Battery SOC estimation method and device based on Kalman filtering, electron and medium

Abstract

The invention belongs to the technical field of battery parameter estimation, and particularly relates to a battery SOC estimation method and device based on Kalman filtering, equipment and a medium. According to the scheme provided by the invention, aiming at the problems of small error and large influence caused by the voltage plateau period of the battery when SOC estimation is carried out on the lithium iron phosphate battery by a traditional Kalman filtering method, on the basis of a Kalman filtering algorithm, the improved formula is introduced, and the value of the gain G in the voltage plateau period in the improved formula is determined, so that the SOC estimation accuracy of the lithium iron phosphate battery is improved. The SOC estimation precision of the lithium iron phosphate battery is effectively improved, the maximum effectiveness of the battery can be ensured through accurate SOC estimation, the cost is reduced, and resource waste is prevented.

Description

Technical Field

[0001] The present application belongs to the technical field of battery parameter estimation, and in particular, relates to a battery SOC estimation method, device, equipment and medium based on Kalman filtering. Background Art

[0002] Battery state of charge (SOC) estimation is one of the core functions of the battery management system (BMS). Improving the accuracy of SOC estimation is of great significance to battery management.

[0003] At present, the commonly used battery SOC estimation methods include open circuit voltage method, ampere-hour integration method, Kalman filter method, and neural network method. In the field of battery SOC estimation based on equivalent circuit model, domestic and foreign scholars generally use various derivative filtering algorithms based on Kalman filter (KF), such as cubic Kalman filter (CKF) algorithm, unscented Kalman filter (UKF) algorithm, and extended Kalman filter (EKF) algorithm.

[0004] However, due to the existence of observation noise and other reasons, there is a certain deviation between the observed value of the terminal voltage and the true value. Especially for batteries such as lithium iron phosphate, even if there is only a small deviation in the terminal voltage during the voltage plateau period, when mapped to the battery SOC, there will be a large deviation in the SOC, which makes it very difficult to "correct" the prior estimate in the EKF, UKF and CKF algorithms, resulting in a decrease in the accuracy of SOC estimation.

[0005] Therefore, the above-mentioned traditional KF algorithms have limitations when estimating the SOC of batteries such as lithium iron phosphate. How to improve the accuracy of SOC estimation has become an urgent problem that needs to be solved. Summary of the invention

[0006] The present application provides a battery SOC estimation method, device, equipment and medium based on Kalman filtering, which solves the limitations of traditional KF algorithms when performing SOC estimation on batteries such as lithium iron phosphate, and improves the SOC estimation accuracy.

[0007] In a first aspect, an embodiment of the present application provides a battery SOC estimation method based on Kalman filtering, the method comprising: constructing a Kalman filter model, introducing an improved formula into the Kalman filter model, constructing a G-CKF model, the improved formula comprising a voltage plateau gain G, and the voltage plateau gain G is used to adjust the SOC estimation result; based on the G-CKF model and the observed current and observed voltage of the battery to be estimated, the battery SOC is estimated.

[0008] In a possible implementation, the improved formula is:

[0009]

[0010] Among them, K k is the Kalman gain, G represents the voltage plateau gain, and its value range is 0 <G<1, is the cross-covariance matrix, is the observation covariance matrix.

[0011] In a possible implementation, before performing battery SOC estimation based on the G-CKF model, the value of the voltage plateau period gain G is determined; determining the value of the voltage plateau period gain G includes: simulating a first preset operating condition and a second preset operating condition based on preset simulation software, the first preset operating condition and the second preset operating condition are used to determine the error of the battery SOC estimation; within the value range of the voltage plateau period gain G, determining the calculation results of the G-CKF model in the first preset operating condition and the second preset operating condition by a preset step length; based on the calculation results, determining the value of the voltage plateau period gain G.

[0012] In a possible implementation, an improved formula is introduced into the Kalman filter model, and constructing the G-CKF model includes: obtaining the initial state variables and the initial state difference covariance matrix in the Kalman filter model; based on the initial state variables and the initial state difference covariance matrix, using the mean weights and covariance weights corresponding to the volume points, calculating the sampled volume points and determining a first volume point set; based on the volume point set, making a preliminary prediction of the state variables and the state difference covariance matrix to determine the state propagation value; based on the state propagation value, performing volume point calculation again to determine a second volume point set; based on the second volume point and the improved formula, constructing the G-CKF model.

[0013] In a possible implementation, constructing the G-CKF model based on the second volume point and the improved formula includes: determining an observation covariance matrix and a cross-covariance matrix based on the second volume point; and constructing the G-CKF model based on the observation covariance matrix, the cross-covariance matrix and the voltage plateau gain.

[0014] In a possible implementation, the observation covariance matrix and the cross-covariance matrix are respectively: The observation covariance matrix:

[0015]

[0016] the cross-covariance matrix;

[0017]

[0018] Among them, y i,k / k -1 is the observed propagation value of the i-th Sigma point, is the estimated value of the observation at time k, is the prior estimate of the state at time k, R k represents the system observation noise v k The positive definite variance matrix of .

[0019] In one possible implementation, the battery SOC estimation based on the G-CKF model and the observed current and observed voltage of the battery to be estimated includes: determining system state variables based on the observed current and observed voltage of the battery to be estimated and the G-CKF model; and determining the battery SOC estimation value based on the system state variables.

[0020] In the second aspect, an embodiment of the present application provides a battery SOC estimation device based on Kalman filtering, the device comprising: a construction module, used to construct a Kalman filter model, introduce an improved formula into the Kalman filter model, and construct a G-CKF model, the improved formula includes a voltage plateau gain G, and the voltage plateau gain G is used to adjust the SOC estimation result; an estimation module, used to estimate the battery SOC based on the G-CKF model and the observed current and observed voltage of the battery to be estimated.

