Battery health state estimation method and device, computer equipment, readable storage medium and program product

Through the multi-attention network, the charging and discharging data of lithium-ion batteries is processed and attention output vectors are generated, which solves the problem of insufficient information extraction and model generalization capabilities of traditional methods, and realizes an accurate assessment of the health status of lithium batteries.

CN119024212APending Publication Date: 2024-11-26SHENZHEN POWER SUPPLY BUREAU
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411339609.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-25
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

When traditional data-driven methods process complex and multi-dimensional operation data of lithium batteries, there are problems such as insufficient information extraction and insufficient generalization capabilities of model, which is difficult to fully reflect the health status of the battery.

Method used

Using multi-attention network combined with deep learning technology, pre-processing by obtaining the charging and discharging data of lithium-ion batteries, and inputting feature extraction networks and multi-attention modules to generate first and second attention output vectors for determining the battery. Health status.

Benefits of technology

By fully extracting charge and discharge information with more attention, the health status of the battery can be accurately evaluated, the gradient disappearance problem in traditional networks can be alleviated, and the global dependence and local connection of features can be effectively established, which improves the accuracy of health status estimation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119024212A_ABST
    Figure CN119024212A_ABST
Patent Text Reader

Abstract

The invention relates to a battery health state estimation method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring charging and discharging data of a lithium ion battery in a current charging and discharging process, preprocessing the charging and discharging data, and acquiring an output vector in a previous charging and discharging process as a previous hidden state; inputting the previous hidden state and the preprocessed charging and discharging data into a feature extraction network to obtain an output vector of the current charging and discharging process; inputting the output vector of the current charging and discharging process into a first attention module and a second attention module of the multi-attention network to obtain a first attention output vector and a second attention output vector; and determining the health state of the battery based on the first attention output vector and the second attention output vector. By adopting the method, the charging and discharging information is fully extracted through multiple attention, and the health state of the battery can be accurately evaluated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of battery technology, and in particular to a battery health status estimation method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art

[0002] With the rapid development of electric vehicles, energy storage devices and other applications, lithium-ion batteries, as the core technology for efficient energy storage, have become one of the key components of various power equipment. However, the state of health (SOH) of lithium-ion batteries directly affects their performance, safety and service life. Therefore, accurate assessment of battery health status is of great significance to ensure safe operation of batteries, optimize charging and discharging strategies, and extend battery life.

[0003] Data-driven methods use a large amount of operating data to estimate the battery health status through machine learning, deep learning and other technologies. This type of method has the advantages of not relying on physical models and strong real-time performance. However, when dealing with complex and multi-dimensional operating data of lithium batteries, traditional data-driven models often have problems such as insufficient information extraction and insufficient model generalization ability, making it difficult to fully reflect the health status of the battery. Summary of the invention

[0004] Based on this, it is necessary to provide a battery health state estimation method, device, computer equipment, computer readable storage medium and computer program product that can accurately evaluate the battery health state in response to the above technical problems.

[0005] In a first aspect, the present application provides a battery health state estimation method, comprising:

[0006] Acquire charge and discharge data of the lithium-ion battery in the current charge and discharge process, preprocess the charge and discharge data to obtain the preprocessed charge and discharge data, and acquire the output vector in the previous charge and discharge process as the previous hidden state;

[0007] Inputting the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network;

[0008] Input the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain a first attention output vector; input the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector;

[0009] A state of health of a battery is determined based on the first attention output vector and the second attention output vector.

[0010] In one embodiment, the acquiring charging and discharging data of the lithium-ion battery during the current charging and discharging process, and preprocessing the charging and discharging data to obtain the preprocessed charging and discharging data, includes:

[0011] The average voltage, average current and current cycle number in the current charge and discharge process are obtained; the average voltage, average current and current cycle number are constructed as a feature vector; the feature vector is standardized to obtain a standardized feature vector; and the standardized feature vector is used as the preprocessed charge and discharge data.

[0012] In one embodiment, the inputting the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process includes:

[0013] Based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the update gate, the output of the update gate is determined; based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the reset gate, the output of the reset gate is determined; based on the output of the reset gate and the parameters of the candidate hidden state, the candidate hidden state is determined; based on the previous hidden state, the output of the update gate and the candidate hidden state, the current hidden state is determined, and the current hidden state is used as the output vector of the current charge and discharge process.

