Lithium battery health degree evaluation method and system and electronic equipment

Abstract

The invention relates to a lithium battery health degree evaluation method and system and electronic equipment, and the method comprises the steps: obtaining historical battery data of a lithium battery, the battery data comprises voltage, current and temperature, and carrying out the feature fusion of the historical battery data through a convolutional neural network; taking the fused features and the corresponding historical battery health as training data, iteratively training a Gaussian process regression model by taking the minimization of a negative logarithm likelihood function as a target based on the training data through a particle swarm algorithm, and obtaining a lithium battery health degree evaluation model; and obtaining target battery data of the lithium battery, and obtaining target battery health through the lithium battery health degree evaluation model based on the target battery data.

Description

Technical Field

[0001] This application relates to the field of lithium battery health assessment, and particularly to a lithium battery health assessment method, system, and electronic device. Background Art

[0002] Accurate estimation of the battery health status is crucial for energy storage batteries. Battery health estimation can evaluate the aging state of the battery. The most commonly used health indicators of the battery include capacity and internal resistance. In actual scenarios, factors such as temperature, charge-discharge cycles, and battery life are considered when calculating the State of Health (SOH) of the battery for precise assessment.

[0003] Existing SOH assessment methods generally include model-based methods and data-driven methods. However, due to the complex internal degradation mechanism of the battery, the established assessment models are often correspondingly complex, resulting in difficult parameter identification and inability to accurately evaluate the battery health. Moreover, complex models often require long-term training, increasing the system response delay and affecting the user experience. In addition, in the face of changes in actual operations (such as temperature fluctuations and load changes), the model may fail, thus affecting the accuracy and reliability of the model assessment. Data-driven methods lack an effective feature fusion mechanism and cannot make full use of the relationships between different features, resulting in information loss and the existence of redundant features. Also, in the feature selection process, there is a lack of understanding of the internal state of the battery and its dynamic changes, resulting in the selected features may not reflect the actual situation, thus leading to the assessment accuracy of data-driven methods. In addition, current SOH assessment methods often rely on a single indicator, resulting in inaccurate battery health assessment results under complex conditions. Especially under the influence of multiple factors, a single indicator cannot comprehensively reflect the health state of the battery.

[0004] Existing lithium battery health assessment methods have the problem of poor assessment efficiency. Summary of the Invention

[0005] Embodiments of this application provide a lithium battery health assessment method, system, and electronic device to at least solve the problem of poor assessment efficiency in the related art of lithium battery health assessment methods.

[0006] In a first aspect, embodiments of this application provide a lithium battery health assessment method, including:

[0007] Obtain historical battery data of the lithium battery, where the battery data includes voltage, current, and temperature, and perform feature fusion on the historical battery data through a convolutional neural network;

[0008] Taking the fused features and the corresponding historical battery health as training data, iteratively train a Gaussian process regression model by using the particle swarm optimization algorithm with the goal of minimizing the negative log-likelihood function based on the training data, to obtain a lithium battery health assessment model;

[0009] Obtain the target battery data of the lithium battery, and obtain the target battery health based on the target battery data through the lithium battery health assessment model.

[0010] In one embodiment, the Gaussian process regression model includes:

[0011] Taking the squared exponential kernel function as the kernel function of the Gaussian process regression model, the kernel function is expressed as:

[0012]

[0013] where k(x i , x j ) represents the covariance between the input features x i and x j , is a hyperparameter representing the signal variance, and θ represents the length scale, characterizing the smoothness of the change of the kernel function.

[0014] In one embodiment, the training of the Gaussian process regression model with the goal of minimizing the negative log-likelihood function includes:

[0015] By optimizing the hyperparameter and the length scale θ, realize the minimization of the negative log-likelihood function of the Gaussian regression process model, and the negative log-likelihood function is expressed as:

[0016] L(σ f , θ) = -logP(Y|X, σ f , θ)

[0017] where L represents the negative log-likelihood estimate, and P(Y|X, σ f , θ) represents the probability of the Gaussian process of the input feature x and the output Y.

[0018] In one embodiment, the training of the Gaussian process regression model with the goal of minimizing the negative log-likelihood function based on the training data by using the particle swarm optimization algorithm includes:

[0019] Initialize the velocity and position of each particle, and calculate the negative log-likelihood value of each particle according to the position, where the position represents the current hyperparameter σ f and the length scale θ, and the velocity represents the moving direction and speed of the particle in the search space;

[0020] Iteratively update the position and velocity of each particle according to the negative log-likelihood value. The velocity update formula is expressed as:

[0021] V i (t + 1) = w·V i (t) + c 1 ·r 1 ·(P i -X i (t)) + c 2 ·r 2 ·(G - X i (t))

[0022] where V i (t) represents the velocity of particle i in the t-th iteration, w represents the inertia weight, c 1 and c 2 represent the acceleration constants, r 1 and r 2 are random numbers in the range of [0, 1], P i represents the best position of particle i in the previous iteration, and G represents the best position among all particles.

