A method, apparatus, and device for determining the state of charge of a lithium battery.

Abstract

This invention discloses a method, apparatus, and device for determining the state of charge (SOC) of a lithium battery, relating to the field of battery modeling technology. Based on the theory of lithium battery relaxation time distribution, this invention first uses a genetic optimization algorithm to accurately obtain the relaxation time distribution parameters and distribution function of the lithium battery. The genetic algorithm aims to minimize the error exponent and smoothing exponent, while also considering the optimal regularization parameter and basis function width. Then, based on the optimal regularization parameter and basis function, the lithium battery relaxation time distribution is determined. Two indices are used to avoid the problems of spurious peaks and unsmoothness caused by overfitting and excessive errors caused by underfitting, respectively, so that the obtained relaxation time distribution result has characteristic peaks without spurious peaks. Subsequently, a rapid characteristic peak parameter extraction technique is proposed, using the eigenvalue of the target characteristic peak most relevant to the lithium battery's SOC to measure the lithium battery's SOC, thereby obtaining a more accurate prediction result of the lithium battery's SOC.

Description

Technical Field

[0001] This invention relates to the field of battery modeling technology, and in particular to a method, apparatus, and device for determining the state of charge of a lithium battery. Background Technology

[0002] Currently, the State of Charge (SOC) of a battery is difficult to estimate. Electrochemical impedance spectroscopy (EIS), a non-invasive and non-destructive measurement technique, can obtain the AC impedance of the battery over an extremely wide frequency range to analyze the complex physicochemical processes inside the battery, making it one of the important tools for studying battery SOC. Existing methods for estimating battery SOC typically include the open-circuit voltage method, the ampere-hour integration method, the data-driven method, and model-based methods. Among these, the open-circuit voltage method has poor accuracy and the correlation between battery voltage and SOC can change; the ampere-hour integration method has poor real-time performance; the data-driven method has poor interpretability and cannot dynamically reflect changes inside the battery; and the model-based method is robust but highly dependent on the accuracy of the model and cannot obtain accurate SOC values ​​by changing parameter values ​​with changes in battery impedance.

[0003] Electrochemical impedance spectroscopy (EIS) of a battery contains a wealth of internal information, which can be used to estimate the battery's state of charge (SOC), health status, and internal temperature. The Distribution of Relaxation Times (DRT) method, a high-fractional-order model, can be used to analyze the impedance spectrum. The core idea of ​​DRT is to deconvolve discrete frequency domain data into a continuous time-domain DRT function. This not only directly yields the distribution of the battery's polarization resistance in the time domain but also provides a more stable and effective way to estimate the SOC. Current research has shown a strong correlation between the battery's relaxation time distribution and SOC and temperature. Compared to other methods, using DRT to estimate the battery's SOC can intuitively reflect changes in battery impedance and dynamically adjust the accuracy of the SOC estimation based on changes in polarization resistance, avoiding model instability caused by changes in internal resistance over long battery lifespan. However, there are still many challenges in obtaining DRT and analyzing its correlation with SOC.

[0004] Existing methods for solving battery relaxation time distribution cannot avoid underfitting and overfitting during the fitting process of the relaxation time distribution function, resulting in relaxation time distribution results that cannot accurately distinguish polarization resistance characteristic peaks. Furthermore, there is currently no scientific method for estimating SOC using DRT; traditional SOC estimation mostly relies on data-driven and battery model-based approaches. Traditional methods have low applicability, poor dynamic response of the models, and cannot reflect the internal dynamic response process of the battery. Using these battery models to estimate the SOC of aging or currently in use results in low accuracy and low robustness. The interpretability and accuracy of lithium battery state-of-charge estimation are not suitable for practical monitoring. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, apparatus, and device for determining the state of charge of a lithium battery to address the aforementioned technical problems.

[0006] The present invention adopts the following technical solution:

[0007] This invention provides a method for determining the state of charge of a lithium battery, comprising:

[0008] Obtain the electrochemical impedance spectroscopy data of the target lithium battery; based on the initial regularization parameters, initial basis function full width at half maximum (FWHM), and electrochemical impedance spectroscopy data of the preset ridge regression regularization method, determine the initial relaxation time distribution function of the target lithium battery using the ridge regression regularization method.

[0009] The fitting impedance spectrum data corresponding to the initial relaxation time distribution function is determined. The optimization objectives are to minimize the deviation between the fitting impedance spectrum data and the electrochemical impedance spectrum data and to smooth the initial relaxation time distribution function. The parameters are optimized from the preset regularization parameter search range and the preset basis function full width at half maximum value search range to determine the preferred regularization parameter and the preferred basis function full width at half maximum value.

[0010] Based on the preferred regularization parameters, preferred full width at half maximum (FWHM) values ​​of the basis functions, and electrochemical impedance spectroscopy data, the relaxation time distribution function of the target lithium battery is determined using the ridge regression regularization method.

[0011] The relaxation time distribution function of the target lithium battery under different states of charge is determined and fitted using the least squares method. Based on the fitted function, the characteristic values ​​of each characteristic peak of the relaxation time distribution function under different states of charge of the same type of battery are determined. Based on the Pearson correlation coefficient between the characteristic values ​​of each characteristic peak and the corresponding state of charge, the target characteristic peak most related to the change of state of charge of the target lithium battery is determined.

[0012] The eigenvalues ​​of the target characteristic peak in the relaxation time distribution function of the target lithium battery are determined, and the state of charge of the target lithium battery is determined based on the correlation between the eigenvalues ​​of the target characteristic peak in the relaxation time distribution function of the target lithium battery and the state of charge.

[0013] Optionally, the initial relaxation time distribution function of the target lithium battery can be determined using the ridge regression regularization method as follows:

[0014]

[0015] Where N is the number of frequencies in the impedance spectrum data, Z fit (f m () represents the frequency f m Fitted impedance under X n φ represents the vector coefficients of the relaxation time distribution function. n (τ) is a basis function, f m Here, τ represents the frequency corresponding to each data point in the electrochemical impedance spectroscopy data, τ is the time constant, i is the imaginary part, and e is the frequency. z (f) represents the fitting error, Z exp ' represents the real part of the electrochemical impedance spectroscopy data, Z exp " represents the imaginary part of the electrochemical impedance spectroscopy data, A is the fitted coefficient matrix, λ is the regularization parameter, and L is the second derivative matrix.