[0014] In one embodiment, the step of inputting the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain the first attention output vector comprises:

[0015] Based on the output vector, query vector and key vector of the current charging and discharging process, determine the attention score of the first head and the attention score of the second head; based on the attention score of the first head, the attention score of the second head and the value vector, determine the output vector of the first head and the output vector of the second head; based on the output vector of the first head, the output vector of the second head and the output vector of the current charging and discharging process, determine the first attention output vector.

[0016] In one embodiment, inputting the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain the second attention output vector includes:

[0017] Based on the output vector of the current charging and discharging process and the convolution kernel, an attention output vector is determined; based on the output vector of the current charging and discharging process and the attention output vector, a second attention output vector is determined.

[0018] In one embodiment, the method further comprises:

[0019] The charging and discharging data and battery capacity data of the lithium-ion battery are acquired and processed to form a charging and discharging data set; a lithium-ion battery health state estimation network is constructed based on a multi-attention mechanism and a deep learning network; and the lithium-ion battery health state estimation network is trained based on the charging and discharging data set to obtain a trained lithium-ion battery health state estimation network.

[0020] In a second aspect, the present application also provides a battery health status estimation device, comprising:

[0021] An acquisition module is used to acquire the charge and discharge data of the lithium-ion battery in the current charge and discharge process, preprocess the charge and discharge data to obtain the preprocessed charge and discharge data, and acquire the output vector of the previous charge and discharge process as the previous hidden state;

[0022] An extraction module is used to input the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of a current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network;

[0023] An attention module, configured to input the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain a first attention output vector; and input the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector;

[0024] An evaluation module is used to determine a health state of a battery based on the first attention output vector and the second attention output vector.

[0025] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0026] Acquire charge and discharge data of the lithium-ion battery in the current charge and discharge process, preprocess the charge and discharge data to obtain the preprocessed charge and discharge data, and acquire the output vector in the previous charge and discharge process as the previous hidden state;

[0027] Inputting the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network;

[0028] Input the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain a first attention output vector; input the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector;

[0029] A state of health of a battery is determined based on the first attention output vector and the second attention output vector.

[0030] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0031] Acquire charge and discharge data of the lithium-ion battery in the current charge and discharge process, preprocess the charge and discharge data to obtain the preprocessed charge and discharge data, and acquire the output vector in the previous charge and discharge process as the previous hidden state;

[0032] Inputting the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network;

[0033] Input the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain a first attention output vector; input the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector;

[0034] A state of health of a battery is determined based on the first attention output vector and the second attention output vector.

[0035] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0036] Acquire charge and discharge data of the lithium-ion battery in the current charge and discharge process, preprocess the charge and discharge data to obtain the preprocessed charge and discharge data, and acquire the output vector in the previous charge and discharge process as the previous hidden state;

[0037] Inputting the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network;

[0038] Input the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain a first attention output vector; input the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector;

[0039] A state of health of a battery is determined based on the first attention output vector and the second attention output vector.

[0040] The above-mentioned battery health state estimation method, device, computer equipment, computer-readable storage medium and computer program product first obtain the charge and discharge data of the lithium-ion battery in the current charge and discharge process, pre-process the charge and discharge data to obtain the pre-processed charge and discharge data, obtain the output vector of the previous charge and discharge process as the previous hidden state; then input the previous hidden state and the pre-processed charge and discharge data into the feature extraction network in the lithium-ion battery health state estimation network to obtain the output vector of the current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network; then input the output vector of the current charge and discharge process into the first attention module of the multi-attention network to obtain the first attention output vector; input the output vector of the current charge and discharge process into the second attention module of the multi-attention network to obtain the second attention output vector; finally, based on the first attention output vector and the second attention output vector, determine the health state of the battery. By fully extracting the charge and discharge information through multi-attention, the health state of the battery can be accurately evaluated. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 is a flowchart of a method for estimating a battery health state in one embodiment;

[0043] Figure 2 A training flow chart of a battery health state estimation network in one embodiment;

[0044] Figure 3 is a structural block diagram of a battery health status estimation device in one embodiment;

[0045] Figure 4 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0047] In one embodiment, Figure 1 As shown, a method for estimating the health status of a battery is provided. This embodiment takes the method applied to a terminal as an example. It can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0048] Step 102, obtaining charging and discharging data of the lithium-ion battery in the current charging and discharging process, preprocessing the charging and discharging data to obtain preprocessed charging and discharging data, and obtaining an output vector in the previous charging and discharging process as the previous hidden state.

