Method, device, apparatus and readable storage medium for determining battery state of charge

Abstract

This application discloses a method, apparatus, device, and readable storage medium for determining the state of charge of a battery, relating to the field of battery management technology. The method includes establishing a time kernel function and a space kernel function based on the historical state parameters of a battery cell, constructing a global integration model based on the time kernel function and the space kernel function, and finally determining the state of charge of the battery cell based on the measured value of the real-time state of charge of the target battery cell and the global integration model. The time kernel function and the space kernel function comprehensively characterize the correlation between the temporal evolution and spatial distribution of the battery cell, thereby constructing a global integration model with global coupling characteristics, thereby improving the accuracy of the state of charge determined by the global integration model.

Description

Technical Field

[0001] The present application relates to the field of battery management technology, and in particular to a method, apparatus, device, and readable storage medium for determining a battery state of charge. Background Art

[0002] In the battery management system, State of Charge (SOC), as a basic parameter for evaluating battery performance, is an important support for achieving full life cycle management of batteries.

[0003] Currently, SOC is typically estimated through direct measurement or by building an equivalent model. While direct measurement is simple to implement, it is susceptible to inaccurate initial values ​​and accumulated sensor errors, causing the estimated result to gradually deviate from the actual value. In contrast, building an equivalent model generally achieves higher estimation accuracy, but often suffers from the model's strong dependence on the application scenario, resulting in poor generalization ability, which in turn limits the accuracy and applicability of SOC estimation. Summary of the Invention

[0004] The present application provides a method, apparatus, device, and readable storage medium for determining the state of charge of a battery, so as to at least solve the problem of limited accuracy in determining the state of charge in the related art.

[0005] This application provides a method for determining a battery state of charge, comprising:

[0006] Obtaining historical status parameters of each battery cell of a target battery; the target battery includes multiple battery cells;

[0007] Constructing the time kernel function and space kernel function of the target battery based on the historical state parameters; the time kernel function is used to characterize the temporal correlation between the historical state parameters of multiple battery cells; the space kernel function is used to characterize the spatial correlation between multiple battery cells;

[0008] Construct a global integration model based on the temporal kernel function and the spatial kernel function;

[0009] Performing hidden state recursive estimation on the global integrated model based on the measured real-time state of charge of the target battery cell in the target battery to determine an estimated value of the state of charge of the target battery cell;

[0010] The state of charge of the target battery cell is determined based on the historical state parameters and the estimated value.

[0011] The present application also provides a device for determining a battery state of charge, comprising: an acquisition module for acquiring historical state parameters of each battery cell of a target battery; the target battery includes a plurality of battery cells;

[0012] A construction module is used to construct a time kernel function and a space kernel function of a target battery based on historical state parameters; the time kernel function is used to characterize the temporal correlation between historical state parameters of multiple battery cells; the space kernel function is used to characterize the spatial correlation between multiple battery cells;

[0013] The building module is also used to build a global integration model based on the time kernel function and the spatial kernel function;

[0014] A processing module, configured to perform hidden state recursive estimation on the global integrated model based on the measured value of the real-time state of charge of the target battery cell in the target battery, and determine an estimated value of the state of charge of the target battery cell;

[0015] The determination module is used to determine the state of charge of the target battery cell based on the historical state parameters and the estimated value.

[0016] The present application also provides an electronic device, comprising: a memory for storing a computer program; and a processor for implementing the steps of the above-mentioned method for determining the battery state of charge when executing the computer program.

[0017] The present application also provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned method for determining the battery state of charge are implemented.

[0018] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned method for determining the battery state of charge when the computer program is executed by a processor.

[0019] Through this application, since it is possible to establish a time kernel function and a space kernel function based on the historical state parameters of the battery cell, and to construct a global integration model based on the time kernel function and the space kernel function, the state of charge of the battery cell is finally determined based on the measured value of the real-time state of charge of the target battery cell and the global integration model. In this way, the correlation between the time evolution and spatial distribution of the battery cell is fully characterized by the time kernel function and the space kernel function, and then a global integration model with global coupling characteristics is constructed, thereby improving the accuracy of the state of charge determined by the global integration model. In addition, it is also possible to robustly recursively infer the state of charge of the battery cell in the presence of observation noise and initial uncertainty, further improving the accuracy and adaptability of the determined state of charge. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0021] Figure 1 A schematic diagram of an application scenario of a method for determining a battery state of charge provided in an embodiment of the present application;

[0022] Figure 2 One of the flow charts of the method for determining the battery state of charge provided in an embodiment of the present application;

[0023] Figure 3 This is a second flowchart of a method for determining a battery state of charge provided in an embodiment of the present application;

[0024] Figure 4 Flowchart 3 of the method for determining the battery state of charge provided in an embodiment of the present application;

[0025] Figure 5 Flowchart 4 of the method for determining the battery state of charge provided in an embodiment of the present application;

[0026] Figure 6 Flowchart 5 of the method for determining the battery state of charge provided in an embodiment of the present application;

[0027] Figure 7 Flowchart 6 of the method for determining the battery state of charge provided in an embodiment of the present application;

[0028] Figure 8 A schematic diagram of the structure of a device for determining the state of charge of a battery provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0030] It should be noted that, in the description of this application, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. The terms "first," "second," etc., in this application are used to distinguish similar objects, and are not used to describe a particular order or sequence.

[0031] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0032] Building a battery state-of-charge estimation system is crucial for dynamically estimating the battery's operating status. This system improves energy efficiency by optimizing charge and discharge threshold settings, while also enabling the analysis of aging characteristics and the establishment of safety monitoring and early warning mechanisms. Therefore, accurate, real-time, and effective battery state-of-charge estimation is not only directly related to improving the operational efficiency of the battery energy management system, but also crucial for driving breakthroughs in battery life and safety performance upgrades.