[0023] The position update formula is expressed as:

[0024] X i (t + 1) = X i (t) + V i (t + 1)

[0025] x i (t) represents the position of particle i in the t-th iteration, and V i (t + 1) represents the velocity of particle i in the (t + 1)-th iteration;

[0026] In response to satisfying the iteration condition, stop the iteration, obtain the optimized hyperparameters σ f and the length scale, and determine the optimized kernel function according to the optimized hyperparameters and the length scale;

[0027] Apply the optimized kernel function to the Gaussian regression model and train the Gaussian regression model with the training data.

[0028] In one embodiment, obtaining the target battery health through the lithium battery health assessment model based on the target battery data includes:

[0029] Determine the lithium battery health assessment model through the optimized kernel function and the length scale;

[0030] Perform feature fusion on the target battery data through the convolutional neural network model to obtain target fusion data;

[0031] Predict the target fusion data through the lithium battery health assessment model to obtain the target battery health.

[0032] In one embodiment, the feature fusion of the historical battery data through the convolutional neural network includes:

[0033] Integrate the historical battery data into a multi-channel input matrix, and extract the feature map of the multi-channel input matrix through the convolutional neural network;

[0034] Process the feature map through an activation function to increase the non-linearity of the feature map;

[0035] Screen the features in the feature map processed by the activation function through a pooling layer to reduce the size of the feature map;

[0036] Perform feature fusion on the feature map processed by the pooling layer through a fully connected layer to reduce the dimension of the feature map.

[0037] In a second aspect, an embodiment of the present application provides a lithium battery health assessment system, including:

[0038] Feature fusion module: used to obtain the historical battery data of the lithium battery, the battery data includes voltage, current and temperature, and perform feature fusion on the historical battery data through a convolutional neural network;

[0039] Model acquisition module: used to use the fused features and the corresponding historical battery health as training data, and iteratively train a Gaussian process regression model based on the training data with the goal of minimizing the negative log-likelihood function to obtain a lithium battery health assessment model;

[0040] Prediction module: used to obtain the target battery data of the lithium battery, and obtain the target battery health based on the target battery data through the lithium battery health assessment model.

[0041] In one embodiment, the prediction module includes:

[0042] Convolution unit: used to integrate the historical battery data into a multi-channel input matrix, and extract the feature map of the multi-channel input matrix through the convolutional neural network;

[0043] Activation function unit: used to process the feature map through an activation function to increase the non-linearity of the feature map;

[0044] Pooling unit: used to screen the features in the feature map processed by the activation function through a pooling layer to reduce the size of the feature map;

[0045] Fully connected unit: It is used to perform feature fusion on the feature map processed by the pooling layer through a fully connected layer to reduce the dimension of the feature map.

[0046] In a third aspect, an embodiment of the present application provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the lithium battery health assessment method as described in the first aspect above.

[0047] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the lithium battery health assessment method as described in the first aspect above.

[0048] A lithium battery health assessment method, system, and electronic device provided by an embodiment of the present application have at least the following technical effects.

[0049] In the present application, multi-feature deep fusion is performed through a convolutional neural network, which can explore the relationship between battery performance and aging from multiple dimensions, effectively capture the complex relationships between various features during the battery aging process, and thus improve the accuracy of SOH estimation. By using the particle swarm algorithm to train the Gaussian process regression model to obtain more accurate parameters suitable for the evaluation model, the evaluation performance of the lithium battery health assessment model is improved. Combining deep learning and ensemble learning can process and analyze a large amount of battery data in a short time, improving the real-time performance and accuracy of the lithium battery health estimation model. Realize real-time monitoring of the battery health status, quickly respond to possible potential faults, reduce accidental faults caused by battery aging, and thus reduce the cost of battery maintenance and replacement. Moreover, accurate SOH estimation helps to optimize the charging and discharging strategies of the battery, reduce the damage to the battery caused by improper use, and thus effectively extend the service life of the battery, further saving resources and costs.