[0049] Optionally, the charge and discharge data may be average voltage, average current and number of charge and discharge cycles.

[0050] Exemplarily, the average voltage, average current and number of charge and discharge cycles of the lithium-ion battery in the current charge and discharge process are obtained, the average voltage, average current and number of charge and discharge cycles are preprocessed to obtain preprocessed charge and discharge data, and the output vector of the previous charge and discharge process is obtained as the previous hidden state.

[0051] Step 104, inputting the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network.

[0052] Optionally, the feature extraction network may be a Gated Recurrent Unit (GRU) network.

[0053] Exemplarily, the previous hidden state and the preprocessed charge and discharge data are input into a GRU network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network.

[0054] Step 106: input the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain a first attention output vector; input the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector.

[0055] Optionally, the first attention module is a global attention module, and the second attention module is a local attention module.

[0056] Exemplarily, the output vector of the current charging and discharging process is input into the global attention module of the multi-attention network to obtain a global attention output vector; the output vector of the current charging and discharging process is input into the local attention module of the multi-attention network to obtain a local attention output vector.

[0057] Step 108: Determine a battery health state based on the first attention output vector and the second attention output vector.

[0058] Exemplarily, the first attention output vector and the second attention output vector are added element by element, and the health state of the battery is determined based on the addition result.

[0059] The above-mentioned battery health state estimation method, device, computer equipment, computer-readable storage medium and computer program product first obtain the charge and discharge data of the lithium-ion battery in the current charge and discharge process, pre-process the charge and discharge data to obtain the pre-processed charge and discharge data, obtain the output vector of the previous charge and discharge process as the previous hidden state; then input the previous hidden state and the pre-processed charge and discharge data into the feature extraction network in the lithium-ion battery health state estimation network to obtain the output vector of the current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network; then input the output vector of the current charge and discharge process into the first attention module of the multi-attention network to obtain the first attention output vector; input the output vector of the current charge and discharge process into the second attention module of the multi-attention network to obtain the second attention output vector; finally, based on the first attention output vector and the second attention output vector, determine the health state of the battery. By fully extracting the charge and discharge information through multi-attention, the health state of the battery can be accurately evaluated.

[0060] In an exemplary embodiment, acquiring charging and discharging data of a lithium-ion battery during a current charging and discharging process, and preprocessing the charging and discharging data to obtain the preprocessed charging and discharging data includes:

[0061] Acquire charge and discharge data in the current charge and discharge process; construct the charge and discharge data into a feature vector; standardize the feature vector to obtain a standardized feature vector; and use the standardized feature vector as the preprocessed charge and discharge data.

[0062] Among them, the eigenvector is Z=[ , , N], where For the The average voltage of the constant current discharge process under the cycle, For the The average current of the constant current discharge process under cycles, N is the number of cycles.

[0063] Among them, standardizing the feature vector is to standardize the data in the feature vector into a feature vector with a mean value of 0 and a standard deviation of 1.

[0064] Exemplarily, the charge and discharge data in the current charge and discharge process is obtained; the charge and discharge data is constructed as a feature vector Z=[ , , N], where For the The average voltage of the constant current discharge process under the cycle, For the The average current of the constant current discharge process under cycles, N is the number of cycles; the characteristic vector is standardized to obtain a standardized characteristic vector; the standardized characteristic vector is used as the preprocessed charge and discharge data.

[0065] In this embodiment, by processing and standardizing the charge and discharge data, numerical problems in the acquired data can be prevented and feature comparison by the network can be facilitated.

[0066] In an exemplary embodiment, the step of inputting the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process includes:

[0067] Based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the update gate, the output of the update gate is determined; based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the reset gate, the output of the reset gate is determined; based on the output of the reset gate and the parameters of the candidate hidden state, the candidate hidden state is determined; based on the previous hidden state, the output of the update gate and the candidate hidden state, the current hidden state is determined, and the current hidden state is used as the output vector of the current charge and discharge process.