[0033] Currently, SOC is typically estimated through direct measurement, equivalent model-based, data-driven methods, and other approaches. The ampere-hour integration method and the power integration method are direct measurement methods for SOC estimation, but they are susceptible to inaccurate initial values, and sensor errors gradually accumulate, leading to increased error in the estimated value.

[0034] Equivalent model-based approaches, such as the open-circuit voltage method, use the open-circuit voltage versus battery state-of-charge curve to estimate the battery's state-of-charge. However, this method requires a long period of resting the battery and cannot estimate the battery's state-of-charge under real-time operating conditions. Another example is the electrochemical model-based approach, which explains battery aging mechanisms by modeling the battery's internal chemical reactions. However, electrochemical models typically involve multiple complex equations and highly coupled model parameters and are primarily used to study internal electrochemical reactions. The basic principle of the equivalent circuit model (ECM) and fractional-order model (FOM) is to use characteristic parameters exhibited by lithium batteries during cycling, establish a relationship between these parameters and the state-of-charge through filtering algorithms, and further estimate the battery's state-of-charge. While these equivalent model-based methods can accurately estimate the battery's state-of-charge, they rely heavily on the accuracy of the model and parameters, require different models to address different situations, and have poor generalization capabilities, limiting the accuracy and applicability of SOC estimation.

[0035] Data-driven methods, such as artificial neural networks (ANNs) and support vector machines (SVMs), require training with large amounts of battery data to learn the nonlinear mapping relationship between input and output, thereby estimating the battery's state of charge. While these methods offer good adaptability and generalization capabilities, they require extensive historical data for training, and the accuracy and reliability of the models depend heavily on the quality and quantity of the data.

[0036] In summary, existing battery state-of-charge estimation methods each have their own strengths and weaknesses. Direct measurement methods are simple and easy to implement, but suffer from serious error accumulation issues. Equivalent model-based methods can deeply characterize the internal characteristics of batteries, but they struggle to balance model complexity and generalization capabilities. Data-driven methods are highly adaptable, but their dependence on data limits their application in certain scenarios.

[0037] To address the above issues, embodiments of the present application provide a method, apparatus, device, and readable storage medium for determining a battery state of charge. The method includes establishing a time kernel function and a space kernel function based on the historical state parameters of the battery cells, constructing a global integration model based on the time kernel function and the space kernel function, and finally determining the state of charge of the battery cells based on the measured value of the real-time state of charge of the target battery cells and the global integration model. In this way, the time kernel function and the space kernel function comprehensively characterize the correlation between the temporal evolution and spatial distribution of the battery cells, and then construct a global integration model with global coupling characteristics, thereby improving the accuracy of the state of charge determined by the global integration model.

[0038] Figure 1 Schematic diagram of an application scenario of a method for determining the state of charge of a battery provided in an embodiment of the present application. Figure 1 As shown, the scenario includes a server 100 and an electronic device 200 .

[0039] The electronic device 200 provided in the embodiment of the present application may have various implementation forms, for example, it may be a mobile phone, a personal computer (PC), a wearable device, an in-vehicle device, etc.

[0040] In some embodiments, upon receiving the battery state of charge determination instruction, the electronic device 200 may communicate data with the server 100. The electronic device 200 may be allowed to communicate with the server 100 via a local area network (LAN) or a wireless local area network (WLAN).

[0041] The server 100 may be a server that provides various services, such as a server that supports historical status parameters uploaded by the electronic device 200. The server 100 may be a server cluster or multiple server clusters, and may include one or more types of servers.

[0042] It should be noted that the method for determining the battery state of charge provided in the embodiment of the present application can be executed by the electronic device 200, can be executed by the server 100, or can be executed jointly by the server 100 and the electronic device 200.

[0043] The battery state of charge determination device provided in the embodiment of the present application can be hardware or software. When the battery state of charge determination device is hardware, it can be various electronic devices 200 with a battery state of charge determination function, including but not limited to smartphones, televisions, tablets, smart watches, computers, AI devices, robots, smart vehicles, etc. When the battery state of charge determination device is software, it can be installed in the electronic devices 200 listed above. It can be implemented as multiple software or software modules (for example, to provide a battery state of charge determination service), or it can be implemented as a single software or software module. No specific limitation is made here.

[0044] Figure 2 This is a flow chart of a method for determining the state of charge of a battery provided in an embodiment of the present application. Figure 2 As shown, the method for determining the battery state of charge may include the following steps:

[0045] S11. Obtain historical status parameters of each battery cell of the target battery.

[0046] The target battery includes multiple battery cells. In some embodiments, the target battery is a lithium battery. The historical state parameters include at least the sampling time, electrical parameters, and the spatial position of the battery cell in the target battery (the spatial position can be represented by a spatial coordinate point). The electrical parameters can include various state parameters related to the battery cell, such as voltage, current, temperature, state of charge, etc.

[0047] Specifically, a method for obtaining the historical state parameters of each battery cell of the target battery may be to obtain historical state data of each battery cell of the target battery in charge and discharge states under different working conditions.

[0048] S12. Constructing a time kernel function and a space kernel function of the target battery according to the historical state parameters.

[0049] Among them, the time kernel function is used to characterize the temporal correlation between the historical state parameters of multiple battery cells; the spatial kernel function is used to characterize the spatial correlation between multiple battery cells.