[0050] Details of one or more embodiments of the present application are presented in the following drawings and description to make other features, purposes, and advantages of the present application more concise and understandable. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The drawings described herein are used to provide a further understanding of the present application, and constitute a part of the present application. The schematic embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation of the present application. In the drawings:

[0052] Figure 1 is a flowchart of a lithium battery health assessment method shown according to an embodiment of the present application;

[0053] Figure 2 is a structural block diagram of a lithium battery health assessment system shown according to an embodiment of the present application;

[0054] Figure 3 This is a schematic structural diagram of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0055] In order to make the objectives, technical solutions and advantages of the present application more clear and understandable, the present application will be described and illustrated below with reference to 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. All other embodiments obtained by those of ordinary skill in the art based on the embodiments provided in the present application without creative efforts fall within the scope of protection of the present application.

[0056] Obviously, the accompanying drawings in the following description are only some examples or embodiments of the present application. For those of ordinary skill in the art, without creative efforts, the present application can also be applied to other similar scenarios based on these drawings. In addition, it can also be understood that although the efforts made in such a development process may be complex and lengthy, for those of ordinary skill in the art related to the content disclosed in the present application, some design, manufacturing or production changes based on the technical content disclosed in the present application are only conventional technical means and should not be understood as the content disclosed in the present application being insufficient.

[0057] Referring to "embodiments" in the present application means that specific features, structures or characteristics described in connection with the embodiments can be included in at least one embodiment of the present application. The phrase appears in various positions in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those of ordinary skill in the art explicitly and implicitly understand that the embodiments described in the present application can be combined with other embodiments without conflict.

[0058] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the ordinary meanings understood by those with ordinary skills in the technical field to which this application belongs. The words such as "a", "an", "one", "the" and the like involved in this application do not indicate a quantity limitation and may represent a singular or plural number. The terms "comprising", "including", "having" and any variations thereof involved in this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or modules (units) is not limited to the listed steps or units, but may further include unlisted steps or units, or may further include other steps or units inherent to these processes, methods, products or devices. The similar words such as "connected", "coupled" and "linked" involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "plurality" involved in this application means two or more. "And / or" describes the association relationship of associated objects and indicates that three relationships may exist. For example, "A and / or B" may represent: A exists alone, A and B exist simultaneously, and B exists alone. The character " / " generally represents an "or" relationship between the front and back associated objects. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific order for the objects.

[0059] The existing model-based method and data-driven method for evaluating the health status of lithium batteries have the following defects: (1) Insufficient accuracy and reliability. The current SOH estimation methods often rely on a single indicator (such as capacity or internal resistance), resulting in inaccurate estimation results under complex conditions. Especially under the influence of multiple factors, a single indicator cannot comprehensively reflect the health status of the battery. Although the model-based method can provide a more in-depth analysis, in the face of changes in actual operation (such as temperature fluctuations, load changes), the model may fail, leading to a decrease in evaluation accuracy. (2) Limitations in feature selection. Many data-driven methods lack an effective feature fusion mechanism and fail to fully utilize the relationships between different features, resulting in information loss and the existence of redundant features, further affecting the performance of the model. And in the feature selection process, due to the lack of understanding of the internal state of the battery and its dynamic changes, the selected features may not reflect the actual situation of the battery. (3) Computational complexity. When dealing with high-dimensional data and feature fusion, the existing methods usually require a large amount of computing resources, which poses a challenge to real-time monitoring and evaluation, especially in the application of energy storage power stations. Complex models often require a long time to train, increasing the response delay of the system and affecting the user experience.

[0060] Based on the above situation, the embodiments of this application provide a method, a system and an electronic device for evaluating the health degree of lithium batteries.

[0061] In a first aspect, an embodiment of the present application provides a method for evaluating the health of a lithium battery. Figure 1 It is a flowchart of a method for evaluating the health of a lithium battery shown according to an embodiment of the present application. As Figure 1 shown, the method includes:

[0062] Step S101, obtain historical battery data of the lithium battery. The battery data includes voltage, current, and temperature, and perform feature fusion on the historical battery data through a convolutional neural network.

[0063] Optionally, the historical battery data can be represented as time series data. The type of the convolutional neural network is not limited and can be any one of ResNet, VGGNet, GoogleNet, etc. In this way, through multi-feature deep fusion by the convolutional neural network, the complex relationships between various features during the battery aging process can be effectively captured, thereby improving the accuracy of SOH estimation.