[0021] In a third aspect, an embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method described in the first aspect or any one of the implementation methods thereof is implemented.

[0022] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described in the first aspect or any one of the implementation methods thereof is implemented.

[0023] In a fifth aspect, an embodiment of the present application provides a computer program product. When the computer program product runs on an electronic device, the electronic device executes the method described in the first aspect or any one of the implementation methods thereof.

[0024] The solution provided by the present application constructs a Kalman filter model, introduces an improved formula into the Kalman filter model, and constructs a G-CKF model. The improved formula includes a voltage plateau gain G, and the voltage plateau gain G is used to adjust the SOC estimation result. The battery SOC is estimated based on the G-CKF model and the observed current and observed voltage of the battery to be estimated, which solves the limitations of traditional KF algorithms in SOC estimation of batteries such as lithium iron phosphate, and improves the accuracy of SOC estimation. Accurate SOC estimation can correctly display the current remaining power and expected cruising range of electric vehicles, helping drivers to make relevant preparations in advance and avoid vehicle failures due to insufficient power; accurate SOC estimation can also prevent the battery from being overcharged or over-discharged, which not only avoids potential system interruption risks and internal damage to the battery, but also reduces the risks of shortened life or explosion; in addition, accurate SOC estimation can ensure the maximum utility of the battery, reduce costs and prevent waste of resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0026] Figure 1 A schematic diagram of a flow chart of a battery SOC estimation method based on Kalman filtering provided in an embodiment of the present application;

[0027] Figure 2 A diagram showing the available capacity of a battery at different current rates provided in an embodiment of the present application;

[0028] Figure 3 A graph showing the HPPC cycle operating condition test current and battery terminal voltage response provided in one embodiment of the present application;

[0029] Figure 4 A fitting curve diagram of the OCV-SO relationship point provided in an embodiment of the present application;

[0030] Figure 5 A schematic diagram of the structure of a second-order RC model provided in an embodiment of the present application;

[0031] Figure 6 An EKF algorithm SOC estimation flow chart provided in an embodiment of the present application;

[0032] Figure 7 A comparison diagram of the estimated SOC value and the actual value under the FUDS cycle condition provided in an embodiment of the present application;

[0033] Figure 8 An SOC estimation error diagram under a FUDS cycle condition provided by an embodiment of the present application;

[0034] Fig. 9 A comparison diagram of the estimated SOC value and the actual value under the DST cycle condition provided in an embodiment of the present application;

[0035] Fig.10 An SOC estimation error diagram under a DST cycle condition provided by an embodiment of the present application;

[0036] Fig.11 A structural block diagram of a battery SOC estimation device based on Kalman filtering provided in one embodiment of the present application;

[0037] Fig.12 A schematic diagram of the structure of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0038] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0039] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0040] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0041] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0042] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0043] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0044] At present, the commonly used battery SOC estimation methods include open circuit voltage method, ampere-hour integration method, Kalman filter method, and neural network method. In the field of battery SOC estimation based on equivalent circuit models, domestic and foreign scholars generally use various derivative filtering algorithms based on Kalman filtering.

[0045] (1) KF algorithm

[0046] The KF algorithm was proposed by RE Kalman in the 1960s. This method estimates the state of the system while taking into account the input information of the system and the accompanying noise, and outputs the optimal state estimate of the system according to the principle of minimizing the mean square error. The KF algorithm is still widely used in fields such as real-time state estimation due to its advantages such as small amount of calculation, fast running speed, and small storage space requirements. However, the application scope of the KF algorithm is limited to a certain extent, and it is only applicable to linear systems.

[0047] (2) EKF algorithm

[0048] The battery system is a highly nonlinear system, and the traditional KF algorithm cannot be used directly to estimate its state. The EKF algorithm linearizes the nonlinear system based on the Taylor expansion principle, and then uses the standard KF algorithm to estimate the system state, so that the basic principles of the classic KF algorithm can be applied to battery SOC estimation.

[0049] Although the EKF algorithm is simple and effective for nonlinear systems, this method ignores the principle of second-order and higher-order terms through Taylor expansion to linearize nonlinear systems, which will inevitably produce truncation errors; when the system is highly nonlinear, excessive first-order approximation errors may cause filtering divergence; in addition, the EKF algorithm needs to calculate the Jacobian matrix, and if the system order is high, the amount of calculation is large.

[0050] The main implementation steps of the EKF algorithm are as follows:

[0051] Step 1, Initialization: Initialize the state variables and the error covariance matrix. When the initial value is unknown, generally take the smaller value and the larger value. Step 2, Prediction: Make a preliminary prediction of the system state variable observation variables and the error covariance matrix. Step 3, Update: Calculate the Kalman gain and perform observation updates on the system state variables and the error covariance matrix. Step 4, Loop: Substitute the system's observation data and repeat steps 2 to 3. By performing one-step prediction and one-step correction, the optimal estimation of the system state is achieved.

[0052] (3) UKF algorithm

[0053] The UKF algorithm was proposed by scholars Julier, Uhlmann and others from the University of Oxford in the 1990s. The algorithm is also a derivative of the KF algorithm. Its core idea is to approximate the probability density distribution of a nonlinear system through the principle of unscented transformation (UT), rather than directly linearizing the nonlinear system, so it has stronger processing capabilities for nonlinear systems. Compared with EKF, the UKF algorithm does not need to linearize the nonlinear system and does not need to ignore high-order terms, which effectively improves the adaptability and calculation accuracy of nonlinear distribution. This method can maintain the characteristics of the Gaussian distribution during the nonlinear mapping process, so that the approximate accuracy of the Gaussian distribution reaches the third order, and the approximate accuracy of the non-Gaussian distribution can also be achieved at least the second order.

