A battery parameter identification method, system, device and medium based on information amount evaluation

Abstract

The application relates to the technical field of battery management, and discloses a battery parameter identification method and system based on information quantity evaluation, which comprises the following steps: establishing an equivalent circuit model of a battery to be identified; calculating the sensitivity information of a model output to each model parameter; quantitatively analyzing the data validity under different forms of current excitation by using a Fisher information matrix and a CRLB; selecting a composite current excitation form containing a periodic pulse component and a sine component according to the parameter information quantity evaluation result, and optimizing the amplitude, frequency and cycle ratio of the composite current; collecting battery terminal voltage and current data under the action of the composite current excitation, and completing the battery equivalent circuit model parameter identification. The application quantitatively analyzes the identifiability of model parameters under different current excitation conditions, avoids blindly selecting test working conditions, optimally designs the amplitude, frequency and cycle of the excitation current, and can improve the accuracy and stability of overall multi-parameter collaborative identification.

Description

Technical Field

[0001] This invention relates to the field of battery management technology, and in particular to a method, system, device and medium for battery parameter identification based on information content assessment. Background Technology

[0002] In existing battery management systems, equivalent circuit models are widely used for battery state estimation and control strategy design due to their simple structure, low computational cost, and ease of engineering implementation. Equivalent circuit models typically contain multiple parameters to be identified, such as capacity parameters, ohmic internal resistance parameters, and polarization parameters. The accuracy of parameter identification directly affects the model's estimation performance and the operational reliability of the battery management system.

[0003] The effectiveness of battery model parameter identification depends not only on the model structure and the identification algorithm itself, but also on the amount of information contained in the test data. Even if the model structure is theoretically identifiable, if the test data fails to provide effective stimulation to the model parameters, the parameter identification results may still be unstable or have large errors. Therefore, by rationally designing the current excitation method to increase the amount of effective information about the model parameters in the test data, the performance of model parameter identification can be improved.

[0004] However, existing battery model parameter identification methods generally employ pre-defined test conditions, such as Hybrid Power Performance Testing (HPPC) or Dynamic Stress Testing (DST). These methods typically use fixed current pulses or predefined dynamic load conditions to obtain model parameters by fitting the battery terminal voltage response. Since test conditions are largely selected based on experience, it is difficult to quantitatively evaluate the excitation effects of different parameters before the experiment. Whether the obtained data contains sufficient information often lacks clear criteria, making it difficult to guarantee the reliability and stability of the parameter identification results. On the other hand, to improve parameter identification accuracy, it is usually necessary to increase the current excitation intensity or expand the current variation range to amplify the battery terminal voltage response characteristics to parameter changes. However, excessive current may lead to battery overheating, accelerated performance degradation, and even safety risks. Therefore, the design of current excitation must be limited by constraints such as the maximum allowable current and the range of state of charge variation. Summary of the Invention

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] Therefore, this invention provides a battery parameter identification method and system based on information content evaluation to solve the problem that current traditional methods mainly rely on experience to select test conditions, lack quantitative evaluation and guidance of parameter identification information in the data, and are difficult to fully guarantee the accuracy and stability of multi-parameter identification within safety constraints.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a battery parameter identification method based on information content evaluation, comprising: Obtain battery parameters and establish an equivalent circuit model of the battery to be identified; Based on the equivalent circuit model, calculate the sensitivity information of the model output to each model parameter; Based on sensitivity information, a parameter information content evaluation index was constructed. Fisher information matrix and CRLB were used to quantitatively analyze the data validity under different forms of current excitation, so as to evaluate the degree of influence of different excitation forms on the identification of parameters of each model. Based on the parameter information evaluation results, a composite current excitation form containing periodic pulse components and sinusoidal components is selected, and the amplitude, frequency and period ratio of the composite current are optimized. Battery terminal voltage and current data are collected under the combined current excitation, and the parameters of the battery equivalent circuit model are identified based on the collected data.