[0016] Optionally, the optimization objective of minimizing the deviation between the fitted impedance spectroscopy data and the electrochemical impedance spectroscopy data and smoothing the initial relaxation time distribution function involves parameter optimization within a preset search range for regularization parameters and a preset search range for the full width at half maximum (FWHM) of the basis functions, to determine the preferred regularization parameters and preferred FWHM values. Specifically, this includes:

[0017] The error exponent of the initial relaxation time distribution function is determined based on the deviation between the fitted impedance spectroscopy data and the electrochemical impedance spectroscopy data.

[0018] The smoothing exponent of the initial relaxation time distribution function is determined based on the number of peaks and the curvature variance of the initial relaxation time distribution function.

[0019] With the goal of minimizing the error exponent and smoothing exponent of the relaxation time distribution function, a multi-objective genetic optimization algorithm is used to optimize the parameters from the preset search range of regularization parameters and the preset search range of the full width at half maximum of the basis functions, and to determine the non-dominated solution set of the regularization parameters and the full width at half maximum of the basis functions.

[0020] Multiple preferred solutions are selected from the non-dominated solution set, and the mean of the regularization parameter is calculated based on each preferred solution as the preferred regularization parameter, and the mean of the full width at half maximum (FWHM) of the basis functions is calculated as the preferred full width at half maximum (FWHM) of the basis functions.

[0021] Optionally, the initial relaxation time distribution function of the target lithium battery can be determined using the ridge regression regularization method as follows:

[0022]

[0023] Where N is the number of frequencies in the impedance spectrum data, Z fit (f m () represents the frequency f m Fitted impedance under X n φ represents the vector coefficients of the relaxation time distribution function. n (τ) is a basis function, f m Here, τ represents the frequency corresponding to each data point in the electrochemical impedance spectroscopy data, τ is the time constant, i is the imaginary part, and e is the frequency. z (f) represents the fitting error, Z exp ' represents the real part of the electrochemical impedance spectroscopy data, Z exp " represents the imaginary part of the electrochemical impedance spectroscopy data, A is the fitted coefficient matrix, λ is the regularization parameter, and L is the second derivative matrix.

[0024] Optionally, the error exponent of the initial relaxation time distribution function can be determined based on the fitted impedance spectroscopy data and electrochemical impedance spectroscopy data using the following formula:

[0025]

[0026] in, Z' is the error exponent of the relaxation time distribution function. exp (f n Z” represents the real part of the electrochemical impedance spectroscopy data. exp (f n Z' is the imaginary part of the electrochemical impedance spectroscopy data. fit (f n Z” represents the real part of the fitted impedance spectrum data. fit (f n ) represents the imaginary part of the fitted impedance spectrum data.

[0027] Alternatively, the smoothing exponent of the initial relaxation time distribution function can be determined by the following formula:

[0028] Idx smth =Sch*100*Var(K(τ)),

[0029]

[0030] Among them, K i S is the curvature of the fitted data. ch The number of peaks in the relaxation time distribution, Var() is the variance, and Idx is the peak value. smth τ is the smoothing exponent, τ is the time constant, and g(τ) is the relaxation time distribution function.

[0031] Optionally, the optimization objective of minimizing the error exponent and smoothing exponent of the relaxation time distribution function, employing a multi-objective genetic optimization algorithm to optimize parameters from a preset search range for regularization parameters and a preset search range for the full width at half maximum (FWHM) of the basis functions, and determining the non-dominated solution set for the regularization parameters and the FWHM of the basis functions, specifically includes:

[0032] An initial population is created by randomly selecting parameters from the preset regularization parameter search range and the preset basis function half-peak full width value search range. The error index and smoothness index of the relaxation time distribution of each individual in the population are determined, and the offspring population is obtained through selection, crossover and mutation among individuals in the initial population.

[0033] An elite retention strategy is used to merge the initial population and the offspring population to obtain a merged population; then the dominance relationship between individuals in the merged population is determined, and the non-dominant solutions and mutually non-dominant solutions are used as the initial optimal solution set.

[0034] The initial optimal solution set is sorted by fast non-dominated sorting and the crowding distance of each individual is calculated. Based on the crowding distance, individuals are selected to form the updated parent population using a targeted optimization method.

[0035] The updated offspring population is obtained by selection, crossover and mutation among individuals in the updated parent population, and the error index and smoothness index of the relaxation time distribution of each individual in the updated offspring population are determined.

[0036] Through multiple iterations until the updated optimal solution set obtained from the updated parent population and the updated child population converges, the converged optimal solution set is used as the non-dominated solution set of the regularization parameter and the full width at half maximum of the basis functions.

[0037] Optionally, determining the relaxation time distribution function of the target lithium battery under different states of charge and fitting it using the least squares method specifically includes:

[0038] For each preset state of charge of the target lithium battery of the same type, determine the relaxation time distribution function under that state of charge, and extract the characteristic peak height, center frequency and full width at half maximum (FWHM) of the relaxation time distribution function under that state of charge.

[0039] Based on the peak height, center frequency, and full width at half maximum (FWHM) of the characteristic peaks in the relaxation time distribution function under this charged state, and using Gaussian functions as basis functions, the following formula is used to fit multiple Gaussian functions corresponding to each characteristic peak using the least squares method:

[0040]

[0041] Among them, G m(x) is the Gaussian function corresponding to the m-th characteristic peak, f m It is the center frequency corresponding to the m-th characteristic peak, A m Hf is the peak height corresponding to the m-th characteristic peak. m It is the full width at half maximum (FWHM) corresponding to the m-th characteristic peak.

[0042] Optionally, the characteristic values ​​of each characteristic peak are the peak area and peak height of each characteristic peak;

[0043] The determination of the state of charge (SOC) of the target lithium battery based on the correlation between the eigenvalues ​​of the target characteristic peak in the relaxation time distribution function of the target lithium battery and the SOC specifically includes:

[0044] The state of charge (SOC) of the target lithium battery is determined by the following formula, based on the correlation between the peak area and peak height of the target characteristic peak in the relaxation time distribution function of the target lithium battery and the SOC:

[0045] SOC = k1S + k2A + k3Asoh;

[0046] Where SOC is the state of charge of the target lithium battery, S is the peak area of ​​the target characteristic peak, A is the peak height of the target characteristic peak, Asoh is the ratio of the first difference between the total impedance of the relaxation time distribution function of the target lithium battery and the impedance of the corresponding target characteristic peak to the second difference between the total impedance of the relaxation time distribution function of the target lithium battery under the rated state of charge and the impedance of the corresponding target characteristic peak, and k1, k2 and k3 are coefficients obtained by fitting the characteristic parameters of the characteristic peak.