[0068] Among them, the previous hidden state is the output vector of the previous charging and discharging process.

[0069] Exemplarily, based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the update gate, the output of the update gate is determined by the formula:

[0070]

[0071] in, is the output of the update gate, is the weight matrix of the update gate, is the bias term of the update gate, is the pre-processed charge and discharge data, is the previous hidden state, The sigmoid activation function compresses the value between [0,1].

[0072] Based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the reset gate, the output of the reset gate is determined, and the formula is:

[0073]

[0074] in, is the output of the reset gate, is the weight matrix of the reset gate, is the bias term for the reset gate.

[0075] Based on the output of the reset gate and the parameters of the candidate hidden state, the candidate hidden state is determined, and the formula is:

[0076]

[0077] in, is the output of the candidate hidden state, is the weight matrix of the candidate hidden states, is the bias term of the candidate hidden state, represents element-wise multiplication, is the hyperbolic tangent activation function, which maps values ​​to [−1, 1].

[0078] Based on the previous hidden state, the output of the update gate and the candidate hidden state, the current hidden state is determined, and the formula is:

[0079]

[0080] In this formula, is the current hidden state, Indicates that the information of the previous hidden state is retained. Indicates the introduction of candidate hidden state information.

[0081] The above process can be expressed as:

[0082]

[0083] in is the output of GRU, that is, the current hidden state, and the output of GRU As the output vector of the current charging and discharging process.

[0084] In this embodiment, the features of the charge and discharge data are extracted through the GRU network, which can alleviate the gradient vanishing problem in the traditional network and can selectively retain or forget previous information.

[0085] In an exemplary embodiment, the output vector of the current charging and discharging process is input into the first attention module of the multi-attention network to obtain the first attention output vector.

[0086] Based on the output vector, query vector and key vector of the current charging and discharging process, determine the attention score of the first head and the attention score of the second head; based on the attention score of the first head, the attention score of the second head and the value vector, determine the output vector of the first head and the output vector of the second head; based on the output vector of the first head, the output vector of the second head and the output vector of the current charging and discharging process, determine the first attention output vector.

[0087] For example, the query vector of the i-th head is , key vector Sum value vector They are as follows:

[0088]

[0089]

[0090]

[0091] in, , , For the i-th head query vector, build the weight matrix of the vector and the value vector, is the output vector of the current charging and discharging process. Based on the output vector of the current charging and discharging process, the query vector and the key vector, the attention score of the i-th head is determined as follows:

[0092]

[0093] in, is the transpose of the key vector, is the scaling factor. The output of the i-th head is obtained based on the attention score of the i-th head for:

[0094]

[0095] Combine the outputs of i heads to get the output :

[0096]

[0097] in, is a weight matrix for linear transformation. In this embodiment, two heads are taken as an example, based on the output of the two heads , the output vector of the current charging and discharging process , residual operation and layer normalization operation, the resulting vector for:

[0098]

[0099]

[0100] in, For normalization operation, is the residual operation, This is the final first attention output vector.

[0101] In this embodiment, by using self-attention to strengthen features, the global dependency of features can be effectively established to capture important information in the global area.

[0102] In an exemplary embodiment, inputting the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain the second attention output vector includes:

[0103] Based on the output vector of the current charging and discharging process and the convolution kernel, an attention output vector is determined; based on the output vector of the current charging and discharging process and the attention output vector, a second attention output vector is determined.

[0104] Exemplarily, based on the output vector and convolution kernel of the current charging and discharging process, the attention output vector is determined according to the following formula:

[0105]

[0106] in, is the convolution kernel in the time series Current location on is the size of the convolution kernel, For the The weight of the convolution kernel, As an activation function, nonlinear characteristics are introduced. Based on the attention output vector , the output vector of the current charging and discharging process , layer normalization operation and residual operation to obtain the second attention output vector , the formula is as follows:

[0107] .

[0108] In this embodiment, by using local attention to strengthen the connection between local features, important features in local areas can be highlighted.

[0109] In an exemplary embodiment, Figure 2 As shown, the method also includes:

[0110] The charging and discharging data and battery capacity data of the lithium-ion battery are acquired and processed to form a charging and discharging data set; a lithium-ion battery health state estimation network is constructed based on a multi-attention mechanism and a deep learning network; and the lithium-ion battery health state estimation network is trained based on the charging and discharging data set to obtain a trained lithium-ion battery health state estimation network.

