering; if the classification result is a low-frequency signal, the battery status signal is subjected to interpolation processing.

3. The battery monitoring method according to claim 2, characterized in that: Performing denoising processing on the battery status signal by using a Kalman filter, including: Defining state variables, state transfer equations, and observation equations of the battery state signal to construct a state space model, wherein the preset parameters of the state space model include state variables, error covariance, process noise covariance, and observation noise covariance; Initializing the state variables of the state-space model, the error covariance, the process noise covariance, and the observation noise covariance; Predicting the state variables at the current moment based on the state variables at the previous moment using the state space model, and predicting the error covariance at the current moment based on the error covariance at the previous moment and the process noise covariance; A Kalman gain of the state-space model is obtained based on the error covariance at the current moment and the observation noise covariance, and the state variable at the current moment and the error covariance at the current moment are updated based on the Kalman gain to complete denoising of the battery status signal.

4. The battery monitoring method according to any one of claims 1 to 3, characterized in that: Generating a battery signal set based on the battery status signal, and embedding the battery signal set to obtain a spatial embedding representation and a temporal embedding representation, including: Aligning the battery status signals according to a time series and generating the battery signal set; Slicing the battery signal set according to a preset time length to obtain a plurality of signal slices; Performing node embedding on each of the signal slices to obtain an embedding vector of the node; Based on the embedding vectors of the nodes, calculating the spatial dependency edge embeddings between nodes of different channels at the same time step, thereby obtaining the spatial embedding representation; Based on the embedding vectors of the nodes, time-dependent edge embeddings between nodes of different channels at adjacent time steps are calculated to obtain the time embedding representation.

5. The battery monitoring method according to claim 4, characterized in that: Performing spatial feature extraction on the spatial embedding representation to obtain spatial dependency features includes: Using binomial distribution as a prior spatial distribution, determining the overall edge distribution between nodes based on the embedding vectors of the nodes; The prior distribution of the first signal slice is derived from the historical signal through a linear transformation on the time series; Eliminating the target parameter in the binomial distribution through a preset Gaussian distribution, and performing a posterior deduction on the prior distribution to obtain a posterior distribution of the second signal slice; Samples are extracted from the spatial dependency distribution of the second signal slice according to the posterior distribution to obtain the spatial dependency feature.

6. The battery monitoring method according to claim 5, characterized in that: Performing temporal feature extraction based on the spatial dependency feature and the temporal embedding representation to obtain a temporal dependency feature includes: defining a Gaussian variable as the time dependency of the first signal slice and the second signal slice, and predicting the distribution of the time embedding representation using the spatial dependency feature based on the Gaussian variable to obtain a conditional Gaussian distribution; Sampling from the conditional Gaussian distribution generates time-dependent features of the first signal slice and the second signal slice.

7. The battery monitoring method according to any one of claims 1 to 3, characterized in that: The spatial dependency feature and the time dependency feature are fused using dual-graph convolution to obtain a fused feature, and battery monitoring is performed based on the fused feature, including: Obtaining a comprehensive feature according to the spatially dependent feature and the temporally dependent feature; Inputting the comprehensive features into the dual-graph convolution for fusion to obtain the fused features; Battery monitoring is performed based on the fusion features to obtain a battery monitoring result.

8. A battery monitoring device, characterized in that: include: A data acquisition module is used to obtain a battery status signal of the power battery, wherein the battery status signal includes voltage, current, temperature and internal resistance; an embedding representation module, configured to generate a battery signal set based on the battery status signal, and perform embedding representation on the battery signal set to obtain a spatial embedding representation and a temporal embedding representation; A feature extraction module is used to extract spatial features from the spatial embedding representation to obtain spatially dependent features, and to extract temporal features based on the spatially dependent features and the temporal embedding representation to obtain temporally dependent features; The battery monitoring module is configured to perform feature fusion on the spatially dependent features and the time-dependent features to obtain fused features, and perform battery monitoring based on the fused features.

9. An electronic device, characterized in that: The electronic device comprises: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the electronic device to implement the battery monitoring method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed by a processor of a computer, the computer is caused to execute the battery monitoring method according to any one of claims 1 to 7.

Images