[0054] The core idea of ​​the UKF algorithm is to approximate the probability density distribution of a nonlinear system through the Unscented Transformation (UT) principle, rather than directly linearizing the nonlinear system, so it has stronger processing capabilities for nonlinear systems. The core of the UT transformation principle is to abandon the idea of ​​directly propagating the system state variables through the nonlinear system, and construct a Sigma point set (composed of 2n+1 sampling points) that is consistent with the mean and error covariance of the system state variables at the previous moment, substitute the point set into the nonlinear system for the propagation of the Sigma point set, and then reversely calculate the approximate value of the original system state variables after propagation based on the criterion of keeping the mean and variance consistent, so as to estimate the state variables of the system at the current moment.

[0055] The main implementation steps of the UKF algorithm are as follows:

[0056] Step 1, initialization: initialize the state variables and the difference covariance matrix. When the initial value is unknown, generally take a smaller value and take a larger value. Step 2, construct the Sigma point set: perform Sigma point calculation on the posterior estimate at time k-1 to obtain the Sigma point set. Step 3, prediction: 1) make a preliminary prediction of the system state variables and state covariance matrix; 2) perform UT transformation to obtain a new Sigma point set. Estimate the observed variables, observation covariance matrix, and mutual covariance matrix. Step 4, update:

[0057] Calculate the Kalman gain and observe and update the system state variables and error covariance matrix. Step 5, loop: Substitute the system observation data and repeat steps 2 to 4. By performing one-step prediction and one-step correction, the optimal estimation of the system state is achieved.

[0058] (4) CKF algorithm

[0059] The CKF algorithm was proposed by Arasaratnam and Haykin in 2009. It is also developed based on the framework of the KF algorithm and is also a derivative algorithm of the KF algorithm. Like the UKF algorithm, the CKF algorithm also approximates the probability density distribution of the nonlinear system, and the mean and variance remain unchanged, but the CKF algorithm uses the spherical-radial Cubature Rule (SRCR) to select the sampling point set, which can be linearly extended to multi-dimensional systems. Therefore, the CKF algorithm is more suitable for state estimation of high-dimensional systems than the UKF algorithm. At the same time, the 2n symmetrical sampling points selected by the CKF algorithm through the SRCR criterion all have equal weights, avoiding the situation where the algorithm is difficult to correct the estimation deviation caused by the intermediate sampling points similar to the UKF. In addition, due to the smaller number of sampling points and simpler weight calculation, CKF is superior to UKF in terms of computational efficiency, especially when dealing with high-dimensional system state estimation, its computational efficiency and stability are better.

[0060] The EKF algorithm is used to estimate the SOC of lithium-ion batteries. This method performs a first-order Taylor series expansion on nonlinear functions and ignores quadratic and higher-order terms. EKF sets the noise to a fixed value, but in actual systems, measurement noise and observation noise are not fixed. Therefore, some scholars have proposed the Adaptive Extended Kalman Filter (AEKF) algorithm, which updates the noise covariance through adaptive iteration to reduce the impact of noise errors on battery SOC estimation. In addition, the linearization link in the EKF inevitably brings linearization errors to the system; at the same time, it is also necessary to calculate the Jacobian matrix. If the system order is high, the amount of calculation is large. Therefore, the UKF algorithm is used to estimate the battery SOC. This method does not require the system to be linearized, but instead realizes the transfer of the mean and variance of the nonlinear function through the UT transformation. The estimation accuracy is improved compared to the EKF, and there is no need to calculate the Jacobian matrix, so the computational complexity is simplified. The CKF algorithm is used to estimate SOC and compared with EKF and UKF. The results show that CKF is superior to EKF and UKF in terms of estimation accuracy and robustness.

[0061] In order to solve the problem of "small error, big impact" caused by the battery voltage plateau period when the traditional Kalman Filter (KF) method estimates the SOC of lithium iron phosphate batteries, a SOC estimation method considering the voltage plateau period gain is proposed based on the Cubature Kalman Filter (CKF) algorithm, which is called G-CKF.

[0062] Figure 1A schematic flow chart of a battery SOC estimation method based on Kalman filtering provided in an embodiment of the present application, the method comprising:

[0063] S110, constructing a Kalman filter model, introducing an improved formula into the Kalman filter model, and constructing a G-CKF model, the improved formula including a voltage plateau gain G, and the voltage plateau gain G is used to adjust the SOC estimation result.

[0064] In a possible implementation, the improved formula is:

[0065]

[0066] Among them, K k is the Kalman gain, G represents the voltage plateau gain, and its value range is 0 <G<1, is the cross-covariance matrix, is the observation covariance matrix.

[0067] In a possible implementation, based on the G-CKF model, before performing battery SOC estimation, the value of the voltage plateau period gain G is determined; determining the value of the voltage plateau period gain G includes: based on a preset simulation software, simulating a first preset operating condition and a second preset operating condition, the first preset operating condition and the second preset operating condition are used to determine the error of the battery SOC estimation; within the value range of the voltage plateau period gain G, by a preset step length, determining the calculation results of the G-CKF model in the first preset operating condition and the second preset operating condition; based on the calculation results, determining the value of the voltage plateau period gain G.

[0068] Specifically, the value of the voltage plateau gain G has a key impact on the performance of the G-CKF algorithm. The larger the G, the smaller the terminal voltage observation value y. k The higher the credibility of G, the smaller the terminal voltage observation value y k The lower the credibility, when G = 1, the credibility of the terminal voltage observation value is not reduced, and G-CKF is equivalent to CKF; when G = 0, the credibility of the terminal voltage observation value is 0, and the correction ability of the prior estimate is lost. At this time, G-CKF is equivalent to the ampere-hour integration method. Therefore, choosing a reasonable G value has a great impact on the performance of the G-CKF algorithm.