[0111] Exemplarily, the charge and discharge data and battery capacity data of the lithium-ion battery are obtained, and the charge and discharge data include: the average voltage of each charge and discharge cycle, the average current of each charge and discharge cycle, and the number of charge and discharge cycles; the battery capacity data include: the battery capacity data of each charge and discharge cycle and the maximum battery capacity data. The average voltage of each charge and discharge cycle, the average current of each charge and discharge cycle, and the number of charge and discharge cycles are used as input; the result of dividing the battery capacity data of each charge and discharge cycle by the maximum battery capacity data is used as the true value of the battery health state, and a data set is established, and divided into a training set and a test set in a ratio of 7:3. Based on global attention, local attention and GRU neural network, a lithium-ion battery health state estimation network is built. The lithium-ion battery health state estimation network is trained based on the training set and the test set to obtain a trained lithium-ion battery health state estimation network. A comparison model is selected for training and verification to obtain a test result; the test result of the trained lithium-ion battery health state estimation network is compared with the test result of the comparison model, and the comparison result of the battery health state (State of health, SOH) is output.

[0112] In this embodiment, by collecting data and creating a data set to train the network, the network can accurately measure the health status of the lithium-ion battery.

[0113] In an exemplary embodiment, the charge and discharge data in the current charge and discharge process is obtained; the charge and discharge data is constructed as a feature vector Z=[ , , N], where For the The average voltage of the constant current discharge process under the cycle, For the The average current of the constant current discharge process under cycles, N is the number of cycles; the feature vector is normalized to obtain a standardized feature vector; the standardized feature vector is used as the preprocessed charge and discharge data. The output vector of the previous charge and discharge process is obtained as the previous hidden state. Based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the update gate, the output of the update gate is determined, and the formula is:

[0114]

[0115] in, is the output of the update gate, is the weight matrix of the update gate, is the bias term of the update gate, is the pre-processed charge and discharge data, is the previous hidden state, is a sigmoid activation function that compresses the value between [0,1]. Based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the reset gate, the output of the reset gate is determined, and the formula is:

[0116]

[0117] in, is the output of the reset gate, is the weight matrix of the reset gate, is the bias term for the reset gate.

[0118] Based on the output of the reset gate and the parameters of the candidate hidden state, the candidate hidden state is determined, and the formula is:

[0119]

[0120] in, is the output of the candidate hidden state, is the weight matrix of the candidate hidden states, is the bias term of the candidate hidden state, represents element-wise multiplication, is a hyperbolic tangent activation function that maps values ​​to [−1, 1]. Based on the previous hidden state, the output of the update gate, and the candidate hidden state, the current hidden state is determined using the formula:

[0121]

[0122] In this formula, is the current hidden state, Indicates that the information of the previous hidden state is retained. Indicates the introduction of candidate hidden state information. The above process can be expressed as:

[0123]

[0124] in is the output of GRU, that is, the current hidden state, and the output of GRU As the output vector of the current charging and discharging process. The lithium-ion battery health status estimation network also includes a multi-attention network. In the global attention module of the multi-attention network, the query vector of the i-th head , key vector Sum value vector They are as follows:

[0125]

[0126]

[0127]

[0128] in, , , For the i-th head query vector, build the weight matrix of the vector and the value vector, is the output vector of the current charging and discharging process. Based on the output vector of the current charging and discharging process, the query vector and the key vector, the attention score of the i-th head is determined as follows:

[0129]

[0130] in, is the transpose of the key vector, is the scaling factor. The output of the i-th head is obtained based on the attention score of the i-th head for:

[0131]

[0132] Combine the outputs of i heads to get the output :

[0133]

[0134] in, is a weight matrix for linear transformation. In this embodiment, two heads are taken as an example, based on the output of the two heads , the output vector of the current charging and discharging process , residual operation and layer normalization operation, the resulting vector for:

[0135]

[0136]

[0137] in, For normalization operation, is the residual operation, That is the final first attention output vector. In the local attention module of the multi-attention network, the attention output vector is determined based on the output vector and convolution kernel of the current charging and discharging process. The formula is as follows:

[0138]

[0139] in, is the convolution kernel in the time series Current location on is the size of the convolution kernel, For the The weight of the convolution kernel, As an activation function, nonlinear characteristics are introduced. Based on the attention output vector , the output vector of the current charging and discharging process , layer normalization operation and residual operation to obtain the second attention output vector , the formula is as follows:

[0140] .