[0064] In one example, step S101:

[0065] Step S1011, integrate the historical battery data into a multi-channel input matrix, and extract the feature map of the multi-channel input matrix through a convolutional neural network. Optionally, the historical battery data is represented as time series data, where the voltage is represented as: V = [V(1), V(2),..., V(t)], the current is represented as I = [I(1), I(2),..., I(t)], and the temperature is represented as T = [T(1), T(2),..., T(t)], where t represents the t-th moment. Integrate the voltage, current, and temperature into a multi-channel input matrix, which is represented as:

[0066]

[0067] The feature representation of the multi-channel input matrix extracted through the convolution operation is:

[0068]

[0069] where G represents the output feature map, that is, the extracted feature map, (i, j) represents the position of the output feature map; K is the convolution kernel; [m, n] represents the size of the convolution kernel; b is the bias term parameter, M represents the row of the feature map, and N represents the column of the feature map.

[0070] Step S1012, process the feature map through an activation function to increase the non-linearity of the feature map. Optionally, the type of the activation function is not limited. Taking the ReLU activation function as an example to increase the non-linearity, the activation function is represented as:

[0071] A[G] = max(0, G)

[0072] Among them, G represents the feature map extracted by the convolutional layer, and A represents the feature map after activation processing.

[0073] Step S1013, screen the features in the feature map after the activation function processing through the pooling layer to reduce the size of the feature map. Optionally, in this application, the size of the feature map is reduced through the max-pooling operation, the features are screened, and the important features are retained.

[0074]

[0075] Among them, Z represents the feature map after pooling processing, s represents the stride of the pooling operation, A represents the feature map after activation processing, i represents the row of the feature map, j represents the column of the feature map, and m and n respectively represent the row and column sizes of the convolutional kernel of the pooling layer.

[0076] Step S1014, perform feature fusion on the feature map processed by the pooling layer through the fully connected layer to reduce the dimension of the feature map. Optionally, perform dimensionality reduction processing on the features processed in step S1013 through the fully connected layer, which is specifically expressed as:

[0077] F = W·A + b

[0078] Among them, F represents the output of the fully connected layer, W represents the weight matrix, A represents the feature map after activation processing, and b represents the bias term.

[0079] In this way, through steps S1011~S1014, through the convolutional layer, activation function, pooling layer and fully connected layer of the convolutional neural network, multi-features are extracted and deeply fused, so as to effectively capture the complex relationships between various features during the battery aging process and improve the accuracy of SOH estimation.

[0080] Step S102, use the fused features and the corresponding historical battery health as training data, and iteratively train the Gaussian process regression model based on the training data with the goal of minimizing the negative log-likelihood function to obtain a lithium battery health assessment model.

[0081] Optionally, train the Gaussian process regression model with the training data, and use the particle swarm optimization algorithm to optimize the kernel function and scale parameter of the Gaussian process regression model to obtain more accurate parameters suitable for the evaluation model, so as to improve the evaluation performance of the lithium battery health assessment model. The kernel function of the Gaussian process regression model includes but is not limited to any one of the linear kernel function, polynomial kernel function and Gaussian kernel function.

[0082] In one example, the Gaussian process regression model in step S102 includes:

[0083] Use the squared exponential kernel function as the kernel function of the Gaussian process regression model, and the kernel function is expressed as:

[0084]

[0085] where k(x i , x j ) represents the covariance between the input features x i and x j . is a hyperparameter representing the signal variance, θ represents the length scale, characterizing the smoothness of the change of the kernel function. Optionally, represents the signal variance and is used to control the variance of the function values. The signal variance and the length scale determine the smoothness and predictive ability of the Gaussian process.

[0086] In one example, step S102 includes:

[0087] By optimizing the hyperparameter and the length scale θ, minimize the negative log-likelihood function of the Gaussian regression process model, and the negative log-likelihood function is expressed as:

[0088] L(σ f , θ) = -logP(Y|X, σ f , θ)

[0089] where L represents the negative log-likelihood estimate, and P(Y|X, σ f , θ) represents the probability of the Gaussian process of the input feature x and the output Y.

[0090] In one example, step S102 includes:

[0091] Step S1021, initialize the velocity and position of each particle, and calculate the negative log-likelihood value of each particle according to the position, where the position represents the current hyperparameters σ f and the length scale θ, and the velocity represents the moving direction and speed of the particle in the search space.

[0092] Step S1022, iteratively update the position and velocity of each particle according to the negative log-likelihood value, and the velocity update formula is expressed as:

[0093] V i (t + 1) = w·V i (t) + c 1 ·r 1 ·(P i - X i (t)) + c 2 ·r 2 ·(G - X i (t))

[0094] where V i(t) represents the velocity of particle i in the t-th iteration, w represents the inertia weight, and c 1 and c 2 represent the acceleration constants, r 1 and r 2 are random numbers within the range [0, 1], P i represents the best position of particle i in previous iterations, G represents the best position among all particles, and x i (t) represents the position of particle i in the t-th iteration.