[0050] Specifically, the temporal and spatial kernel functions of the target battery can be constructed based on historical state parameters. The temporal kernel function can be constructed based on sampling time and electrical parameters, while the spatial kernel function can be constructed based on spatial position and electrical parameters. Specifically, the temporal kernel function can be constructed based on sampling time by analyzing the temporal correlations between the historical state parameters of multiple battery cells at different time points to create a temporal kernel function that reflects the temporal evolution of the state. The spatial kernel function can be constructed based on historical state parameters by analyzing the spatial correlations between the historical state parameters of different battery cells based on their spatial positions within the target battery to create a spatial kernel function that reflects the spatial dependencies between cells. This spatial kernel function can express the similarity and spatial structure of adjacent cell states, thereby helping to capture the overall spatiotemporal dynamics of the target battery. In this approach, constructing the temporal kernel function based on sampling time and electrical parameters helps reveal temporal variations in battery performance. Constructing the spatial kernel function based on the spatial position and electrical parameters of the battery cells within the target battery can reflect the differences and spatial distribution characteristics between cells at different locations. In this way, the operating status of the target battery is comprehensively characterized from the two dimensions of time and space, which improves the accuracy of state modeling and provides a more reliable data basis for subsequent state of charge assessment.

[0051] S13. Construct a global integration model based on the time kernel function and the space kernel function.

[0052] Among them, the global integration model is used to describe the evolution process of the state parameters of the battery cells in the target battery at different time points and spatial positions.

[0053] Specifically, the global integration model can be constructed based on the temporal and spatial kernel functions by combining them to form a joint spatiotemporal kernel function, thereby constructing a global integration model that considers both temporal and spatial dependencies. This model can capture the dynamic evolution of battery cell state parameters at different time points and spatial locations, enabling a comprehensive description and prediction of the target battery state.

[0054] S14 , performing hidden state recursive estimation on the global integrated model according to the measured value of the real-time state of charge of the target battery cell in the target battery, and determining an estimated value of the state of charge of the target battery cell.

[0055] The measured value is the target battery cell state of charge measured in real time by a direct measurement method (such as the ampere-hour integration method and the power integration method), which is used to assist in estimating the target battery cell state of charge, thereby obtaining an estimated value of the target battery cell state of charge.

[0056] Specifically, based on the measured value of the real-time state of charge of the target battery cell in the target battery, the hidden state of the global integrated model is recursively estimated to determine the estimated value of the state of charge of the target battery cell. The method can be based on the measured value of the real-time state of charge of the target battery cell, combined with the evolution law of the battery state in the time and space dimensions described by the global integrated model, and through an algorithm or model with inference estimation capabilities, recursively estimate the state of charge of the target battery cell at the current moment, thereby obtaining an estimated value.

[0057] S15. Determine the state of charge of the target battery cell based on the historical state parameters and the estimated value.

[0058] Specifically, the SOC of the target battery cell can be determined based on the historical state parameters and the estimated value by first using a spatial correlation algorithm to calculate the spatial correlation parameters between the target battery cell and other battery cells based on the historical state parameters; then combining the spatial correlation parameters with the estimated value to determine the SOC of the target battery cell. The spatial correlation algorithm can be the spatial kernel function in S12, or other spatial correlation algorithms, or a pre-trained model can be used as the spatial correlation algorithm to calculate the spatial correlation parameters.

[0059] In the above scheme, since time and space kernel functions can be established based on the historical state parameters of the battery cells, and a global integration model is constructed based on the time and space kernel functions, the battery cell's state of charge is finally determined based on the measured value of the target battery cell's real-time state of charge and the global integration model. In this way, the time and space kernel functions comprehensively characterize the correlation between the battery cells' temporal evolution and spatial distribution, and then a global integration model with global coupling characteristics is constructed, thereby improving the accuracy of the state of charge determined by the global integration model. In addition, the state of charge of the battery cells can be robustly recursively in the presence of observation noise and initial uncertainty, further improving the accuracy and adaptability of the determined state of charge.

[0060] In some embodiments, as Figure 3 As shown, the electrical parameters include the historical state of charge; the method of constructing the time kernel function of the target battery according to the sampling time and the electrical parameters may include the following steps:

[0061] S1201: Determine the standard deviation of historical states of charge of multiple battery cells.

[0062] Specifically, the method for determining the standard deviation of the historical state of charge of multiple battery cells can be to first determine the average value of the historical state of charge of the multiple battery cells; then, determine the standard deviation of the historical state of charge of the multiple battery cells based on the historical state of charge of the multiple battery cells, the average value and the total number of historical state of charge data.

[0063] For example, according to the formula To determine the standard deviation of the historical state of charge of multiple battery cells. Used to represent the standard deviation of the historical state of charge of multiple battery cells. Used to indicate the historical state of charge of any battery cell. Used to represent the average value of the historical state of charge of multiple battery cells. n The total number of data points used to represent the historical state of charge.

[0064] S1202 : Perform fitting processing on the electrical parameters according to the sampling time to obtain a time-dependent decay rate of the electrical parameters.

[0065] Specifically, fitting the electrical parameters according to the sampling time to obtain the time-correlated decay rate of the electrical parameters can be performed by constructing a function model of the electrical parameters' time-varying behavior based on the electrical parameters acquired at multiple sampling times. For example, this model can be fitted using an exponential decay function, a power function, or a Gaussian process. By analyzing the time factor in the fitting model, the time-correlated decay characteristics of the electrical parameters can be extracted, thereby obtaining their time-correlated decay rate.

[0066] It should be noted that the time-dependent decay rate is a value that changes over time according to the state of charge during use. It has the following relationship with the state of charge: when the state of charge changes rapidly, the weight of the historical state parameter is low, and the time-dependent decay rate will become high; when the state of charge changes slowly, the weight of the historical state parameter is high, and the time-dependent decay rate value will decrease.