[0069] Exemplarily, the present application is based on a preset simulation software (Matlab simulation software), and takes the first preset operating condition (FUDS cycle condition) and the second preset operating condition (DST cycle condition) as simulation conditions, and uses the G-CKF algorithm with different G values ​​to perform SOC estimation simulation on the battery. The voltage plateau gain G value is optimized through the simulation results (the preset step length of the present application is set to 0.1), and the MAE value and RMSE value of the SOC estimation value and the actual value (the battery SOC value obtained by the ampere-hour integration method under laboratory conditions, which can be approximately considered to be the true value) are used as optimization evaluation indicators, and the value of G is determined through comprehensive judgment.

[0070] In a possible implementation, a Kalman filter model is constructed, an improved formula is introduced into the Kalman filter model, and a G-CKF model is constructed, including: improving the integral of the Gaussian nonlinear function in the traditional volumetric Kalman filter model by the improved formula; the integral of the Gaussian nonlinear function is:

[0071]

[0072] Among them, f(x) is a nonlinear function, I(f) is the Gaussian nonlinear function integral, Rn is the n-dimensional integration domain, and x is the state vector.

[0073] In a possible implementation, after improving the integral of the Gaussian nonlinear function in the traditional volumetric Kalman filter model by improving the formula, the construction of the G-CKF model also includes: converting the integral of the improved Gaussian nonlinear function into a spherical radial integral, converting the state vector x in the Cartesian coordinate system into a radius r and a direction vector y to obtain a double integral; converting the double integral into a radial integral and a spherical integral; determining the spherical radial volume criterion formula by approximate conversion of the radial integral and the spherical integral; and determining the G-CKF model based on the spherical radial volume criterion formula.

[0074] Furthermore, we convert equation (1) into the spherical radial integral form and convert the state vector x in the Cartesian coordinate system into the radius r and the direction vector y, that is, let x = ry, y T y=1,r∈[0,+∞), so x T x=r 2 . Formula (1) can be expressed as the following double integral:

[0075]

[0076] Where r is the radius and y is the direction vector. Equation (2) can be further written in the form of radial integral and spherical integral:

[0077] Radial integration:

[0078] Surface integral:

[0079] The radial integral (3) can be used to r The spherical surface integral (4) can be obtained by using the Gauss-Hermitian approximation of m points. s The spherical volume criterion of each point is approximated to obtain the following formula:

[0080]

[0081] Among them, a i 、b j For the corresponding weight, combining the above two equations, we can get (m r ×m s) The spherical radial volume criterion formula is:

[0082]

[0083] Take m r =1,m s =2n, where n is the dimension of the state vector. Equation (7) can be further expressed as the following equation with a mean of zero and a covariance of the identity matrix:

[0084]

[0085] Where m is the number of volume points, ωj=1 / m represents the corresponding weight of the volume point, is the volume point, [δ] j represents the jth volume point, as shown in formula (9):

[0086]

[0087] Select 2n sampling points according to the SRCR criterion. For nonlinear systems, select the following 2n volume points according to the SRCR criterion:

[0088]

[0089] In the formula, and P k-1 The posterior estimate and error covariance matrix of the state variables at time k-1, n is the state dimension, e is i Represents the i-th column of the n×n identity matrix.

[0090] The mean weights and covariance weights corresponding to the above 2n volume points are:

[0091]

[0092] Furthermore, the G-CKF model is determined according to formula (11).

[0093] The steps to implement the CKF model algorithm are as follows:

[0094] Step 1, Initialization:

[0095] For state variables Initialize with the state difference covariance matrix P0. When the initial value is unknown, Generally, a smaller value is taken and a larger value is taken for P0.

[0096]

[0097] Step 2, calculate the volume point set:

[0098] According to equations (10) and (11), the posterior estimate at time k-1 is Perform volume point calculation to obtain the first volume point set in Represents the i-th volume point.

[0099] Step 3, prediction:

[0100] 1) System state variables State covariance matrix Make a preliminary forecast:

[0101]

[0102] In the formula, x i,k / k-1 is the state propagation value of the i-th Sigma point, is the prior estimate of the state at time k, Q k-1 represents the system process noise w k-1 The non-negative definite variance matrix of .

[0103] 2) Yes Perform volume point calculation to obtain the second volume point set And the observed variables Observation covariance matrix Cross-covariance matrix Make an estimate:

[0104]

[0105] In the formula, y i,k / k-1 is the observed propagation value of the i-th Sigma point, is the estimated value of the observation at time k, R k represents the system observation noise v k The positive definite variance matrix of .

[0106] Step 4, Update:

[0107] Calculate the Kalman gain Kk , for the system state variables Error covariance matrix P k To perform an observation update:

[0108]

[0109] Step 5, loop:

[0110] Substitute the observed data of the system and repeat steps 2 to 4. By performing one-step prediction and one-step correction in a continuous cycle, the optimal estimate of the system state can be achieved.

[0111] The G-CKF model algorithm calculates the Kalman gain K k , based on formula (20), the gain G is introduced to obtain:

[0112]

[0113] S120 , estimating the battery SOC based on the G-CKF model and the observed current and observed voltage of the battery to be estimated.

[0114] Specifically, after the above steps, based on the G-CKF model and the observed current and observed voltage of the battery to be estimated, the final result is the system state variable The state variable value includes related parameters, one of which is the estimated SOC value of the battery.