[0095] The position update formula is expressed as:

[0096] X i (t + 1) = X i (t) + V i (t + 1)

[0097] x i (t) represents the position of particle i in the t-th iteration, and V i (t + 1) represents the velocity of particle i in the (t + 1)-th iteration.

[0098] Optionally, in the process of optimizing the Gaussian regression model through the particle swarm algorithm, the hyperparameters in the Gaussian regression model are adjusted so that the model can better fit the data and improve the prediction accuracy. Among them, the smaller the negative log-likelihood value, the smaller the error of the model prediction, and the better the corresponding model parameters.

[0099] Step S1023, in response to meeting the iteration condition, stop the iteration, obtain the optimized hyperparameter σ f and the length scale, and determine the optimized kernel function according to the optimized hyperparameters and length scale.

[0100] Optionally, the iteration condition includes that the number of iterations reaches the maximum number of iterations, or the negative log-likelihood estimate value has not changed significantly in several iterations, indicating that the particle swarm has converged. Then, the parameter hyperparameters corresponding to the minimum negative log-likelihood estimate value during the training process are used as the model parameters.

[0101] Step S1024, apply the optimized kernel function to the Gaussian regression model and train the Gaussian regression model with the training data.

[0102] Optionally, the training process can be expressed as:

[0103]

[0104] where k represents the kernel function, θ represents the length scale, PSO represents the particle swarm optimization algorithm, and k * represents the finally solved optimal kernel function, and θ *Denote the optimal length scale for the final solution, F represents the fused features obtained by performing feature fusion on the input multi-dimensional data, and the feature SOH train Denote the historical battery health corresponding to the input data, RMSE represents the root mean square error, m represents the number of training data, and SOH i Denote the battery health predicted during the training process. Usually, the iteration stops when the number of iterations reaches the maximum number of iterations or the error is less than the preset threshold. In this example, the iteration stops when the error value is less than 0.001, and the model parameters at the time of the minimum error are taken as the optimal parameters. It should be noted that both the number of iterations and the error threshold at the end of the iteration can be defined according to the actual application scenario and are not limited to the examples listed in this application. In this way, combined with the Gaussian process regression model, the extracted fused features are processed, further improving the performance of the model in different battery SOH estimation tasks. Reducing the overfitting risk that may be brought by a single model, making the SOH estimation efficient and stable in various application scenarios.

[0105] Step S103, obtain the target battery data of the lithium battery, and obtain the target battery health based on the target battery data through the lithium battery health assessment model.

[0106] Optionally, the target battery data can be the battery data obtained in real time or the data obtained by other means such as a web page. Perform feature fusion on the obtained data, and predict the fused features through the lithium battery health assessment model to obtain the target battery health. In this way, the real-time monitoring of the battery health status is realized, the possible fault hidden dangers can be quickly responded to, the accidental faults caused by battery aging are reduced, thereby reducing the cost of maintenance and battery replacement. Moreover, accurate SOH estimation helps to optimize the charge and discharge strategy of the battery, reduce the damage to the battery caused by improper use, thereby effectively extending the service life of the battery and further saving resources and costs.

[0107] In one example, step S103 includes:

[0108] Step S301, determine the lithium battery health assessment model through the optimized kernel function and length scale. Optionally, apply the optimized kernel function to the Gaussian regression model and train the Gaussian regression model with the training data.

[0109] Step S302, perform feature fusion on the target battery data through the convolutional neural network model to obtain the target fused data.

[0110] Step S303, predict the target fused data through the lithium battery health assessment model to obtain the target battery health. Optionally, the prediction process can be expressed as:

[0111] SOH final = GP(F; k*, θ*)

[0112] Among them, SOH final is the battery health prediction result under the given target fusion data and optimization parameters k * and θ * and θ

[0113] In this way, by combining deep learning and ensemble learning, a large amount of battery data can be processed and analyzed in a short time, making the SOH estimation more efficient and improving the real-time performance and accuracy of the lithium battery health estimation model.