[0067] S1203: Construct a time kernel function of the target battery according to the standard deviation, the time correlation decay rate, and the time difference.

[0068] The time difference is the time difference between any two sampling times.

[0069] Specifically, based on the standard deviation, time-correlated decay rate and time difference, the time kernel function of the target battery is constructed in such a way that the output of the time kernel function is determined as the product of the standard deviation and a first function, wherein the first function is an exponential function with the natural logarithm as the base and the product of the negative time-correlated decay rate and the absolute value of the time difference as the exponent.

[0070] In some embodiments, the time kernel function can be expressed as: .

[0071] in, Used to represent the output of the time kernel function, Used to represent standard deviation, Used to express the time-dependent decay rate, Used to express time difference, is the base of natural logarithms.

[0072] In the above scheme, the standard deviation reflects the consistency of the state of charge between battery cells, the time-correlated decay rate reveals the decay trend of electrical parameters over time, and the time difference quantifies the time interval between samples. The three work together to construct a time kernel function that can effectively capture the state correlation of the target battery at different time points, thereby improving the accuracy and robustness of the global integration model subsequently constructed based on the time kernel function.

[0073] In some embodiments, as Figure 4 As shown, the electrical parameters include at least the historical state of charge; the method of constructing the spatial kernel function of the target battery according to the spatial position and the electrical parameters may include the following steps:

[0074] S1211: Determine the standard deviation of the historical states of charge of multiple battery cells.

[0075] Specifically, the method for determining the standard deviation is the same as S1201 and will not be repeated here.

[0076] S1212: Construct a spatial kernel function of the target battery according to the standard deviation, spatial distance, and electrical parameters.

[0077] The spatial distance is the distance between any two battery cells among the plurality of battery cells.

[0078] Specifically, the spatial kernel function can be: .

[0079] in, and are used to represent electrical parameters, and and corresponding to different battery cells among the plurality of battery cells; Used to represent the output of the spatial kernel function, for and The corresponding spatial distance between the two battery cells, Used to represent standard deviation, is the base of natural logarithms.

[0080] In the above scheme, the standard deviation of the historical state of charge reflects the consistency of battery cell performance, and the spatial distance reflects the actual physical layout of the battery cells. Introducing these two together with electrical parameters into the spatial kernel function helps capture the correlation and coupling effects between battery cells in the local area, thereby improving the accuracy and robustness of the subsequent global integration model constructed based on the spatial kernel function.

[0081] In some embodiments, as Figure 5 As shown in FIG, the method of constructing a global integration model based on the time kernel function and the space kernel function may include the following steps:

[0082] S131. Perform rational spectral decomposition on the time kernel function to obtain hidden state variables of battery cells in the target battery.

[0083] Rational spectral decomposition is a matrix factorization method that decomposes a matrix into multiple components, each corresponding to a rational irreducible polynomial in the matrix's minimal polynomial. Through rational spectral decomposition, a matrix is ​​represented as the sum of several projection matrices, each of which reflects the matrix's behavior under the corresponding rational irreducible polynomial.

[0084] Specifically, a method for performing rational spectral decomposition on the time kernel function to obtain the hidden state variables of the battery cells in the target battery can be to perform rational spectral decomposition on the time kernel function to obtain the hidden state variables of multiple battery cells in the target battery, where one battery cell corresponds to one hidden state variable, and one hidden state variable corresponds to a rational irreducible polynomial.

[0085] S132. Construct a continuous-time state-space model of the target battery according to the hidden state variables, and construct a global state-space model according to the continuous-time state-space model and the spatial kernel function.

[0086] First, a continuous-time state-space model of the target battery is constructed based on the hidden state variables.

[0087] Specifically, the continuous-time state space model can be expressed as: ; .

[0088] in, Used to indicate the i The hidden state change rate of each battery cell, Used to express the time-dependent decay rate of electrical parameters, Used to indicate the i The hidden state variables of each battery cell, The Gaussian white noise used to represent the driving state equation is Used to indicate the i Each battery cell contains a time-dynamic state-of-charge component, Used to represent the standard deviation of the historical state of charge of multiple battery cells.

[0089] Secondly, a global state space model is constructed based on the continuous-time state space model and the spatial kernel function.

[0090] Specifically, the global state space model can be expressed as: ; .

[0091] in, Used to indicate The set of discrete state vectors corresponding to the k-th moment after discretization, Used to indicate k The discrete observation vector at time t, Used to represent discretization time interval; Used to represent process noise, is used to represent the measurement noise, It is used to represent the lower triangular matrix after Cholesky decomposition of the spatial kernel matrix (calculated according to the spatial kernel function). Used to represent the Kronecker product, is the base of natural logarithms, Used to express the time-dependent decay rate of electrical parameters, Used to represent the standard deviation of the historical state of charge of multiple battery cells.

[0092] S133 : Integrate the battery cells in the target battery according to the continuous-time state-space model and the global state-space model to obtain a global integrated model.

[0093] In some embodiments, the global integration model may be: ; .

[0094] in, Used to express the time-dependent decay rate of electrical parameters, Used to represent the standard deviation of the historical state of charge of multiple battery cells. Used to represent discretization time interval, Used to represent process noise, is used to represent the measurement noise, It is used to represent the lower triangular matrix after Cholesky decomposition of the spatial kernel matrix. Used to represent the Kronecker product, Used to indicate The set of discrete state vectors at time , Used to indicate k The discrete observation vector at time t, is the base of natural logarithms; for M dimensional global identity matrix, M is the number of battery cells in the target battery.