[0047] This invention provides a device for determining the state of charge of a lithium battery, comprising:

[0048] The acquisition module is used to acquire the electrochemical impedance spectroscopy data of the target lithium battery; based on the initial regularization parameters, initial basis function full width at half maximum and electrochemical impedance spectroscopy data of the preset ridge regression regularization method, the initial relaxation time distribution function of the target lithium battery is determined by the ridge regression regularization method.

[0049] The optimization module is used to determine the fitting impedance spectrum data corresponding to the initial relaxation time distribution function. The optimization objectives are to minimize the deviation between the fitting impedance spectrum data and the electrochemical impedance spectrum data and to smooth the initial relaxation time distribution function. The module optimizes the parameters from the preset regularization parameter search range and the preset basis function full width at half maximum value search range to determine the preferred regularization parameter and the preferred basis function full width at half maximum value.

[0050] The fitting module is used to determine the relaxation time distribution function of the target lithium battery using the ridge regression regularization method based on the preferred regularization parameters, the preferred full width at half maximum of the basis function and the electrochemical impedance spectroscopy data.

[0051] A selection module is used to determine the relaxation time distribution function of the same type of target lithium battery under different states of charge, and fit it using the least squares method; based on the fitted function, the characteristic values ​​of each characteristic peak of the relaxation time distribution function under different states of charge of the same type of battery are determined, and based on the Pearson correlation coefficient between the characteristic values ​​of each characteristic peak and the corresponding state of charge, the target characteristic peak most related to the change of state of charge of the target lithium battery is determined.

[0052] The determination module is used to determine the characteristic value of the target characteristic peak in the relaxation time distribution function of the target lithium battery, and to determine the state of charge of the target lithium battery based on the correlation between the characteristic value of the target characteristic peak in the relaxation time distribution function of the target lithium battery and the state of charge.

[0053] The present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-mentioned method for determining the state of charge of a lithium battery.

[0054] The above-mentioned at least one technical solution adopted in this invention can achieve the following beneficial effects:

[0055] This invention first optimizes the relaxation time distribution function by minimizing its error exponent and smoothing exponent, while also considering the optimal regularization parameter and basis function width. This process determines the relaxation time distribution function of the target lithium battery, avoiding the problems of spurious peaks and unsmoothness caused by overfitting and excessive errors caused by underfitting. The resulting relaxation time distribution has characteristic peaks without spurious peaks, improving the discriminability of the analytical results and better reflecting the response of the battery's internal polarization resistance. Furthermore, based on the correlation between the eigenvalues ​​of the characteristic peaks of the relaxation time distribution function of the target lithium battery under different states of charge (SDCs) and their corresponding SDCs, the characteristic peaks in the target lithium battery's relaxation time distribution function are optimized. By determining the correlation between the eigenvalues ​​of the optimized target characteristic peaks and the SDC, the SDC of the target lithium battery is determined, improving the interpretability, accuracy, and applicability of lithium battery SDC estimation. Attached Figure Description

[0056] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0057] Figure 1 A schematic flowchart of a method for determining the state of charge of a lithium battery provided by the present invention;

[0058] Figure 2 A flowchart of a relaxation process distributed optimizer is provided for this invention;

[0059] Figure 3 A schematic diagram illustrating the final DRT result obtained from the impedance spectrum within the range of 0.1Hz to 3000Hz under different SOCs, below the optimal value, provided by this invention;

[0060] Figure 4 A schematic diagram of fitting parameters extracted by DRT using a Gaussian function is provided for this invention;

[0061] Figure 5 A schematic diagram of a lithium battery state-of-charge determination device provided by the present invention;

[0062] Figure 6 A schematic diagram of a computer device for implementing a method for determining the state of charge of a lithium battery, as provided by the present invention. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0064] Chinese invention patent CN104914312A discloses a method for calculating the relaxation time distribution of AC impedance spectra. However, this method directly specifies the regularization coefficient when applying the Tikhonov regularization method, without explicitly proposing a method for selecting the coefficient. The proper selection of the regularization parameter is a core issue in regularization methods; an excessively large regularization parameter can cause characteristic peaks to couple together, making them difficult to separate, while an excessively small regularization parameter can lead to unnecessary spurious peaks and spikes. Furthermore, this method does not consider the pure capacitive behavior of battery EIS at low frequencies, which may cause the Tikhonov regularization method to exhibit non-convergence.

[0065] Chinese invention patent CN112540316A discloses a method for analyzing the impedance spectrum of complex batteries. This method does not perform hypothetical extensions in the low-frequency region, resulting in insufficient completeness of the DRT function. Furthermore, it does not consider the selection of the basis function width, which significantly affects the results. A large basis function width can lead to excessive interference from multiple functions at a certain point, resulting in characteristic coupling peaks. Conversely, a small basis function width causes each peak to affect only a short period, ultimately leading to oscillations in the results.

[0066] Currently, there are two main methods for selecting regularization parameters: The first is the cross-validation method, which calculates the errors of the regularization results obtained from the real and imaginary parts of the electrochemical impedance spectroscopy (EIS) data separately. The parameter with the smallest difference between the two errors is the best regularization parameter. However, this method requires continuous verification of the differences, making the calculations cumbersome, and it cannot guarantee that the result will not have spurious peaks or overfitting if the difference is minimized. The second method is the L-curve method, which calculates the errors of all DRT results corresponding to different regularization parameters and selects the regularization parameter with the smallest error as the best parameter. However, the second method involves a large amount of computation, is time-consuming, and cannot guarantee smooth results or the occurrence of overfitting.

[0067] Currently, there is no clear method for selecting the basis function width. It is usually chosen empirically as the basis function width for Tikhonov regularization. However, the impedance spectrum at different frequencies and in different frequency bands cannot be selected solely based on experience. The coupling degree of polarization resistance varies in different frequency bands, and the requirements for the basis function width also vary with the number of frequency points. When there are many frequency points, the basis function width needs to be reduced. When there are many frequency points, the coupling relationship is not obvious and spurious peaks are prone to appear, so the basis function width needs to be increased. Therefore, a clear selection method is needed to select the appropriate basis function width for different impedance spectrum data.