[0114] In summary, through multi-feature deep fusion by the convolutional neural network in this application, the relationship between battery performance and aging can be mined from multiple dimensions, and the complex relationship between various features during the battery aging process can be effectively captured, thereby improving the accuracy of SOH estimation. By using the particle swarm optimization algorithm to optimize the kernel function and scale parameters of the Gaussian process regression model to obtain more accurate parameters suitable for the evaluation model, the evaluation performance of the lithium battery health evaluation model is improved. By combining deep learning and ensemble learning, a large amount of battery data can be processed and analyzed in a short time, improving the real-time performance and accuracy of the lithium battery health estimation model. Realize real-time monitoring of the battery health status, quickly respond to possible potential faults, reduce accidental faults caused by battery aging, thereby reducing the cost of maintenance and battery replacement. And accurate SOH estimation helps to optimize the charging and discharging strategies of the battery, reduce the damage to the battery caused by improper use, thereby effectively extending the service life of the battery and further saving resources and costs.

[0115] In a second aspect, this application provides a lithium battery health evaluation system, Figure 2 which is a structural block diagram of a lithium battery health evaluation system shown according to an embodiment of this application, as Figure 2 shown, the system includes:

[0116] Feature fusion module 100: used to obtain historical battery data of the lithium battery, the battery data includes voltage, current and temperature, and perform feature fusion on the historical battery data through a convolutional neural network.

[0117] Model acquisition module 200: used to take the fused features and the corresponding historical battery health as training data, and iteratively train a Gaussian process regression model based on the training data with the goal of minimizing the negative log-likelihood function to obtain a lithium battery health evaluation model.

[0118] Prediction module 300: used to obtain target battery data of the lithium battery, and obtain target battery health based on the target battery data through the lithium battery health evaluation model.

[0119] In one example, the prediction module 300 includes:

[0120] Convolution unit: used to integrate historical battery data into a multi-channel input matrix, and extract the feature map of the multi-channel input matrix through a convolutional neural network.

[0121] Activation function unit: used to process the feature map through an activation function to increase the non-linearity of the feature map.

[0122] Pooling unit: used to screen the features in the feature map processed by the activation function through a pooling layer to reduce the size of the feature map.

[0123] Fully connected unit: used to perform feature fusion on the feature map processed by the pooling layer through a fully connected layer to reduce the dimension of the feature map.

[0124] In one example, the Gaussian process regression model in the model acquisition module 200 includes:

[0125] Taking the squared exponential kernel function as the kernel function of the Gaussian process regression model, the kernel function is expressed as:

[0126]

[0127] where k(x i , x j ) represents the covariance between the input features x i and x j , is a hyperparameter representing the signal variance, and θ represents the length scale, characterizing the smoothness of the change of the kernel function.

[0128] In one example, the model acquisition module 200 includes:

[0129] By optimizing the hyperparameter and the length scale θ, the negative log-likelihood function of the Gaussian regression process model is minimized, and the negative log-likelihood function is expressed as:

[0130] L(σ f , θ) = -logP(Y|X, σ f , θ)

[0131] where L represents the negative log-likelihood estimation, and P(Y|X, σ f , θ) represents the probability of the Gaussian process of the input feature x and the output Y.

[0132] In one example, the model acquisition module 200 includes:

[0133] Initialize the velocity and position of each particle, and calculate the negative log-likelihood value of each particle according to the position, where the position represents the current hyperparameter σ f and the length scale θ, and the velocity represents the moving direction and speed of the particle in the search space.

[0134] Iteratively update the position and velocity of each particle according to the negative log-likelihood value. The velocity update formula is expressed as:

[0135] V i (t + 1) = w·V i (t) + c 1 ·r 1 ·(P i -X i (t)) + c 2 ·r 2 ·(G - X i (t))

[0136] Among them, V i (t) represents the velocity of particle i in the t-th iteration, w represents the inertia weight, c 1 and c 2 represent the acceleration constants, r 1 and r 2 are random numbers in the range of [0, 1], P i represents the best position of particle i in the previous iteration, and G represents the best position among all particles.

[0137] The position update formula is expressed as:

[0138] X i (t + 1) = X i (t) + V i (t + 1)

[0139] x i (t) represents the position of particle i in the t-th iteration, and V i (t + 1) represents the velocity of particle i in the t + 1-th iteration.

[0140] In response to satisfying the iteration condition, stop the iteration, obtain the optimized hyperparameter σ f and the length scale, and determine the optimized kernel function according to the optimized hyperparameter and length scale.

[0141] Apply the optimized kernel function to the Gaussian regression model and train the Gaussian regression model with the training data.

[0142] In one example, the prediction module 300 includes:

[0143] Determine the lithium battery health assessment model through the optimized kernel function and length scale.

[0144] Perform feature fusion on the target battery data through the convolutional neural network model to obtain the target fusion data.