[0095] In this approach, a rational spectral decomposition of the time kernel function is performed to extract the hidden state variables of the battery cells, enabling a deep analysis of the battery's dynamic characteristics. The continuous-time state-space model constructed using the hidden state variables accurately depicts the behavior of the battery cells as they evolve over time. Furthermore, a global state-space model constructed in conjunction with the spatial kernel function effectively integrates temporal and spatial information, enhancing the model's ability to describe the overall behavior of the system. Finally, a global integrated model is used to uniformly characterize the coevolutionary process of each cell in the battery system, improving the accuracy of state-of-charge estimation.

[0096] In some embodiments, as Figure 6 As shown, a method for performing hidden state recursive estimation on the global integrated model based on the measured value of the real-time state of charge of the target battery cell in the target battery to determine the estimated value of the state of charge of the target battery cell may include the following steps:

[0097] S141 , using Kalman filtering to perform hidden state recursive estimation on the global integrated model according to the measured value of the real-time state of charge of the target battery cell in the target battery, to obtain the optimal posterior estimation value at the current moment.

[0098] Among them, the Kalman filter recursive update process is as follows:

[0099] Prediction step: .

[0100] Update step: .

[0101] in, A is the discrete state transfer matrix, ; The real-time state of charge measurement value of the target battery cell in the target battery; is the Kalman gain, which is used to dynamically balance the confidence of the estimated value and the measured value. ; C is the output matrix of the global integration model, , for C The transposed matrix of R is the observation noise covariance matrix, ; is the measurement noise variance; is the output of the prediction step, is the output of the update step.

[0102] Specifically, a Kalman filter is used to recursively estimate the hidden state of the global integrated model based on the measured real-time state of charge of the target battery cells in the target battery. The optimal posterior estimate at the current moment can be obtained by temporally updating the hidden state using the state transition matrix according to the Kalman filter's prediction step to obtain the current a priori estimate and covariance matrix. Subsequently, the Kalman gain is calculated in the update step based on the measured state of charge of the target battery cells at the current moment. This gain is used to weight and modify the prior estimate to obtain the optimal posterior estimate at the current moment. This recursive process is repeated at each sampling moment to achieve real-time dynamic estimation of the target battery state of charge.

[0103] S142. Obtain the output matrix of the global integration model.

[0104] Specifically, the output matrix of the global integration model is .

[0105] S143. Determine the product of the output matrix and the optimal posterior estimation value as the estimation value.

[0106] Specifically, the estimated value can be calculated as .in, is an estimated value, C is the output matrix, is the optimal posterior estimate at the current moment.

[0107] In this solution, a Kalman filter is introduced to perform recursive estimation of the hidden state of the global integrated model, effectively integrating the real-time measurements of the target battery cells with the model's predictions, thereby improving the real-time performance and accuracy of state estimation. The Kalman filter can obtain the optimal a posteriori estimate at the current moment in the presence of noise and uncertainty, ensuring highly reliable dynamic updates of the hidden state.

[0108] In some embodiments, as Figure 7 As shown, the method of determining the state of charge of the target battery cell according to the historical state parameters and the estimated value may include the following steps:

[0109] S151 : Determine, based on historical state parameters, a spatial correlation vector between a target battery cell and multiple battery cells, and a spatial kernel matrix between other battery cells.

[0110] The other battery cells are battery cells other than the target battery cell in the target battery.

[0111] Specifically, a method for determining the spatial kernel matrix between other battery cells based on historical state parameters may be to input the historical state parameters of other battery cells into a spatial kernel function to obtain a spatial kernel matrix. A method for determining the spatial correlation vector between a target battery cell and multiple battery cells based on historical state parameters may be to first determine a global spatial kernel matrix based on the historical state parameters and spatial kernel function of all battery cells in the target battery, and then extract the spatial correlation vector related to the target battery cell from the global spatial kernel matrix.

[0112] S152 : Determine the product of the estimated value, the spatial correlation vector, and the inverse matrix of the spatial kernel matrix as the state of charge of the target battery cell.

[0113] Specifically, the state of charge of the target battery cell can be expressed as: .in, Used to indicate the state of charge of the target battery cell, Used to represent the spatial correlation vector, , used to represent the set of electrical parameters of a battery cell; Used to represent the electrical parameters of the target battery cell; Used to represent the inverse matrix of the spatial kernel matrix, Used to express estimated values.

[0114] In this approach, the spatial correlation vector reflects the degree of historical correlation between the target battery cell and other cells, while the spatial kernel matrix captures the covariance relationship of the overall spatial structure. Fusion of this spatial information with the estimated values ​​facilitates robust inference of the target battery cell state even when some data is missing or measurements are inaccurate, thereby enhancing the reliability and robustness of SOC estimation in actual operating scenarios.

[0115] In some embodiments, before constructing the temporal kernel function and spatial kernel function of the target battery based on the historical state parameters, the method further includes detecting and removing abnormal parameters from the historical state parameters. Specifically, abnormal parameters from the historical state parameters can be detected by setting a reasonable threshold range, or by using statistical methods (such as mean and standard deviation, box plot analysis) to identify outliers that significantly deviate from the normal distribution, or by employing anomaly detection algorithms in machine learning (such as isolation forest and local anomaly factor) to detect abnormal parameters from the historical state parameters. After detecting abnormal parameters from the historical state parameters, they are removed to obtain the historical state parameters, thereby ensuring the reliability of the historical state parameters.

[0116] In this way, before constructing the time kernel function and space kernel function of the target battery, abnormal parameters in the historical state parameters are first detected and eliminated, which helps to eliminate abnormal values ​​caused by sensor failure, data acquisition errors or extreme working conditions, and prevent them from interfering with the subsequent model construction, thereby improving the accuracy and stability of kernel function modeling.