[0068] There is currently no method for simultaneously optimizing regularization parameters and basis function widths. This invention provides a fast optimization method based on a multi-objective genetic optimization algorithm to select the optimal regularization parameters and basis function widths to avoid overfitting and underfitting, and finally obtain characteristic DRT results.

[0069] Based on DRT, SOC can be estimated and is well applied to most lithium-ion batteries. A relatively accurate SOC value can be obtained by analyzing and calculating the relaxation time distribution function. It can also dynamically determine the change of polarization impedance contribution during aging. For a specific battery model, the changes in aging degree can be analyzed for each DRT peak measured each time. In this way, the relationship between the SOC and DRT characteristic peaks of the lithium battery throughout the entire life cycle can be obtained, and a large amount of preliminary data research is avoided.

[0070] The technical solutions provided by the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0071] Figure 1 This is a schematic flowchart of a method for determining the state of charge of a lithium battery according to the present invention, which specifically includes the following steps:

[0072] S101: Obtain the electrochemical impedance spectroscopy data of the target lithium battery; based on the initial regularization parameters, initial basis function full width at half maximum (FWHM), and electrochemical impedance spectroscopy data of the preset ridge regression regularization method, determine the initial relaxation time distribution function of the target lithium battery using the ridge regression regularization method.

[0073] Generally, before determining the relaxation time distribution of a lithium battery, electrochemical impedance spectroscopy (EIS) data can be measured using instruments such as an electrochemical workstation for the target lithium battery. This EIS data can be full-frequency EIS data.

[0074] Based on this, in one or more embodiments of the present invention, the server of the business platform can first perform KK verification on the obtained electrochemical impedance spectroscopy data, and filter out the unqualified impedance spectroscopy data through KK verification.

[0075] Furthermore, in one or more embodiments of the present invention, after performing KK verification, the server can fit the high-frequency portion of the electrochemical impedance spectroscopy data based on inductance, and the low-frequency portion based on insertion capacitance, using the following formula:

[0076]

[0077] In the formula, ω is the angular velocity, R0 is the ohmic internal resistance, and R pol Let g(τ) be the polarization resistance, g(τ) be the relaxation time distribution function, τ be the time constant, and C be the polarization resistance. in Let w be the number of inserted capacitors.

[0078] During the fitting process, the insertion capacitance is obtained as follows: A fully charged battery is discharged using a current of 1 / 100C, from a state of charge (SOC) of 1 to the cutoff voltage. The discharge IC curve is recorded, and the value of the insertion capacitance is calculated using the following formula:

[0079]

[0080] In the formula, Cin is the insertion capacitance, and Q... li dQ represents the actual capacity of the battery, soc represents the state of charge of the battery, ocv represents the open-circuit voltage of the battery, dQ represents the change in battery charge, and dV represents the change in voltage.

[0081] For the mid-frequency portion of the electrochemical impedance spectroscopy data of the target lithium battery, the relaxation time distribution function can be fitted, and then the relaxation time distribution can be calculated.

[0082] Subsequently, the server can determine the initial relaxation time distribution function of the target lithium battery using the ridge regression regularization method based on the preset initial regularization parameters, initial basis function full width at half maximum (FWHM) values, and electrochemical impedance spectroscopy data. Then, it can optimize the regularization parameters and basis function FWHM values ​​in subsequent operations.

[0083] The server mentioned in this invention can be a server set up on a business platform, or a device such as a desktop computer or laptop computer capable of executing the solution of this invention. For ease of explanation, the following description will only focus on the server as the executing entity.

[0084] S102: Determine the fitted impedance spectrum data corresponding to the initial relaxation time distribution function. With the goal of minimizing the deviation between the fitted impedance spectrum data and the electrochemical impedance spectrum data and smoothing the initial relaxation time distribution function, perform parameter optimization from the preset regularization parameter search range and the preset basis function full width at half maximum (FWHM) value search range to determine the preferred regularization parameter and the preferred basis function FWHM value.

[0085] S103: Based on the preferred regularization parameters, preferred basis function full width at half maximum (FWHM) values, and electrochemical impedance spectroscopy data, the relaxation time distribution function of the target lithium battery is determined using the ridge regression regularization method.

[0086] In determining the initial relaxation time distribution function of the target lithium battery, the selection of regularization parameters and basis functions has a significant impact on the accuracy of the final result. Therefore, in one or more embodiments of this invention, the regularization parameters and basis functions are first optimized.

[0087] For details, please refer to Figure 2 , Figure 2 This is a flowchart of a distributed optimizer for a relaxation process in this invention. The server can first obtain a preset search range for regularization parameters, for example, assuming the preset search range for regularization parameter λ is 0.1 to 10. -15 Then the server can determine the minimum boundary value of the search range, 10. -15 This is the initial regularization parameter. Of course, the specific selection of the initial regularization parameter can be determined as needed, and this invention does not impose any restrictions on it. Similarly, the server can obtain a preset search range for the regularization parameter. For example, assuming the preset search range for the full width at half maximum (FWHM) of the basis function is 0.1Δlnτ to 10Δlnτ, the server can determine the minimum boundary value of this search range as 10. -15 10Δlnτ is the full width at half maximum (FWHM) of the initial basis functions, and the basis functions are determined from this value.

[0088] Furthermore, the server can also simultaneously determine the search speed; the minimum search speeds for λ and FWHM are 0.1 * 10^25. -15The maximum search speeds of 0.05Δlnτ, λ, and FWHM are 10, respectively. -15 0.2Δlnτ. The server can randomly determine the initial search speed of λ and FWHM within the range of minimum and maximum search speeds, where λ can be selected according to a logarithmic scale.

[0089] The server can then begin DRT iterative calculations. This invention introduces two metrics: the error exponent R. 2 opt And smoothing exponent Idx smth .

[0090] Specifically, the error exponent of the initial relaxation time distribution function can be determined using the following formula based on the fitted impedance spectroscopy data and electrochemical impedance spectroscopy data:

[0091]

[0092] In the formula, Z' is the error exponent of the relaxation time distribution function. exp (f n Z” represents the real part of the electrochemical impedance spectroscopy data. exp (f n Z' is the imaginary part of the electrochemical impedance spectroscopy data. fit (f n Z” represents the real part of the fitted impedance spectrum data. fit (f n ) represents the imaginary part of the fitted impedance spectrum data.