[0145] Predict the target fusion data through the lithium battery health assessment model to obtain the target battery health.

[0146] In summary, through multi-feature deep fusion by means of a convolutional neural network, this application can explore the relationship between battery performance and aging from multiple dimensions, effectively capture the complex relationships among various features during the battery aging process, thereby improving the accuracy of SOH estimation. By using the particle swarm optimization algorithm to optimize the kernel function and scale parameters of the Gaussian process regression model to obtain more accurate parameters suitable for the evaluation model, the evaluation performance of the lithium battery health assessment model is improved. Combining deep learning and ensemble learning can process and analyze a large amount of battery data in a short time, improving the real-time performance and accuracy of the lithium battery health estimation model. It realizes real-time monitoring of the battery health status, quickly responds to possible potential faults, reduces accidental faults caused by battery aging, thereby reducing the costs of maintenance and battery replacement. Moreover, accurate SOH estimation helps to optimize the charging and discharging strategies of the battery, reduce the damage to the battery caused by improper use, thereby effectively extending the service life of the battery and further saving resources and costs.

[0147] In a third aspect, an embodiment of this application provides an electronic device. Figure 3 It is a schematic structural diagram of an electronic device provided by an embodiment of this application. The electronic device includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the lithium battery health assessment method provided in the first aspect. Figure 3 The illustrated electronic device 60 is merely an example and should not impose any limitation on the functions and usage scope of the embodiments of this application.

[0148] The electronic device 60 may be presented in the form of a general computing device. For example, it may be a server device. The components of the electronic device 60 may include, but are not limited to: at least one of the above-mentioned processors 61, at least one of the above-mentioned memories 62, and a bus 63 connecting different system components (including the memory 62 and the processor 61).

[0149] The bus 63 includes a data bus, an address bus, and a control bus.

[0150] The memory 62 may include volatile memory, such as a random access memory (RAM) 621 and / or a cache memory 622, and may further include a read-only memory (ROM) 623.

[0151] The memory 62 may also include a program / utilities 625 having a set (at least one) of program modules 624. Such program modules 624 include, but are not limited to: an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment.

[0152] The processor 61 executes various functional applications and data processing by running computer programs stored in the memory 62, such as the lithium battery health assessment method provided in the first aspect of the present application.

[0153] The electronic device 60 may also communicate with one or more external devices 64 (such as a keyboard, a pointing device, etc.). Such communication may be carried out through an input / output (I / O) interface 65. And, the device 60 for generating a model may also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 66. As shown in the figure, the network adapter 66 communicates with other modules of the device 60 for generating a model through a bus 63. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in combination with the device 60 for generating a model, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (redundant array of independent disks) systems, magnetic tape drives, and data backup storage systems, etc.

[0154] It should be noted that, although several units / modules or sub-units / modules of the electronic device are mentioned in the above detailed description, such a division is merely exemplary and not mandatory. In fact, according to the embodiments of the present invention, the features and functions of two or more of the above-described units / modules may be embodied in one unit / module. Conversely, the features and functions of one unit / module described above may be further divided and embodied by multiple units / modules.

[0155] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a program is stored. When the program is executed by a processor, the lithium battery health assessment method provided in the first aspect is implemented.

[0156] Among them, the more specific forms that the readable storage medium may adopt may include, but are not limited to: a portable disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0157] In a possible implementation manner, the present invention may also be implemented in the form of a program product, which includes program code. When the program product runs on a terminal device, the program code is used to cause the terminal device to execute the steps of implementing the lithium battery health assessment method provided in the first aspect.

[0158] Among them, the program code for implementing the present invention can be written in any combination of one or more programming languages, and the program code can be executed entirely on the user device, partially on the user device, executed as an independent software package, partially on the user device and partially on a remote device, or entirely on a remote device.

[0159] The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0160] The above-described embodiments merely represent several implementation manners of the present application, and the description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.

Claims

1. A lithium battery health assessment method, characterized in that: include: Acquire historical battery data of a lithium battery, the battery data including voltage, current and temperature, and perform feature fusion on the historical battery data through a convolutional neural network; The fused features and the corresponding historical battery health are used as training data, and a Gaussian process regression model is iteratively trained based on the training data by a particle swarm algorithm with the goal of minimizing the negative log-likelihood function to obtain a lithium battery health assessment model; Target battery data of the lithium battery is acquired, and target battery health is obtained through the lithium battery health assessment model based on the target battery data.