[0141] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0142] In an exemplary embodiment, Figure 3 As shown, a battery health state estimation device is provided, including: an acquisition module, an extraction module, an attention module and an evaluation module, wherein:

[0143] An acquisition module is used to acquire the charge and discharge data of the lithium-ion battery in the current charge and discharge process, preprocess the charge and discharge data to obtain the preprocessed charge and discharge data, and acquire the output vector of the previous charge and discharge process as the previous hidden state;

[0144] An extraction module is used to input the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of a current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network;

[0145] An attention module, configured to input the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain a first attention output vector; and input the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector;

[0146] An evaluation module is used to determine a health state of a battery based on the first attention output vector and the second attention output vector.

[0147] In one embodiment, the acquisition module is further used to:

[0148] Acquire charge and discharge data in the current charge and discharge process; construct the charge and discharge data into a feature vector; standardize the feature vector to obtain a standardized feature vector; and use the standardized feature vector as the preprocessed charge and discharge data.

[0149] In one embodiment, the extraction module is further used to:

[0150] Based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the update gate, the output of the update gate is determined; based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the reset gate, the output of the reset gate is determined; based on the output of the reset gate and the parameters of the candidate hidden state, the candidate hidden state is determined; based on the previous hidden state, the output of the update gate and the candidate hidden state, the current hidden state is determined, and the current hidden state is used as the output vector of the current charge and discharge process.

[0151] In one embodiment, the attention module is further configured to:

[0152] Based on the output vector, query vector and key vector of the current charging and discharging process, determine the attention score of the first head and the attention score of the second head; based on the attention score of the first head, the attention score of the second head and the value vector, determine the output vector of the first head and the output vector of the second head; based on the output vector of the first head, the output vector of the second head and the output vector of the current charging and discharging process, determine the first attention output vector.

[0153] In one embodiment, the attention module is further configured to:

[0154] Based on the output vector of the current charging and discharging process and the convolution kernel, an attention output vector is determined; based on the output vector of the current charging and discharging process and the attention output vector, a second attention output vector is determined.

[0155] In one embodiment, the battery health status estimation device is further used for:

[0156] The charging and discharging data and battery capacity data of the lithium-ion battery are acquired and processed to form a charging and discharging data set; a lithium-ion battery health state estimation network is constructed based on a multi-attention mechanism and a deep learning network; and the lithium-ion battery health state estimation network is trained based on the charging and discharging data set to obtain a trained lithium-ion battery health state estimation network.

[0157] Each module in the above-mentioned battery health status estimation device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.

[0158] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 4As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store hidden state data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a battery health state estimation method is implemented.

[0159] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0160] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0161] Acquire charge and discharge data of the lithium-ion battery in the current charge and discharge process, preprocess the charge and discharge data to obtain the preprocessed charge and discharge data, and acquire the output vector in the previous charge and discharge process as the previous hidden state;

[0162] Inputting the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network;

[0163] Input the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain a first attention output vector; input the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector;

[0164] A state of health of a battery is determined based on the first attention output vector and the second attention output vector.

[0165] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0166] The average voltage, average current and current cycle number in the current charge and discharge process are obtained; the average voltage, average current and current cycle number are constructed as a feature vector; the feature vector is standardized to obtain a standardized feature vector; and the standardized feature vector is used as the preprocessed charge and discharge data.

[0167] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0168] Based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the update gate, the output of the update gate is determined; based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the reset gate, the output of the reset gate is determined; based on the output of the reset gate and the parameters of the candidate hidden state, the candidate hidden state is determined; based on the previous hidden state, the output of the update gate and the candidate hidden state, the current hidden state is determined, and the current hidden state is used as the output vector of the current charge and discharge process.

[0169] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0170] Based on the output vector, query vector and key vector of the current charging and discharging process, determine the attention score of the first head and the attention score of the second head; based on the attention score of the first head, the attention score of the second head and the value vector, determine the output vector of the first head and the output vector of the second head; based on the output vector of the first head, the output vector of the second head and the output vector of the current charging and discharging process, determine the first attention output vector.