[0093] The error exponent can effectively reflect the deviation between the fitted result and the true result. Considering both the real and imaginary parts simultaneously can avoid situations where a single real or imaginary part is excessively offset.

[0094] The smoothing exponent of the initial relaxation time distribution function is determined by the following formula:

[0095] Idx smth =Sch*100*Var(K(τ))

[0096]

[0097] In the formula, K i S is the curvature of the fitted data. ch The number of peaks in the relaxation time distribution, Var() is the variance, and Idx is the peak value. smth Let g(τ) be the smoothing exponent, τ be the time constant, and g(τ) be the relaxation time distribution function. By introducing the number of peaks and the curvature variance, oscillations and the appearance of spurious peaks can be avoided.

[0098] By combining the error exponent and the smoothness exponent, the optimal Pareto solution can be obtained using an improved multi-objective genetic optimization algorithm.

[0099] Specifically, the server can randomly select parameters from the preset regularization parameter search range and the preset basis function half-peak full width value search range to create an initial population, determine the error index and smoothness index of the relaxation time distribution of each individual in the population, and obtain the offspring population through selection, crossover and mutation among individuals in the initial population.

[0100] An elite retention strategy is adopted to merge the initial population and the offspring population to obtain a merged population; and the dominance relationship between individuals in the merged population is determined, and the non-dominated solutions and mutually non-dominated solutions are used as the initial optimal solution set.

[0101] The initial optimal solution set is sorted by fast non-dominated sorting and the crowding distance of each individual is calculated. Based on the crowding distance, individuals are selected to form the updated parent population using a targeted optimization method.

[0102] An updated offspring population is obtained by selection, crossover, and mutation among individuals in the updated parent population, and the error index and smoothness index of the relaxation time distribution of each individual in the updated offspring population are determined.

[0103] Through multiple iterations until the updated optimal solution set obtained from the updated parent population and the updated child population converges, the converged optimal solution set is used as the non-dominated solution set of the regularization parameter and the full width at half maximum of the basis functions.

[0104] Furthermore, in one or more embodiments of the present invention, in order to accelerate the iteration speed of this method, E can also be introduced. set The preset parameters have been further improved. set To achieve convergence, a preset value is typically set to 0.1. After each iteration, the error exponent R is adjusted. 2 opt Individuals with a value greater than 0.1 are excluded from the optimal solution set to accelerate the iteration process. A convergence preset value is used to accelerate the search for local and global optima. This value can be dynamically changed if the final result is not less than E. set The solution is E. set It will dynamically increase by 0.01 and perform another iterative search.

[0105] After optimizing the regularization parameters and the full width at half maximum (FWHM) of the basis functions to obtain the Pareto solution set, the server can perform mesh generation on the Pareto solution set. For example, the horizontal axis can be set to λ, with each log10 interval representing one unit, and the vertical axis to μ, with μ = 2 representing one unit. This can be obtained from the following formula:

[0106]

[0107] The above equation represents the relationship between μ and the full width at half maximum (FWHM). When the basis functions are Gaussian functions, the relationship between μ and the width of the basis functions is easier to calculate by substituting μ into the basis functions. Therefore, μ is used to substitute into the basis functions for fitting calculation.

[0108] Then, refer to Figure 2 The server can first use the minimum error index corresponding to each solution in the non-dominated solution set as a benchmark, and then select the first grid containing the solution within a preset error index fluctuation range from each grid in the grid coordinate system. For example, select the first grid S1 containing the solution whose accuracy error index is within 15% of the minimum value.

[0109] Next, calculate the grid density and grid position corresponding to each solution in the first grid, and identify unreasonable edge solutions in each solution whose grid density is less than the preset value or which are located at the grid edge.

[0110] Secondly, based on the minimum smoothing exponent corresponding to each solution in the non-dominated solution set, a second grid containing the solution within a preset smoothing exponent fluctuation range is selected from the first grid. For example, the minimum smoothing exponent Idx is calculated from the first grid S1. min Then select a smoothing exponent < 5 * Idx min The solution is located in the second grid S2.

[0111] Next, unreasonable edge solutions are removed from the second grid to obtain the third grid S3, and the remaining solutions are selected as multiple optimal solutions. The mean value of the regularization parameter and the mean value of the full width at half maximum of the basis functions are calculated based on each optimal solution. For example, the optimal λ and μ are obtained by averaging all solutions in the third grid S3.

[0112] Finally, the server can determine the relaxation time distribution function of the target lithium battery using the Tikhonov regularization method based on the mean full width at half maximum (FWHM) values ​​of the optimal λ and μ basis functions and the electrochemical impedance spectroscopy data.

[0113] Table 1. Optimizer results for different frequency points.

[0114] R2opt Idxsmth λ μ FWHM ppds=30 0.0115 2.1164e-05 10.8611 9.53926 1.75806△lnτ ppds=20 0.0021 2.80128e-04 10.38825 7.514933 2.0856△lnτ ppds=10 0.0033 4.8824e-04 9.742448 5.185193 2.8659△lnτ ppds=5 0.0092 9.83369e-12 6.38971 1.77413 1.962△lnτ

[0115] Table 1 shows the optimal solutions obtained by implementing this scheme for electrochemical impedance spectroscopy at different frequency points in the frequency range of 0.1Hz to 3000Hz, along with the error index and smoothness index corresponding to the optimal solutions. Figure 3 This is a schematic diagram of the final DRT result obtained by impedance spectrum in the range of 0.1Hz to 3000Hz under different SOCs in this invention, below the optimal value.

[0116] S104: Determine the relaxation time distribution function of the target lithium battery under different states of charge and fit it using the least squares method; determine the characteristic values ​​of each characteristic peak of the relaxation time distribution function under different states of charge of the same type of battery based on the fitted function, and determine the target characteristic peak most relevant to the change of state of charge of the target lithium battery based on the characteristic value of each characteristic peak and the Pearson correlation coefficient of the corresponding state of charge.