[0117] In some embodiments, detecting abnormal parameters in historical state parameters includes: detecting abnormal parameters in historical state parameters using an outlier factor detection method.

[0118] Specifically, the process of detecting abnormal parameters in historical state parameters using the outlier factor detection method may include the following steps:

[0119] (1) According to Calculate the first k The distance from each point in the neighborhood k Reachable distance. To represent data points o To data point p No. k Reachable distance, Used to represent points o No. k distance, Used to represent points o Arrive p distance.

[0120] (2) According to Calculate the local k Locally accessible density. Used to represent points p Part of k Local reachability density, point p No. k All point-to-point distances within the neighborhood p The average k The inverse of the reachable distance, To represent data points o To data point p No. k Reachable distance, Used to represent points p No. k The set of all points in the neighborhood, including the k Points at distance.

[0121] (3) According to Calculate the first k Local outlier factor. Among them, Used to represent points p No.k Local outlier factor, Used to represent points p No. k The set of all points in the neighborhood, Used to represent points p Part of k Locally accessible density, Used to represent points o Part of k Locally accessible density.

[0122] (4) According to the k The local outlier factor determines whether the data point is an abnormal point, that is, whether the historical state parameters of the data point are abnormal parameters.

[0123] Specifically, according to the k The local outlier factor determines whether the data point is an outlier by k When the local outlier factor is greater than the outlier factor threshold, the data point is determined to be an outlier. The outlier factor threshold is a preset value, for example, a default value or a value set according to actual conditions, for example, the outlier factor threshold is 1.

[0124] In some embodiments, the k The local outlier factor can be used to determine whether the data point is an outlier by directly u The local outlier factor of each data point is used to judge and determine the abnormal points to save computing resources.

[0125] In the above scheme, the outlier factor detection method is used to identify abnormal parameters in the historical state parameters, which helps to accurately find data points that deviate from the normal distribution and ensure the reliability of the data used in modeling.

[0126] In response to the large amount of data and nonlinear characteristics of battery charging and discharging, the present embodiment utilizes the LOF method to eliminate abnormal data, eliminating its impact on the regularity of characteristic quantities and reducing the runtime of the subsequently established temporal kernel function, spatial kernel function, continuous-time state-space model, global state-space model, and global integrated model (hereinafter collectively referred to as the model). Furthermore, a model is established using a spatiotemporal Gaussian process, modeling the battery pack SOC evolution as a spatiotemporal Gaussian process. Its mean function characterizes the expected SOC decay trend, and its covariance function captures the spatial coupling and temporal dynamics between cells. The infinite-dimensional Gaussian process is converted into a finite-dimensional Kalman state-space model, achieving real-time SOC estimation through recursive updating, reducing computational complexity. Furthermore, the system supports the online addition and removal of monitoring cells, maintaining prediction consistency through covariance matrix reconstruction and state projection, avoiding the recalculation of historical data, improving the accuracy of state-of-charge estimation, and improving the accuracy of the model.

[0127] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus the necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method.

[0128] The embodiment of the present application also provides a device for determining the state of charge of a battery, such as Figure 8 Said device comprises:

[0129] An acquisition module 701 is used to acquire historical state parameters of each battery cell of a target battery; the target battery includes a plurality of battery cells; a construction module 702 is used to construct a time kernel function and a space kernel function of the target battery based on the historical state parameters; the time kernel function is used to characterize the time correlation between the historical state parameters of a plurality of battery cells; the space kernel function is used to characterize the spatial correlation between a plurality of battery cells; the construction module 702 is also used to construct a global integration model based on the time kernel function and the space kernel function; a processing module 703 is used to perform hidden state recursive estimation on the global integration model based on the measured value of the real-time state of charge of the target battery cell in the target battery, and determine the estimated value of the state of charge of the target battery cell; a determination module 704 is used to determine the state of charge of the target battery cell based on the historical state parameters and the estimated value.

[0130] In some embodiments, the historical state parameters include at least sampling time, electrical parameters, and the spatial position of the battery cell in the target battery; the construction module 702 is specifically used to: construct a time kernel function of the target battery based on the sampling time and electrical parameters; and construct a spatial kernel function of the target battery based on the spatial position and electrical parameters.

[0131] In some embodiments, the electrical parameters include historical state of charge; a construction module 702 is specifically used to: determine the standard deviation of the historical state of charge of multiple battery cells; fit the electrical parameters according to the sampling time to obtain the time-correlated decay rate of the electrical parameters; construct a time kernel function of the target battery based on the standard deviation, the time-correlated decay rate and the time difference; the time difference is the time difference between any two sampling times.

[0132] In some embodiments, the time kernel function is: ;in, Used to represent the output of the time kernel function, Used to represent standard deviation, Used to express the time-dependent decay rate, Used to express time difference, is the base of natural logarithms.

[0133] In some embodiments, the electrical parameters include at least a historical state of charge; a construction module 702 is specifically used to: determine the standard deviation of the historical state of charge of multiple battery cells; construct a spatial kernel function of the target battery based on the standard deviation, spatial distance and electrical parameters; the spatial distance is the distance between the spatial positions of any two battery cells among the multiple battery cells.

[0134] In some embodiments, the spatial kernel function is: ;in, and are used to represent electrical parameters, and and corresponding to different battery cells among the plurality of battery cells; Used to represent the output of the spatial kernel function, for and The corresponding spatial distance between the two battery cells, Used to represent standard deviation, is the base of natural logarithms.

[0135] In some embodiments, the construction module 702 is specifically used to: perform rational spectral decomposition on the time kernel function to obtain hidden state variables of the battery cells in the target battery; construct a continuous-time state space model of the target battery based on the hidden state variables, and construct a global state space model based on the continuous-time state space model and the spatial kernel function; integrate the battery cells in the target battery based on the continuous-time state space model and the global state space model to obtain a global integrated model.