[0171] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0172] Based on the output vector of the current charging and discharging process and the convolution kernel, an attention output vector is determined; based on the output vector of the current charging and discharging process and the attention output vector, a second attention output vector is determined.

[0173] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0174] The charging and discharging data and battery capacity data of the lithium-ion battery are acquired and processed to form a charging and discharging data set; a lithium-ion battery health state estimation network is constructed based on a multi-attention mechanism and a deep learning network; and the lithium-ion battery health state estimation network is trained based on the charging and discharging data set to obtain a trained lithium-ion battery health state estimation network.

[0175] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0176] Acquire charge and discharge data of the lithium-ion battery in the current charge and discharge process, preprocess the charge and discharge data to obtain the preprocessed charge and discharge data, and acquire the output vector in the previous charge and discharge process as the previous hidden state;

[0177] Inputting the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network;

[0178] Input the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain a first attention output vector; input the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector;

[0179] A state of health of a battery is determined based on the first attention output vector and the second attention output vector.

[0180] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0181] The average voltage, average current and current cycle number in the current charge and discharge process are obtained; the average voltage, average current and current cycle number are constructed as a feature vector; the feature vector is standardized to obtain a standardized feature vector; and the standardized feature vector is used as the preprocessed charge and discharge data.

[0182] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0183] Based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the update gate, the output of the update gate is determined; based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the reset gate, the output of the reset gate is determined; based on the output of the reset gate and the parameters of the candidate hidden state, the candidate hidden state is determined; based on the previous hidden state, the output of the update gate and the candidate hidden state, the current hidden state is determined, and the current hidden state is used as the output vector of the current charge and discharge process.

[0184] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0185] Based on the output vector, query vector and key vector of the current charging and discharging process, determine the attention score of the first head and the attention score of the second head; based on the attention score of the first head, the attention score of the second head and the value vector, determine the output vector of the first head and the output vector of the second head; based on the output vector of the first head, the output vector of the second head and the output vector of the current charging and discharging process, determine the first attention output vector.

[0186] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0187] Based on the output vector of the current charging and discharging process and the convolution kernel, an attention output vector is determined; based on the output vector of the current charging and discharging process and the attention output vector, a second attention output vector is determined.

[0188] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0189] The charging and discharging data and battery capacity data of the lithium-ion battery are acquired and processed to form a charging and discharging data set; a lithium-ion battery health state estimation network is constructed based on a multi-attention mechanism and a deep learning network; and the lithium-ion battery health state estimation network is trained based on the charging and discharging data set to obtain a trained lithium-ion battery health state estimation network.

[0190] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0191] Acquire charge and discharge data of the lithium-ion battery in the current charge and discharge process, preprocess the charge and discharge data to obtain the preprocessed charge and discharge data, and acquire the output vector in the previous charge and discharge process as the previous hidden state;

[0192] Inputting the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network;

[0193] Input the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain a first attention output vector; input the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector;

[0194] A state of health of a battery is determined based on the first attention output vector and the second attention output vector.

[0195] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0196] The average voltage, average current and current cycle number in the current charge and discharge process are obtained; the average voltage, average current and current cycle number are constructed as a feature vector; the feature vector is standardized to obtain a standardized feature vector; and the standardized feature vector is used as the preprocessed charge and discharge data.

[0197] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0198] Based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the update gate, the output of the update gate is determined; based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the reset gate, the output of the reset gate is determined; based on the output of the reset gate and the parameters of the candidate hidden state, the candidate hidden state is determined; based on the previous hidden state, the output of the update gate and the candidate hidden state, the current hidden state is determined, and the current hidden state is used as the output vector of the current charge and discharge process.

[0199] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0200] Based on the output vector, query vector and key vector of the current charging and discharging process, determine the attention score of the first head and the attention score of the second head; based on the attention score of the first head, the attention score of the second head and the value vector, determine the output vector of the first head and the output vector of the second head; based on the output vector of the first head, the output vector of the second head and the output vector of the current charging and discharging process, determine the first attention output vector.

[0201] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0202] Based on the output vector of the current charging and discharging process and the convolution kernel, an attention output vector is determined; based on the output vector of the current charging and discharging process and the attention output vector, a second attention output vector is determined.