[0115] The Kalman filter model is a recursive algorithm widely used to estimate the state of a system, and is particularly suitable for dealing with linear dynamic systems containing noise. It mainly includes the following features:

[0116] (1) Linearity

[0117] The standard Kalman filter assumes that the system dynamics and observation model are linear. If the system is nonlinear, variants such as the Extended Kalman Filter (EKF) or the Unscented Kalman Filter (UKF) can be used.

[0118] (2) Gaussian noise

[0119] The system process noise and observation noise are assumed to be Gaussian distributed. This allows the Kalman filter to estimate the system state through the mean and covariance.

[0120] (3) Recursive estimation

[0121] The filter recursively calculates the state estimate and covariance matrix, does not need to store historical data, and is suitable for real-time processing.

[0122] (4) Optimality

[0123] Under the assumption of Gaussian noise and a linear system model, the Kalman filter provides optimal linear state estimation.

[0124] SOC (State of Charge) estimation is a key task in battery management systems. The Kalman filter model has several important applications in SOC estimation:

[0125] (1) Dynamic system modeling

[0126] The SOC changes of the battery can be described by a linear model, such as by a battery discharge or charge model. A Kalman filter can be used to dynamically track these changes.

[0127] (2) Noise processing

[0128] Battery voltage and current measurements are often subject to noise. Kalman filters can effectively filter these noises, providing more accurate SOC estimates.

[0129] (3) Recursive Update

[0130] Since the battery state is constantly changing over time, the recursive nature of the Kalman filter makes it well suited for updating the SOC estimate in real time.

[0131] (4) Information fusion

[0132] The Kalman filter can fuse data from different sensors (such as current sensors, voltage sensors, etc.) to improve the accuracy of SOC estimation.

[0133] Based on the G-CKF model, the battery SOC is estimated. The "MAE value and RMSE value" are calculated based on the results and the actual battery SOC, and the accuracy of the SOC estimation is determined by the "MAE value and RMSE value", thereby illustrating the accuracy of the model. Furthermore, in order to verify the accuracy of the battery SOC estimation based on the G-CKF model, this application uses comparative tests to prove that the battery SOC estimation based on the G-CKF model is more accurate than other methods.

[0134] Battery experiment

[0135] (1) Battery available capacity test

[0136] The nominal capacity of a lithium-ion battery refers to the maximum amount of electricity that the battery can release under specific conditions (usually the standard charge and discharge method specified by the battery manufacturer). However, in actual battery operation, the charge and discharge methods do not always follow the standard conditions specified by the battery manufacturer; at the same time, due to factors such as battery materials and production processes, there may be inconsistencies in the available capacity even between single cells of the same type and batch; in addition, the setting of capacity parameters has a great impact on the subsequent hybrid pulse power characterization (HPPC) cycle condition test, parameter identification research, and charge state estimation research in this article. Therefore, it is necessary to test the available capacity of the battery. The available capacity of the battery at different current rates is as follows: Figure 2 shown.

[0137] (2) HPPC cycle test

[0138] HPPC test is to charge and discharge the battery by applying a specific pulse current to obtain the terminal voltage response of the battery. The original intention was to evaluate and test the dynamic power performance of the battery. Since this method can analyze and calculate the battery current and terminal voltage response to obtain the parameters of the lithium-ion battery equivalent circuit model, this method was later widely used for offline parameter identification of battery equivalent models. HPPC cycle test current and battery terminal voltage response Figure 3 shown.

[0139] (3) OCV-SOC curve acquisition

[0140] It can be seen from the experimental steps and processes of the HPPC cycle test that before each HPPC test, the battery has been standing for a long enough time, and it can be considered that the battery polarization effect has disappeared and the battery has reached a fully balanced state. At this time, the battery terminal voltage can be considered to be the battery open circuit voltage (Open Circuit Voltage, OCV). Since subsequent battery parameter identification and SOC estimation research need to rely on the OCV-SOC relationship curve, it is necessary to fit the obtained OCV-SOC relationship points. This paper uses the MATLAB fitting toolbox Curve Fitting Tool to perform polynomial fitting on the above curve. In order to determine the order of the fitted polynomial function, the adjusted R2 (Adjusted R-squared) is used as a measure of the fitting effect. Taking into account the fitting effect and computational cost, this application uses a 7th-order polynomial to fit the OCV-SO relationship points. The fitting curve is as follows Figure 4 shown.

[0141] Battery Modeling

[0142] The equivalent circuit model simulates the performance and behavior of the battery under actual working conditions by using a combination of basic circuit elements (such as resistors and capacitors). This model can intuitively represent the various characteristics of the battery's working state. Each parameter of the model can be obtained through identification, and then the battery SOC can be estimated through the battery model. The second-order RC model not only simulates the resistance effect of the battery, but also distinguishes between the electrochemical polarization and concentration polarization phenomena of the battery. The model has high accuracy and parameter identification is relatively easy, taking into account the accuracy of the model and the difficulty of parameter identification. This patent selects the second-order RC model as the research basis. The structure of the second-order RC model is as follows: Figure 5 As shown in the figure, U OCV represents the open circuit voltage of the battery, R0 represents the ohmic internal resistance of the battery, R1 and R2 represent the electrochemical polarization internal resistance and concentration polarization internal resistance of the battery respectively, C1 and C2 represent the electrochemical polarization capacitance and concentration polarization capacitance of the battery respectively, U L Represents the terminal voltage of the battery, I L Indicates the current when the battery is discharging.

[0143] Parameter identification

[0144] (1) RLS algorithm

[0145] The Least Squares (LS) method is a method for finding the optimal solution of a function by minimizing the sum of squared errors. This method is widely used in various scenarios due to its advantages such as simple calculation and fast convergence. However, the traditional least squares method needs to be completely recalculated at each iteration. As the matrix dimension increases, the computational time cost of matrix inversion increases sharply, limiting the online applicability of the LS algorithm. To address this problem, scholars proposed the Recursive Least Squares (RLS) method.