[0136] In some embodiments, the continuous-time state-space model is: ; ;in, Used to indicate the i The hidden state change rate of each battery cell, Used to express the time-dependent decay rate of electrical parameters, Used to indicate the i The hidden state variables of each battery cell, The Gaussian white noise used to represent the driving state equation is Used to indicate the i Each battery cell contains a time-dynamic state-of-charge component, Used to represent the standard deviation of the historical state of charge of multiple battery cells;

[0137] The global state space model is: ; ;in, Used to indicate After discretization k The set of discrete state vectors corresponding to the moment, Used to indicate k The discrete observation vector at time t, Used to represent discretization time interval; Used to represent process noise, is used to represent the measurement noise, It is used to represent the lower triangular matrix after Cholesky decomposition of the spatial kernel matrix. Used to represent the Kronecker product, is the base of natural logarithms.

[0138] In some embodiments, the global integration model is: ; ;in, Used to express the time-dependent decay rate of electrical parameters, Used to represent the standard deviation of the historical state of charge of multiple battery cells. Used to represent discretized time intervals, Used to represent process noise, is used to represent the measurement noise, It is used to represent the lower triangular matrix after Cholesky decomposition of the spatial kernel matrix. Used to represent the Kronecker product, Used to indicate The set of discrete state vectors at time , is the base of natural logarithms; for M dimensional global identity matrix, M is the number of battery cells in the target battery.

[0139] In some embodiments, the processing module 703 is specifically used to: use Kalman filtering to perform hidden state recursive estimation on the global integration model based on the measured value of the real-time state of charge of the target battery cell in the target battery to obtain the optimal posterior estimation value at the current moment; obtain the output matrix of the global integration model; and determine the product of the output matrix and the optimal posterior estimation value as the estimation value.

[0140] In some embodiments, the determination module 704 is specifically used to: determine the spatial correlation vector between the target battery cell and multiple battery cells, and the spatial kernel matrix between other battery cells based on historical state parameters; the other battery cells are battery cells in the target battery other than the target battery cell; and determine the product of the estimated value, the spatial correlation vector, and the inverse matrix of the spatial kernel matrix as the state of charge of the target battery cell.

[0141] In some embodiments, the battery state of charge determination device also includes a detection module; the detection module is used to detect abnormal parameters in the historical state parameters and eliminate the abnormal parameters before constructing the time kernel function and space kernel function of the target battery based on the historical state parameters.

[0142] In some embodiments, the detection module is specifically configured to detect abnormal parameters in historical state parameters using an outlier factor detection method.

[0143] In some embodiments, the target battery is a lithium battery.

[0144] For the description of the features in the embodiment corresponding to the device for determining the battery state of charge, reference can be made to the relevant description of the embodiment corresponding to the method for determining the battery state of charge, which will not be repeated here.

[0145] An embodiment of the present application further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any of the above-mentioned battery state of charge determination method embodiments.

[0146] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above-mentioned battery state of charge determination method embodiments when running.

[0147] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk, or an optical disk.

[0148] An embodiment of the present application further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the steps of any of the above-mentioned battery state of charge determination method embodiments.

[0149] An embodiment of the present application also provides another computer program product, including a non-volatile computer-readable storage medium, the non-volatile computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, implementing the steps in any of the above-mentioned battery state of charge determination method embodiments.

[0150] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0151] The above is a detailed introduction to the method, device, equipment, and readable storage medium for determining the state of charge of a battery provided by the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only intended to help understand the method and core ideas of the present application. It should be noted that for ordinary technicians in this technical field, without departing from the principles of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the claims of the present application.

Claims

1. A method for determining a battery state of charge, characterized in that: include: Acquiring historical state parameters of each battery cell of a target battery, wherein the target battery includes a plurality of battery cells; Constructing a time kernel function and a space kernel function of the target battery according to the historical state parameters; the time kernel function is used to characterize the time correlation between the historical state parameters of the multiple battery cells; The spatial kernel function is used to characterize the spatial correlation between the multiple battery cells; constructing a global integration model according to the temporal kernel function and the spatial kernel function; Performing hidden state recursive estimation on the global integrated model based on the measured real-time state of charge of the target battery cell in the target battery to determine an estimated value of the state of charge of the target battery cell; determining the state of charge of the target battery cell according to the historical state parameter and the estimated value; The constructing of a global integration model based on the time kernel function and the space kernel function includes: Performing rational spectral decomposition on the time kernel function to obtain hidden state variables of battery cells in the target battery; Constructing a continuous-time state-space model of the target battery according to the hidden state variables, and constructing a global state-space model according to the continuous-time state-space model and the spatial kernel function; The battery cells in the target battery are integrated according to the continuous-time state-space model and the global state-space model to obtain a global integrated model.

2. The determination method according to claim 1, characterized in that The historical state parameters include at least sampling time, electrical parameters, and the spatial position of the battery cell in the target battery; The constructing the time kernel function and the space kernel function of the target battery according to the historical state parameters includes: Constructing a time kernel function of the target battery according to the sampling time and the electrical parameters; A spatial kernel function of the target battery is constructed according to the spatial position and the electrical parameters.

3. The determination method according to claim 2, characterized in that: The electrical parameters include a historical state of charge; and constructing a time kernel function of the target battery according to the sampling time and the electrical parameters includes: determining a standard deviation of the historical states of charge of the plurality of battery cells; Performing fitting processing on the electrical parameter according to the sampling time to obtain a time-dependent decay rate of the electrical parameter; A time kernel function of the target battery is constructed according to the standard deviation, the time correlation decay rate, and the time difference; the time difference is the time difference between any two sampling times.