[0203] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0204] The charging and discharging data and battery capacity data of the lithium-ion battery are acquired and processed to form a charging and discharging data set; a lithium-ion battery health state estimation network is constructed based on a multi-attention mechanism and a deep learning network; and the lithium-ion battery health state estimation network is trained based on the charging and discharging data set to obtain a trained lithium-ion battery health state estimation network.

[0205] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0206] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0207] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for estimating a battery health state, characterized in that: The method comprises: Acquire charge and discharge data of the lithium-ion battery in the current charge and discharge process, preprocess the charge and discharge data to obtain the preprocessed charge and discharge data, and acquire the output vector in the previous charge and discharge process as the previous hidden state; Inputting the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network; Input the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain a first attention output vector; input the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector; A state of health of a battery is determined based on the first attention output vector and the second attention output vector.

2. The method according to claim 1, characterized in that: The acquiring charging and discharging data of the lithium-ion battery during the current charging and discharging process, and preprocessing the charging and discharging data to obtain the preprocessed charging and discharging data, comprises: Obtain the charging and discharging data during the current charging and discharging process; constructing the charge and discharge data into a feature vector; Normalize the feature vector to obtain a standardized feature vector; The standardized feature vector is used as the pre-processed charge and discharge data.

3. The method according to claim 1, characterized in that The step of inputting the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of the current charge and discharge process includes: Determine the output of the update gate based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the update gate; determine the output of the reset gate based on the previous hidden state, the preprocessed charge and discharge data and the parameters of the reset gate; Based on the output of the reset gate and the parameters of the candidate hidden state, a candidate hidden state is determined; based on the previous hidden state, the output of the update gate and the candidate hidden state, a current hidden state is determined, and the current hidden state is used as an output vector of the current charge and discharge process.

4. The method according to claim 1, characterized in that: The step of inputting the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain the first attention output vector comprises: Determine an attention score of the first head and an attention score of the second head based on the output vector, the query vector and the key vector of the current charging and discharging process; Determine the output vector of the first head and the output vector of the second head based on the attention score of the first head, the attention score of the second head, and the value vector; Based on the output vector of the first head, the output vector of the second head and the output vector of the current charging and discharging process, a first attention output vector is determined.

5. The method according to claim 1, characterized in that The step of inputting the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector includes: Determine an attention output vector based on the output vector and the convolution kernel of the current charging and discharging process; Based on the output vector of the current charging and discharging process and the attention output vector, a second attention output vector is determined.

6. The method according to claim 1, characterized in that The method further comprises: Acquire the charge and discharge data and battery capacity data of the lithium-ion battery, and process them to form a charge and discharge data set; Based on the multi-attention mechanism and the deep learning network, a lithium-ion battery health state estimation network is built; based on the charge and discharge data set, the lithium-ion battery health state estimation network is trained to obtain a trained lithium-ion battery health state estimation network.

7. A battery health status estimation device, characterized in that: The device comprises: An acquisition module is used to acquire the charge and discharge data of the lithium-ion battery in the current charge and discharge process, preprocess the charge and discharge data to obtain the preprocessed charge and discharge data, and acquire the output vector of the previous charge and discharge process as the previous hidden state; An extraction module is used to input the previous hidden state and the preprocessed charge and discharge data into a feature extraction network in a lithium-ion battery health state estimation network to obtain an output vector of a current charge and discharge process; the lithium-ion battery health state estimation network also includes a multi-attention network; An attention module, configured to input the output vector of the current charging and discharging process into the first attention module of the multi-attention network to obtain a first attention output vector; and input the output vector of the current charging and discharging process into the second attention module of the multi-attention network to obtain a second attention output vector; An evaluation module is used to determine a health state of a battery based on the first attention output vector and the second attention output vector.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Citation Information

Patent Citations

  • Lithium battery health state estimation method based on equal-voltage-rise energy and improved GRU

    CN115389946A

  • Modeling method of prediction network for lithium battery health state estimation

    CN116298934A

  • Lithium ion battery SOH prediction method based on Informer neural network

    CN116739164A

  • Nickel-cadmium storage battery life prediction method and device, computer equipment and storage medium

    CN117930009A

  • Deep learning lithium battery SOH prediction method based on transformer architecture

    CN118131044A