[0146] (2)FFRLS algorithm

[0147] Although the RLS algorithm is simple, efficient, and has a small amount of calculation, it is very suitable for online parameter identification, but it also has certain limitations. The data memory length of the RLS algorithm is infinite and the new and old data are given equal weights. As time goes by, the accumulation of old data causes the algorithm to have poor tracking performance. At the same time, the positive covariance matrix P(k) in the algorithm will continue to approach zero as the number of recursions increases, causing the correction factor K(k) to also approach zero, resulting in the algorithm losing its correction ability, which is called the "data saturation" phenomenon. In order to improve the tracking performance of the algorithm and overcome the "data saturation" problem, some scholars have introduced the forgetting factor λ based on the RLS algorithm to weaken the weight of historical data. This method is called the Forgetting Factor Recursive Least Squares (FFRLS).

[0148] (3) Recursive least squares algorithm with limited memory

[0149] The limited memory recursive least squares algorithm can theoretically effectively solve the "data saturation" problem of the RLS algorithm and improve the tracking ability of the algorithm, but the limited memory rule will make the algorithm less stable. Multi-information input uses the data of the most recent multiple sampling moments as the input of the model, and more new information data as the basis for reasoning, which can theoretically effectively improve the stability of the algorithm.

[0150] The battery model is identified using online parameter identification based on the multi-innovation recursive least squares method with limited memory. The battery parameters are obtained, laying the foundation for the subsequent battery SOC estimation.

[0151] Comparative Analysis of SOC Estimation Algorithms

[0152] (1) SOC estimation based on EKF algorithm

[0153] The EKF algorithm linearizes the nonlinear system based on the Taylor expansion principle, and then uses the standard KF algorithm to estimate the state of the system, so that the basic principles of the classic KF algorithm can be applied to battery SOC estimation.

[0154] Although the performance of the EKF algorithm does not completely depend on the initial value setting, choosing an initial value that is closer to the actual state can effectively shorten the algorithm's recursive time and improve its dynamic response performance. There is no unified selection standard for the initial values ​​of these parameters, and they are usually set based on practical tests and experience. Although the EKF algorithm is simple and effective for nonlinear systems, this method will inevitably produce truncation errors when linearizing nonlinear systems through Taylor expansion, ignoring the principle of second-order and higher-order terms; when the system is highly nonlinear, excessive first-order approximation errors may cause filter divergence; in addition, the EKF algorithm needs to calculate the Jacobian matrix, and if the system order is high, the amount of calculation is large. Figure 6 This is the SOC estimation flow chart of the EKF algorithm.

[0155] (2) SOC estimation based on G-CKF algorithm

[0156] For batteries such as lithium iron phosphate, the OCV-SOC curve of lithium iron phosphate batteries has two voltage plateaus. In the voltage plateau period, even if the battery SOC changes greatly, the open circuit voltage changes very little. This paper divides the battery OCV-SOC correspondence into six intervals, namely SOC = 0% ~ 10%, 10% ~ 40%, 40% ~ 70%, 70% ~ 80%, 80% ~ 95%, 95% ~ 100%, and calculates how much the open circuit voltage OCV changes when the SOC changes by 1% in each interval. The calculation results are shown in Table 1.

[0157] Table 1 OCV changes caused by 1% SOC change in different SOC intervals

[0158]

[0159]

[0160] It can be seen from Table 1 that for the same SOC change, the OCV change caused by the voltage plateau period is much smaller than that caused by the non-voltage plateau period; conversely, even if the OCV changes only slightly during the voltage plateau period, the SOC will change greatly when mapped to the battery SOC.

[0161] Due to the existence of observation noise and other reasons, there is a certain deviation between the observed value and the true value of the terminal voltage. According to the above analysis, for batteries such as lithium iron phosphate, even if there is only a slight deviation in the terminal voltage during the voltage plateau period, when mapped to the battery SOC, there will be a large deviation in the SOC, which makes it very difficult to "correct" the prior estimate in the EKF, UKF and CKF algorithms, resulting in a decrease in the accuracy of SOC estimation. Therefore, the above-mentioned traditional KF algorithms have limitations when estimating the SOC of batteries such as lithium iron phosphate. In response to the above-mentioned "small error, big impact" problem, this paper proposes a volumetric Kalman filter algorithm that takes the voltage plateau period into consideration. Its principle is to introduce a voltage plateau gain on the basis of the CKF algorithm, and to reduce the credibility of the terminal voltage observation value during the voltage plateau period to reduce the impact of the terminal voltage observation error. This article calls this algorithm G-CKF, where G represents the voltage plateau gain, and its value range is 0. <G<1。

[0162] (3) G value optimization

[0163] First, according to G-CKF, determine the value of G (G value optimization). The value of the voltage plateau gain G has a key impact on the performance of the G-CKF algorithm. The larger G represents the terminal voltage observation value yk The higher the credibility of G, the smaller the terminal voltage observation value y k The lower the credibility. When G = 1, the credibility of the terminal voltage observation value is not reduced, and G-CKF is equivalent to CKF; when G = 0, the credibility of the terminal voltage observation value is 0, and the correction ability of the prior estimate is lost. At this time, G-CKF is equivalent to the ampere-hour integration method. Therefore, choosing a reasonable G value has a great impact on the performance of the G-CKF algorithm.