2. The lithium battery health assessment method according to claim 1, characterized in that: The Gaussian process regression model includes: The square exponential kernel function is used as the kernel function of the Gaussian process regression model, and the kernel function is expressed as: Among them, k(x i ,x j ) represents the input feature x i and x j The covariance between is a hyperparameter, representing the signal variance, and θ represents the length scale, which characterizes the smoothness of the change of the kernel function.

3. The lithium battery health assessment method according to claim 2, characterized in that: The Gaussian process regression model is trained with the goal of minimizing the negative log-likelihood function, including: By optimizing the hyperparameters and length scale θ, to minimize the negative log-likelihood function of the Gaussian regression process model, which is expressed as: L(s f ,θ)=-logP(Y|X,σ f ,i) Among them, L represents the negative log-likelihood estimate, P(Y|X,σ f , θ) represents the probability of the Gaussian process of input feature X and output Y.

4. The lithium battery health assessment method according to claim 3, characterized in that: The method of training the Gaussian process regression model based on the training data by using a particle swarm algorithm with the goal of minimizing a negative log-likelihood function comprises: Initialize the velocity and position of each particle and calculate the negative log-likelihood of each particle based on the position, which represents the current hyperparameter σ f and length scale θ, the velocity characterizes the moving direction and speed of the particle in the search space; The position and velocity of each particle are iteratively updated according to the negative log-likelihood value, and the velocity update formula is expressed as: V i (t+1)=w·V i (t)+c1·r1·(P i -X i (t))+c2·r2·(GX i (t)) where V i (t) represents the velocity of particle i in the tth iteration, w represents the inertia weight, c1 and c2 represent acceleration constants, r1 and r2 are random numbers in the range [0,1], P i represents the best position of particle i in the previous iteration, G represents the best position among all particles, The position update formula is expressed as: X i (t+1)=X i (t)+V i (t+1) X i (t) represents the position of particle i in the tth iteration, V i (t+1) represents the velocity of particle i in the t+1th iteration; In response to satisfying the iteration condition, stop the iteration and obtain the optimized hyperparameter σ f and length scale, determining the optimized kernel function according to the optimized hyperparameters and length scale; The optimized kernel function is applied to the Gaussian regression model, and the Gaussian regression model is trained using the training data.

5. The lithium battery health assessment method according to claim 4, characterized in that: The step of obtaining a target battery health by using the lithium battery health assessment model based on the target battery data includes: Determine the lithium battery health assessment model through the optimized kernel function and length scale; Performing feature fusion on the target battery data through the convolutional neural network model to obtain target fusion data; The target fusion data is predicted by the lithium battery health assessment model to obtain the target battery health.

6. The lithium battery health assessment method according to claim 1, characterized in that: The feature fusion of the historical battery data by using a convolutional neural network includes: Integrate the historical battery data into a multi-channel input matrix, and extract a feature map of the multi-channel input matrix through the convolutional neural network; Processing the feature map by an activation function to increase the nonlinearity of the feature map; The features in the feature map processed by the activation function are filtered through a pooling layer to reduce the size of the feature map; The feature map processed by the pooling layer is subjected to feature fusion through a fully connected layer to reduce the dimension of the feature map.

7. A lithium battery health assessment system, characterized in that: include: Feature fusion module: used to obtain historical battery data of lithium batteries, the battery data including voltage, current and temperature, and perform feature fusion on the historical battery data through a convolutional neural network; Model acquisition module: used to use the fused features and the corresponding historical battery health as training data, and iteratively train the Gaussian process regression model based on the training data with the goal of minimizing the negative log-likelihood function through the particle swarm algorithm to obtain a lithium battery health assessment model; Prediction module: used to obtain target battery data of the lithium battery, and obtain the target battery health based on the target battery data through the lithium battery health assessment model.

8. The lithium battery health assessment system according to claim 7, characterized in that: The prediction module comprises: Convolution unit: used to integrate the historical battery data into a multi-channel input matrix, and extract a feature map of the multi-channel input matrix through the convolutional neural network; An activation function unit: used for processing the feature map through an activation function to increase the nonlinearity of the feature map; Pooling unit: used for filtering features in the feature map after the activation function is processed through a pooling layer to reduce the size of the feature map; Fully connected unit: used to perform feature fusion on the feature map processed by the pooling layer through the fully connected layer to reduce the dimension of the feature map.

9. An electronic device, characterized in that: include Memory, processor, and A computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the lithium battery health assessment method as described in any one of claims 1 to 6 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the lithium battery health assessment method as described in any one of claims 1 to 6 is implemented.

Images