[0117] S105: Determine the characteristic value of the target characteristic peak in the relaxation time distribution function of the target lithium battery, and determine the state of charge of the target lithium battery based on the correlation between the characteristic value of the target characteristic peak in the relaxation time distribution function of the target lithium battery and the state of charge.

[0118] After determining the relaxation time distribution function of the target lithium battery, the server can further analyze its state of charge based on the characteristic peaks of the target lithium battery's relaxation time.

[0119] Based on this, in one or more embodiments of the present invention, the server can pre-determine the relaxation time distribution function of different states of charge of the target lithium battery of the same type, and use the characteristic peak rapid extraction technology. First, for each pre-determined state of charge of the target lithium battery of the same type, the relaxation time distribution function of that state of charge is determined, and the highest peak value and the corresponding center frequency of the characteristic peak in the relaxation time distribution function of the target lithium battery of the same type under that state of charge are extracted. Then, the full width at half maximum (FWHM) value of each peak is found in turn. In this way, three parameters of each characteristic peak are obtained: peak height, center frequency, and full width at half maximum (FWHM).

[0120] Then, the server can fit the characteristic peak height, center frequency, and full width at half maximum (FWHM) of the relaxation time distribution function under different states of charge using the Gaussian function as the basis function and the least squares method according to the following formula, to obtain multiple Gaussian functions corresponding to each characteristic peak:

[0121] Based on the peak height, center frequency, and full width at half maximum (FWHM) of the characteristic peaks in the relaxation time distribution function under this charged state, and using Gaussian functions as basis functions, the following formula is used to fit multiple Gaussian functions corresponding to each characteristic peak using the least squares method:

[0122]

[0123] In the formula, G m (x) is the Gaussian function corresponding to the m-th characteristic peak, f m It is the center frequency corresponding to the m-th characteristic peak, A m Hf is the peak height corresponding to the m-th characteristic peak. m It is the full width at half maximum (FWHM) corresponding to the m-th characteristic peak.

[0124] Each Gaussian function corresponds to a characteristic peak. Integrating each Gaussian function yields the area of ​​the corresponding characteristic peak at different center frequencies. Using the peak area and peak height of all characteristic peaks as evaluation indicators, these two evaluation indicator values ​​are measured and calculated under different states of charge. Then, the Pearson correlation coefficients between the two evaluation indicators and the state of charge are calculated. The characteristic peak corresponding to the Gaussian function most correlated with the change in state of charge is selected, and the state of charge is estimated using the peak area and peak height of this characteristic peak in the relaxation time distribution function of the target lithium battery. Figure 4 This is a schematic diagram of fitting parameters extracted by DRT using a Gaussian function in this invention.

[0125] In one or more embodiments of the present invention, the state of charge (SOC) and the peak area and peak height of this characteristic peak can be nonlinearly fitted, and the SOC of the target lithium battery can be determined by the following formula based on the correlation between the peak area and peak height of the target characteristic peak and the SOC in the relaxation time distribution function of the target lithium battery:

[0126] SOC=k1S+k2A+k3Asoh

[0127] Wherein, SOC is the state of charge of the target lithium battery, S is the peak area of ​​the target characteristic peak, A is the peak height of the target characteristic peak, Asoh is the ratio of the first difference between the total impedance of the relaxation time distribution function of the target lithium battery and the impedance of the corresponding target characteristic peak to the second difference between the total impedance of the relaxation time distribution function of the target lithium battery under the rated state of charge and the impedance of the corresponding target characteristic peak. Asoh can reflect the change relationship of the state of health (SOH) of the target lithium battery to a certain extent, and k1, k2 and k3 are coefficients obtained by fitting the characteristic value parameters of the characteristic peak.

[0128] based on Figure 1 The method for determining the state of charge (SOC) of a lithium battery, as shown, uses inductors and interpolated capacitors to fit impedance spectrum data across the entire frequency band, making it suitable for DRT fitting across the entire frequency band. The fast optimization method for regularization parameters and basis function widths, employing error and smoothness exponents, effectively avoids overfitting and underfitting problems in ridge regression regularization, resulting in eigenpeaks without spurious peaks. The automatic selection of a fast convergence architecture for multi-objective optimization uses a convergence preset value to accelerate convergence and dynamically adjusts the preset value to provide the optimal solution under the current conditions.

[0129] The optimization objective is to minimize the error exponent and smoothing exponent of the relaxation time distribution function. The optimization of regularization parameters and basis function widths is also considered. The relaxation time distribution of lithium batteries is then determined based on the optimal regularization parameters and basis functions. Two indices are used to avoid the problems of spurious peaks and unsmoothness caused by overfitting and excessive errors caused by underfitting, respectively. This results in a relaxation time distribution with characteristic peaks and no spurious peaks, which improves the recognizability of the analytical results of the lithium battery relaxation time distribution and better reflects the response of the internal polarization resistance of the battery.

[0130] Meanwhile, the DRT distribution parameter extraction method used in this invention can accurately extract parameters such as peak height and peak area of ​​DRT. At the same time, the fitting regression method is used to obtain the characteristic peak parameters most related to SOC. The influence of aging changes on battery estimation accuracy is also considered. It combines data-driven and model recognition methods, but does not require a large amount of data for pre-calculation. While obtaining SOC, the polarization resistance parameters of the battery's internal reaction can also be obtained, which better reflects the ability to estimate the battery's state of charge and has good dynamic response.

[0131] This invention is applicable to different states of charge (SOC = 0 to 1) and different frequency points (ppd), and it is applicable to any impedance spectrum data.

[0132] When applying the lithium battery state-of-charge determination method provided by this invention, it is not necessary to rely on... Figure 1 The steps shown are executed in sequence. The specific execution order of each step can be determined as needed, and this invention does not impose any restrictions on it.

[0133] The above describes a method for determining the state of charge (SOC) of a lithium battery according to one or more embodiments of the present invention. Based on the same idea, the present invention also provides a corresponding device for determining the SOC of a lithium battery, such as... Figure 5 As shown.

[0134] Figure 5 A schematic diagram of a lithium battery state-of-charge determination device provided by the present invention includes:

[0135] The acquisition module 201 is used to acquire the electrochemical impedance spectroscopy data of the target lithium battery; based on the initial regularization parameters, the full width at half maximum (FWHM) of the initial basis function and the electrochemical impedance spectroscopy data of the preset ridge regression regularization method, the initial relaxation time distribution function of the target lithium battery is determined by the ridge regression regularization method.