[0164] This experiment is based on Matlab simulation software. The Full Urban Driving Schedule (FUDS) cycle and the Dynamic Stress Test (DST) cycle are used as simulation conditions. The G-CKF algorithm with different G values ​​is used to simulate the SOC estimation of the battery. The voltage plateau gain G value is optimized through the simulation results (the optimization step length is set to 0.1 in this paper), and the MAE value and RMSE value of the SOC estimation value and the actual value (the battery SOC value obtained by the ampere-hour integration method under laboratory conditions, which can be approximately considered as the true value) are used as optimization evaluation indicators. In order to simulate the actual situation that the SOC initial value set by the estimation algorithm generally has an error with the actual initial value, this paper sets the SOC initial value to 70% (the actual initial value is 100%). The optimization results are shown in Tables 2 and 3 respectively.

[0165] Table 2 Optimization results of platform gain G under FUDS cycle conditions

[0166] G-Value 0 0.1 0.2 0.3 MAE value (%) 12.19 7.03 6.00 5.03 RMSE value (%) 17.03 12.01 8.54 6.81 G-Value 0.4 0.5 0.6 0.7 MAE value (%) 4.64 2.25 1.07 1.05 RMSE value (%) 6.00 3.84 2.41 2.19 G-Value 0.8 0.9 1.0 MAE value (%) 1.07 1.11 1.16 RMSE value (%) 2.08 2.05 2.08

[0167] Table 3 Optimization results of platform gain G under DST cycle conditions

[0168] G-Value 0 0.1 0.2 0.3 MAE value (%) 12.57 6.92 5.53 4.37 RMSE value (%) 17.40 12.62 8.12 6.02 G-Value 0.4 0.5 0.6 0.7 MAE value (%) 2.48 1.13 0.99 0.93 RMSE value (%) 4.03 2.98 2.59 2.33 G-Value 0.8 0.9 1.0 MAE value (%) 0.91 0.99 0.99 RMSE value (%) 2.19 2.60 2.51

[0169] According to the optimization results in Table 2, it can be seen that the minimum value of MAE under the FUDS cycle condition occurs when the voltage plateau gain G = 0.7, and its value is 1.05%, which means that the SOC estimation value under this condition is generally closest to the true value, while the minimum value of RMSE occurs when the voltage plateau gain G = 0.9, and its value is 2.05%, which means that the SOC estimation value under this condition is generally closest to the true value and has a small fluctuation. When G = 0.8, although neither the MAE value nor the RMSE value is the minimum value, both are suboptimal values ​​and very close to the minimum value. According to the optimization results in Table 5-3, it can be seen that the MAE and RMSE values ​​under the FUDS cycle condition both appear when the voltage plateau gain G = 0.8, and their values ​​are 0.91% and 2.19% respectively, which means that the SOC estimation value under this condition is generally closest to the true value and has a small fluctuation.

[0170] Taking into account the MAE and RMSE values ​​of SOC estimation under two working conditions, this paper chooses to set the voltage plateau gain to G = 0.8.

[0171] (4) FUDS cycle condition simulation and comparison

[0172] Under the FUDS cycle condition, the SOC estimation simulation results of the EKF, UKF, CKF and G-CKF algorithms are as follows: The SOC estimation value and the actual value under the FUDS cycle condition are as follows: Figure 7 As shown in Figure 2, the SOC estimation error under the FUDS cycle condition is as follows: Figure 8 As shown, the MAE value and RMSE value error indicators of SOC estimation are shown in Table 4.

[0173] Table 4 Comparison of error indicators of different SOC estimation algorithms under FUDS cycle conditions

[0174] algorithm EKF UKF CKF G-CKF MAE value (%) 2.12 1.36 1.16 1.07 RMSE value (%) 3.14 2.25 2.08 2.08

[0175] (5) Simulation and comparison of DST cycle conditions

[0176] Under the DST cycle condition, the SOC estimation simulation results of the EKF, UKF, CKF and G-CKF algorithms are as follows: The SOC estimation value and the actual value under the DST cycle condition are as follows Fig. 9 As shown in Figure 2, the SOC estimation error under the DST cycle condition is as follows: Fig.10 As shown, the MAE and RMSE values ​​of SOC estimation are shown in Table 5.

[0177] Table 5 Comparison of error indicators of different SOC estimation algorithms under DST cycle conditions

[0178]

[0179]

[0180] Under the two working conditions, the volatility of the estimation results of the EKF algorithm, the maximum absolute value of the error, and the overall error are the largest, followed by the UKF algorithm; the volatility and the maximum absolute value of the error of the estimation results of the CKF algorithm and the G-CKF algorithm are relatively close, but the overall error of the G-CKF algorithm is significantly smaller than that of the CKF algorithm. Specifically, the MAE value and RMSE value of the SOC estimation of the G-CKF algorithm are about 52.24% and 39.51%, 22.66% and 8.16%, and 7.92% and 6.38% lower than the average of EKF, UKF, and CKF, respectively, which means that the G-CKF algorithm has higher accuracy and less error volatility for the SOC estimation of lithium iron phosphate batteries than the other three algorithms.

[0181] In summary, compared with the EKF, UKF and CKF algorithms, the G-CKF algorithm proposed in this paper has better accuracy for SOC estimation of lithium iron phosphate batteries.

[0182] The solution provided in the present application aims to solve the problem of "small error, big impact" caused by the battery voltage plateau period when the traditional Kalman filter method estimates the SOC of lithium iron phosphate batteries. On the basis of the volumetric Kalman filter algorithm, an improved formula is introduced and the value of the voltage plateau period gain G in the improved formula is determined to effectively improve the SOC estimation accuracy of lithium iron phosphate batteries. Accurate SOC estimation can ensure the maximum utility of the battery, reduce costs and prevent waste of resources.

[0183] Fig.11 This is a structural block diagram of a battery SOC estimation device based on Kalman filtering provided in one embodiment of the present application. For the sake of convenience, only the part related to the embodiment of the present application is shown. Fig.11 The battery SOC estimation device 1100 includes a construction module 1101 and an execution module 1102 .