4. The determination method according to claim 3, characterized in that: The time kernel function is: ； in, Used to represent the output of the time kernel function, is used to express the standard deviation, is used to represent the time-dependent decay rate, is used to express the time difference, is the base of natural logarithms.

5. The determination method according to claim 2, characterized in that: The electrical parameters include at least a historical state of charge; and constructing the spatial kernel function of the target battery according to the spatial position and the electrical parameters includes: determining a standard deviation of the historical states of charge of the plurality of battery cells; A spatial kernel function of the target battery is constructed according to the standard deviation, the spatial distance, and the electrical parameter; the spatial distance is the distance between the spatial positions of any two battery cells among the plurality of battery cells.

6. The determination method according to claim 5, characterized in that: The spatial kernel function is: ； in, and are used to represent electrical parameters, and and corresponding to different battery cells among the plurality of battery cells; Used to represent the output of the spatial kernel function, for and The corresponding spatial distance between the two battery cells, is used to express the standard deviation, is the base of natural logarithms.

7. The determination method according to claim 1, characterized in that: The continuous-time state-space model is: ； ； in, Used to indicate the i The hidden state change rate of each battery cell, Used to express the time-dependent decay rate of electrical parameters, Used to indicate the i The hidden state variables of each battery cell, The Gaussian white noise used to represent the driving state equation is Used to indicate the i Each battery cell contains a time-dynamic state-of-charge component, Used to represent the standard deviation of the historical state of charge of multiple battery cells; The global state space model is: ； ； in, Used to indicate After discretization k The set of discrete state vectors corresponding to the moment, Used to indicate k The discrete observation vector at time t, Used to represent discretization time interval; Used to represent process noise, is used to represent the measurement noise, It is used to represent the lower triangular matrix after Cholesky decomposition of the spatial kernel matrix. Used to represent the Kronecker product, is the base of natural logarithms.

8. The determination method according to claim 7, characterized in that: The global integration model is: ； ； in, Used to express the time-dependent decay rate of electrical parameters, Used to represent the standard deviation of the historical state of charge of multiple battery cells. Used to represent discretized time intervals, Used to represent process noise, is used to represent the measurement noise, It is used to represent the lower triangular matrix after Cholesky decomposition of the spatial kernel matrix. Used to represent the Kronecker product, Used to indicate The set of discrete state vectors at time , is the base of natural logarithms; is an M-dimensional global identity matrix, where M is the number of battery cells in the target battery.

9. The determination method according to claim 1, characterized in that: The step of performing hidden state recursive estimation on the global integrated model based on the measured value of the real-time state of charge of the target battery cell in the target battery to determine the estimated value of the state of charge of the target battery cell includes: Using a Kalman filter to perform hidden state recursive estimation on the global integrated model based on the measured value of the real-time state of charge of the target battery cell in the target battery, to obtain an optimal posterior estimate at the current moment; Obtaining an output matrix of the global integration model; The product of the output matrix and the optimal a posteriori estimate is determined as an estimate.

10. The determination method according to claim 1, characterized in that: The step of determining the state of charge of the target battery cell according to the historical state parameter and the estimated value includes: Determining, based on the historical state parameters, a spatial correlation vector between the target battery cell and the plurality of battery cells, and a spatial kernel matrix between other battery cells; the other battery cells are battery cells in the target battery other than the target battery cell; The product of the estimated value, the spatial correlation vector, and the inverse matrix of the spatial kernel matrix is ​​determined as the state of charge of the target battery cell.

11. The determination method according to claim 1, characterized in that: Before constructing the time kernel function and the space kernel function of the target battery according to the historical state parameters, the method further includes: Detect abnormal parameters in the historical state parameters and eliminate the abnormal parameters.

12. The determination method according to claim 11, characterized in that: The detecting of abnormal parameters in the historical state parameters includes: An outlier factor detection method is used to detect abnormal parameters in the historical state parameters.

13. The determination method according to claim 1, characterized in that: The target battery is a lithium battery.

14. A device for determining a battery state of charge, characterized in that: include: an acquisition module, configured to acquire historical state parameters of each battery cell of a target battery; the target battery includes a plurality of battery cells; A construction module, configured to construct a time kernel function and a space kernel function of the target battery according to the historical state parameters; the time kernel function is used to characterize the time correlation between the historical state parameters of the plurality of battery cells; The spatial kernel function is used to characterize the spatial correlation between the multiple battery cells; The construction module is further used to construct a global integration model based on the time kernel function and the space kernel function; a processing module, configured to perform hidden state recursive estimation on the global integrated model based on a measured value of the real-time state of charge of a target battery cell in the target battery, and determine an estimated value of the state of charge of the target battery cell; a determination module, configured to determine the state of charge of the target battery cell based on the historical state parameter and the estimated value; The construction module is further configured to construct a global integration model based on the time kernel function and the space kernel function, including: Performing rational spectral decomposition on the time kernel function to obtain hidden state variables of battery cells in the target battery; Constructing a continuous-time state-space model of the target battery according to the hidden state variables, and constructing a global state-space model according to the continuous-time state-space model and the spatial kernel function; The battery cells in the target battery are integrated according to the continuous-time state-space model and the global state-space model to obtain a global integrated model.

15. An electronic device, characterized in that: include: memory for storing computer programs; A processor, configured to implement the steps of the method for determining the battery state of charge according to any one of claims 1 to 13 when executing the computer program.

16. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the steps of the method for determining the battery state of charge according to any one of claims 1 to 13 are implemented.

Images