[0136] The optimization module 202 is used to determine the fitted impedance spectrum data corresponding to the initial relaxation time distribution function. The optimization objectives are to minimize the deviation between the fitted impedance spectrum data and the electrochemical impedance spectrum data and to smooth the initial relaxation time distribution function. The module optimizes the parameters from the preset regularization parameter search range and the preset basis function full width at half maximum value search range to determine the preferred regularization parameter and the preferred basis function full width at half maximum value.

[0137] The fitting module 203 is used to determine the relaxation time distribution function of the target lithium battery by using the ridge regression regularization method based on the preferred regularization parameters, the preferred full width at half maximum of the basis function and the electrochemical impedance spectroscopy data.

[0138] Module 204 is selected to determine the relaxation time distribution function of the same type of target lithium battery under different states of charge, and fit it using the least squares method; based on the fitted function, the characteristic values ​​of each characteristic peak of the relaxation time distribution function under different states of charge of the same type of battery are determined, and based on the characteristic values ​​of each characteristic peak and the Pearson correlation coefficient of the corresponding state of charge, the target characteristic peak most related to the change of state of charge of the target lithium battery is determined.

[0139] The determination module 205 is used to determine the characteristic value of the target characteristic peak in the relaxation time distribution function of the target lithium battery, and to determine the state of charge of the target lithium battery based on the correlation between the characteristic value of the target characteristic peak in the relaxation time distribution function of the target lithium battery and the state of charge.

[0140] Specific limitations regarding the lithium battery state-of-charge determination device can be found in the limitations of the lithium battery state-of-charge determination method described above, and will not be repeated here. Each module in the aforementioned lithium battery state-of-charge determination device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0141] The present invention also provides a computer-readable storage medium storing a computer program that can be used to execute the above-described... Figure 1 The provided method for determining the state of charge of lithium batteries.

[0142] The present invention also provides Figure 6 The schematic diagram of the computer device shown is as follows: Figure 6 As shown, at the hardware level, this computer device includes a processor, internal bus, network interface, memory, and non-volatile memory, and may also include other hardware required for business operations. The processor reads the corresponding computer program from the non-volatile memory into memory and then executes it to achieve the above. Figure 1The provided method for determining the state of charge of lithium batteries.

[0143] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0144] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this invention.

Claims

1. A method for determining the state of charge of a lithium battery, characterized in that, include: Obtain electrochemical impedance spectroscopy data of the target lithium battery; Based on the initial regularization parameters, initial basis function full width at half maximum (FWHM), and electrochemical impedance spectroscopy data of the pre-defined ridge regression regularization method, the initial relaxation time distribution function of the target lithium battery is determined using the ridge regression regularization method. The fitting impedance spectrum data corresponding to the initial relaxation time distribution function is determined. The optimization objectives are to minimize the deviation between the fitting impedance spectrum data and the electrochemical impedance spectrum data and to smooth the initial relaxation time distribution function. The parameters are optimized from the preset regularization parameter search range and the preset basis function full width at half maximum value search range to determine the preferred regularization parameter and the preferred basis function full width at half maximum value. Based on the preferred regularization parameters, preferred full width at half maximum (FWHM) values ​​of the basis functions, and electrochemical impedance spectroscopy data, the relaxation time distribution function of the target lithium battery is determined using the ridge regression regularization method. The relaxation time distribution function of the target lithium battery under different states of charge is determined and fitted using the least squares method. Based on the fitted function, the characteristic values ​​of each characteristic peak of the relaxation time distribution function under different states of charge of the same type of battery are determined. Based on the Pearson correlation coefficient between the characteristic values ​​of each characteristic peak and the corresponding state of charge, the target characteristic peak most related to the change of state of charge of the target lithium battery is determined. The eigenvalues ​​of the target characteristic peak in the relaxation time distribution function of the target lithium battery are determined, and the state of charge of the target lithium battery is determined based on the correlation between the eigenvalues ​​of the target characteristic peak in the relaxation time distribution function of the target lithium battery and the state of charge.

2. The method for determining the state of charge of a lithium battery as described in claim 1, characterized in that, The initial relaxation time distribution function of the target lithium battery is determined using the ridge regression regularization method according to the following formula: Where N is the number of frequencies in the impedance spectrum data, Z fit (f m () represents the frequency f m Fitted impedance under X n φ represents the vector coefficients of the relaxation time distribution function. n (τ) is a basis function, f m Here, τ represents the frequency corresponding to each data point in the electrochemical impedance spectroscopy data, τ is the time constant, i is the imaginary part, and e is the frequency. z (f) represents the fitting error, Z exp ' represents the real part of the electrochemical impedance spectroscopy data, Z exp " represents the imaginary part of the electrochemical impedance spectroscopy data, A is the fitted coefficient matrix, λ is the regularization parameter, and L is the second derivative matrix.

3. The method for determining the state of charge of a lithium battery as described in claim 1, characterized in that, The optimization objectives are to minimize the deviation between the fitted impedance spectroscopy data and the electrochemical impedance spectroscopy data, and to smooth the initial relaxation time distribution function. Parameter optimization is performed from a preset search range for regularization parameters and a preset search range for the full width at half maximum (FWHM) of the basis functions. The optimal regularization parameters and optimal FWHM values ​​are determined, specifically including: The error exponent of the initial relaxation time distribution function is determined based on the deviation between the fitted impedance spectroscopy data and the electrochemical impedance spectroscopy data. The smoothing exponent of the initial relaxation time distribution function is determined based on the number of peaks and the curvature variance of the initial relaxation time distribution function. With the goal of minimizing the error exponent and smoothing exponent of the relaxation time distribution function, a multi-objective genetic optimization algorithm is used to optimize the parameters from the preset search range of regularization parameters and the preset search range of the full width at half maximum of the basis functions, and to determine the non-dominated solution set of the regularization parameters and the full width at half maximum of the basis functions. Multiple preferred solutions are selected from the non-dominated solution set, and the mean of the regularization parameter is calculated based on each preferred solution as the preferred regularization parameter, and the mean of the full width at half maximum (FWHM) of the basis functions is calculated as the preferred full width at half maximum (FWHM) of the basis functions.