[0184] In one implementation, the apparatus 1100 may be used to implement the above Figure 1 For example, the construction module 1101 is used to implement S110, and the execution module 1102 is used to implement S120.

[0185] The solution provided in the present application aims to solve the problem of "small error, big impact" caused by the battery voltage plateau period when the traditional Kalman filter method estimates the SOC of lithium iron phosphate batteries. On the basis of the volumetric Kalman filter algorithm, an improved formula is introduced and the value of the voltage plateau period gain G in the improved formula is determined to effectively improve the SOC estimation accuracy of lithium iron phosphate batteries. Accurate SOC estimation can ensure the maximum utility of the battery, reduce costs and prevent waste of resources.

[0186] Fig.12 This is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Fig.12 As shown, the electronic device 12 of this embodiment includes: at least one processor 120 ( Fig.12 Only one is shown in the figure) a processor, a memory 121, and a computer program 122 stored in the memory 121 and executable on the at least one processor 120, and when the processor 120 executes the computer program 122, the steps in any of the above method embodiments are implemented.

[0187] The electronic device 12 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The electronic device may include, but is not limited to, a processor 120 and a memory 121. Those skilled in the art will appreciate that Fig.12 It is only an example of the electronic device 12 and does not constitute a limitation on the electronic device 12. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, it may also include input and output devices, network access devices, etc.

[0188] The processor 120 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0189] In some embodiments, the memory 121 may be an internal storage unit of the electronic device 12, such as a hard disk or memory of the electronic device 12. In other embodiments, the memory 121 may also be an external storage device of the electronic device 12, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 12. Further, the memory 121 may also include both an internal storage unit of the electronic device 12 and an external storage device. The memory 121 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program. The memory 121 may also be used to temporarily store data that has been output or is to be output.

[0190] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0191] An embodiment of the present application also provides an electronic device, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor implements the steps of any of the above-mentioned method embodiments when executing the computer program.

[0192] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0193] An embodiment of the present application provides a computer program product. When the computer program product runs on a mobile terminal, the mobile terminal can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0194] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device that can carry the computer program code to the camera / electronic device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electric carrier signal, a telecommunication signal, and a software distribution medium. For example, a USB flash drive, a mobile hard disk, a disk or an optical disk.

[0195] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0196] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0197] In the embodiments provided in the present application, it should be understood that the disclosed devices / electronic devices and methods can be implemented in other ways. For example, the device / electronic device embodiments described above are merely schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0198] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0199] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A battery SOC estimation method based on Kalman filtering, characterized in that: The method comprises: Constructing a Kalman filter model, introducing an improved formula into the Kalman filter model, and constructing a G-CKF model, wherein the improved formula includes a voltage plateau gain G, and the voltage plateau gain G is used to adjust the SOC estimation result; The battery SOC is estimated based on the G-CKF model and the observed current and observed voltage of the battery to be estimated.

2. The method according to claim 1, characterized in that The improved formula is: Among them, K k is the Kalman gain, G represents the voltage plateau gain, and its value range is 0 <G<1, is the cross-covariance matrix, is the observation covariance matrix.

3. The method according to claim 2, characterized in that Before estimating the battery SOC based on the G-CKF model, determining the value of the voltage plateau gain G; Determining the value of the voltage plateau gain G includes: Based on preset simulation software, simulating a first preset operating condition and a second preset operating condition, wherein the first preset operating condition and the second preset operating condition are used to determine an error in battery SOC estimation; Within the value range of the voltage plateau gain G, by preset step length, determining the calculation results of the G-CKF model under the first preset working condition and the second preset working condition; Based on the calculation result, a value of the voltage plateau gain G is determined.

4. The method according to claim 2, characterized in that: The improved formula is introduced into the Kalman filter model to construct the G-CKF model, including: Obtain the initial state variables and initial state difference covariance matrix in the Kalman filter model; Based on the initial state variables and the initial state difference covariance matrix, the mean weights and covariance weights corresponding to the volume points are used to calculate the sum of the sampled volume points to determine a first volume point set; Based on the volume point set, a preliminary prediction is made on the state variables and the state difference covariance matrix to determine the state propagation value; performing volume point calculation again based on the state propagation value to determine a second volume point set; A G-CKF model is constructed based on the second volume point and the improved formula.

5. The method according to claim 4, characterized in that The constructing of the G-CKF model based on the second volume point and the improved formula includes: determining an observation covariance matrix and a cross-covariance matrix based on the second volume points; A G-CKF model is constructed based on the observation covariance matrix, the cross-covariance matrix and the voltage plateau gain.

6. The method according to claim 5, characterized in that The observation covariance matrix and the cross-covariance matrix are respectively: The observation covariance matrix: the cross-covariance matrix; Among them, y i,k / k-1 is the observed propagation value of the i-th Sigma point, is the estimated value of the observation at time k, is the prior estimate of the state at time k, R k represents the system observation noise v k The positive definite variance matrix of .

7. The method according to claim 1, characterized in that The estimating the battery SOC based on the G-CKF model and the observed current and observed voltage of the battery to be estimated includes: Determining a system state variable based on the observed current and the observed voltage of the battery to be estimated and the G-CKF model; Based on the system state variables, a battery SOC estimate is determined.

8. A battery SOC estimation device based on Kalman filtering, characterized in that: The device comprises: A construction module is used to construct a Kalman filter model, introduce an improved formula into the Kalman filter model, and construct a G-CKF model, wherein the improved formula includes a voltage plateau gain G, and the voltage plateau gain G is used to adjust the SOC estimation result; The estimation module is used to estimate the battery SOC based on the G-CKF model and the observed current and observed voltage of the battery to be estimated.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images