4. The method for determining the state of charge of a lithium battery as described in claim 3, characterized in that, The error exponent of the initial relaxation time distribution function is determined using the following formula based on the fitted impedance spectroscopy data and electrochemical impedance spectroscopy data: in, Z' is the error exponent of the relaxation time distribution function. exp (f n Z” represents the real part of the electrochemical impedance spectroscopy data. exp (f n Z' is the imaginary part of the electrochemical impedance spectroscopy data. fit (f n Z” represents the real part of the fitted impedance spectrum data. fit (f n ) represents the imaginary part of the fitted impedance spectrum data.

5. The method for determining the state of charge of a lithium battery as described in claim 3, characterized in that, The smoothing exponent of the initial relaxation time distribution function is determined by the following formula: IDX smth <Sch*100*Var(K(τ)), Among them, K i S is the curvature of the fitted data. ch The number of peaks in the relaxation time distribution, Var() is the variance, and Idx is the peak value. smth τ is the smoothing exponent, τ is the time constant, and g(τ) is the relaxation time distribution function.

6. The method for determining the state of charge of a lithium battery as described in claim 3, characterized in that, The optimization objective is to minimize the error exponent and smoothing exponent of the relaxation time distribution function. A multi-objective genetic optimization algorithm is used to optimize parameters from a preset search range for regularization parameters and a preset search range for the full width at half maximum (FWHM) of the basis functions. This determines the non-dominated solution set for the regularization parameters and the FWHM of the basis functions, specifically including: An initial population is created by randomly selecting parameters from the preset regularization parameter search range and the preset basis function half-peak full width value search range. The error index and smoothness index of the relaxation time distribution of each individual in the population are determined, and the offspring population is obtained through selection, crossover and mutation among individuals in the initial population. An elite retention strategy is used to merge the initial population and the offspring population to obtain a merged population; then the dominance relationship between individuals in the merged population is determined, and the non-dominant solutions and mutually non-dominant solutions are used as the initial optimal solution set. The initial optimal solution set is sorted by fast non-dominated sorting and the crowding distance of each individual is calculated. Based on the crowding distance, individuals are selected to form the updated parent population using a targeted optimization method. The updated offspring population is obtained by selection, crossover and mutation among individuals in the updated parent population, and the error index and smoothness index of the relaxation time distribution of each individual in the updated offspring population are determined. Through multiple iterations until the updated optimal solution set obtained from the updated parent population and the updated child population converges, the converged optimal solution set is used as the non-dominated solution set of the regularization parameter and the full width at half maximum of the basis functions.

7. The method for determining the state of charge of a lithium battery as described in claim 1, characterized in that, The process of determining the relaxation time distribution function of the target lithium battery under different states of charge and fitting it using the least squares method specifically includes: For each preset state of charge of the target lithium battery of the same type, determine the relaxation time distribution function under that state of charge, and extract the characteristic peak height, center frequency and full width at half maximum (FWHM) of the relaxation time distribution function under that state of charge. Based on the peak height, center frequency, and full width at half maximum (FWHM) of the characteristic peaks in the relaxation time distribution function under this charged state, and using Gaussian functions as basis functions, the following formula is used to fit multiple Gaussian functions corresponding to each characteristic peak using the least squares method: Among them, G m (x) is the Gaussian function corresponding to the m-th characteristic peak, f m It is the center frequency corresponding to the m-th characteristic peak, A m Hf is the peak height corresponding to the m-th characteristic peak. m It is the full width at half maximum (FWHM) corresponding to the m-th characteristic peak.

8. The method for determining the state of charge of a lithium battery as described in claim 1, characterized in that, The characteristic values ​​of each characteristic peak are the peak area and peak height of each characteristic peak; The determination of the state of charge (SOC) of the target lithium battery based on the correlation between the eigenvalues ​​of the target characteristic peak in the relaxation time distribution function of the target lithium battery and the SOC specifically includes: The state of charge (SOC) of the target lithium battery is determined by the following formula, based on the correlation between the peak area and peak height of the target characteristic peak in the relaxation time distribution function of the target lithium battery and the SOC: SOC = k1S + k2A + k3Asoh; Where SOC is the state of charge of the target lithium battery, S is the peak area of ​​the target characteristic peak, A is the peak height of the target characteristic peak, Asoh is the ratio of the first difference between the total impedance of the relaxation time distribution function of the target lithium battery and the impedance of the corresponding target characteristic peak to the second difference between the total impedance of the relaxation time distribution function of the target lithium battery under the rated state of charge and the impedance of the corresponding target characteristic peak, and k1, k2 and k3 are coefficients obtained by fitting the characteristic parameters of the characteristic peak.

9. A device for determining the state of charge of a lithium battery, characterized in that, include: The acquisition module is used to acquire electrochemical impedance spectroscopy data of the target lithium battery; Based on the initial regularization parameters, initial basis function full width at half maximum (FWHM), and electrochemical impedance spectroscopy data of the pre-defined ridge regression regularization method, the initial relaxation time distribution function of the target lithium battery is determined using the ridge regression regularization method. The optimization module is used to determine the fitting impedance spectrum data corresponding to the initial relaxation time distribution function. The optimization objectives are to minimize the deviation between the fitting impedance spectrum data and the electrochemical impedance spectrum data and to smooth the initial relaxation time distribution function. The module optimizes the parameters from the preset regularization parameter search range and the preset basis function full width at half maximum value search range to determine the preferred regularization parameter and the preferred basis function full width at half maximum value. The fitting module is used to determine the relaxation time distribution function of the target lithium battery using the ridge regression regularization method based on the preferred regularization parameters, the preferred full width at half maximum of the basis function and the electrochemical impedance spectroscopy data. A selection module is used to determine the relaxation time distribution function of the same type of target lithium battery under different states of charge, and fit it using the least squares method; based on the fitted function, the characteristic values ​​of each characteristic peak of the relaxation time distribution function under different states of charge of the same type of battery are determined, and based on the Pearson correlation coefficient between the characteristic values ​​of each characteristic peak and the corresponding state of charge, the target characteristic peak most related to the change of state of charge of the target lithium battery is determined. The determination module is used to determine the characteristic value of the target characteristic peak in the relaxation time distribution function of the target lithium battery, and to determine the state of charge of the target lithium battery based on the correlation between the characteristic value of the target characteristic peak in the relaxation time distribution function of the target lithium battery and the state of charge.

10. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described in any one of claims 1 to 8.

Images