[0008] As a preferred embodiment of the battery parameter identification method based on information content evaluation described in this invention, the step of establishing an equivalent circuit model of the battery to be identified includes: A first-order resistor-capacitor equivalent circuit model is constructed. The model parameters include capacity parameters, ohmic internal resistance parameters, and polarization branch parameters used to describe the dynamic polarization characteristics of the battery. Based on the model parameters, the state vector and parameter vector are defined. The system input is the load current, and the system output is the battery terminal voltage.

[0009] As a preferred embodiment of the battery parameter identification method based on information content evaluation described in this invention, the method employs Fisher information matrix and CRLB to quantitatively analyze the data validity under different forms of current excitation, including: Based on the principle of maximum likelihood estimation, let the observed noise of the battery terminal voltage have zero mean and variance. The independent and identically distributed random noise is used to obtain the joint probability density function of the observed voltage sequence; The Fisher information matrix is ​​calculated based on the joint probability density function. The Fisher information matrix is ​​defined as the statistical expectation of the log-likelihood function with respect to the parameter gradient, and is expressed as a weighted cumulative form of the model output with respect to the parameter sensitivity. The optimal estimates of the model parameters are obtained by solving the extrema of the log-likelihood function; Based on the optimal estimate and the diagonal elements of the inverse of the Fisher information matrix, the Cramér-Rao lower bounds of each model parameter under unbiased estimation conditions are determined. The Cramér-Rao lower bounds are used to characterize the theoretical minimum value that the variance of the estimation error of each parameter can reach.

[0010] As a preferred embodiment of the battery parameter identification method based on information content evaluation described in this invention, the method includes: selecting a composite current excitation form containing periodic pulse components and sinusoidal components based on the parameter information content evaluation results, and optimizing the amplitude, frequency, and period ratio of the composite current, including: Construct a composite current excitation formed by the superposition of periodic pulse components and sinusoidal components; Under periodic or continuous current excitation conditions, a time-averaged Fisher information matrix is ​​constructed by averaging the Fisher information matrix in the time domain to evaluate the comprehensive identifiability of model parameters throughout the entire excitation cycle. When the rate of change of the battery's open-circuit voltage-state-of-charge curve is lower than a preset threshold within the target operating range, the composite current excitation is repeated over multiple cycles, and the elements related to the battery capacity parameters in the time-averaged Fisher information matrix are corrected.

[0011] As a preferred embodiment of the battery parameter identification method based on information content evaluation described in this invention, the method further includes: selecting a composite current excitation form containing periodic pulse components and sinusoidal components based on the parameter information content evaluation results, and optimizing the amplitude, frequency, and period ratio of the composite current; Normalize the time-averaged Fisher information matrix; The optimization objective is to maximize the minimum eigenvalue of the normalized average Fisher information matrix. An optimization problem is constructed for the submatrices corresponding to the ohmic internal resistance parameters and polarization branch parameters. The pulse current amplitude and the ratio coefficient of the pulse period to the sine period are optimized.

[0012] As a preferred embodiment of the battery parameter identification method based on information content evaluation described in this invention, the method further includes: finding an optimal solution for the optimization objective; The solution is obtained under the premise of satisfying the preset constraints. Among them, the instantaneous amplitude of the composite excitation current shall not exceed the maximum allowable current of the battery; The sinusoidal current angular frequency satisfies the excitation requirements of the dynamic response characteristics of the polarization branch and is limited by the sampling frequency of the battery testing system; During multi-cycle composite excitation, the cumulative change in the battery's state of charge does not exceed the preset maximum change in state of charge.

[0013] As a preferred embodiment of the battery parameter identification method based on information content evaluation described in this invention, the method further includes: finding the optimal solution for the optimization objective. The genetic algorithm is used to find the optimal solution for the variable to be optimized. After obtaining the optimal values ​​of the pulse current amplitude and the ratio of the pulse period to the sine period, the change in the state of charge within a single excitation cycle is calculated. The number of repetition cycles of the composite current excitation is determined based on the maximum allowable change in state of charge, and the amplitude of the sinusoidal current is determined based on the difference between the maximum allowable current of the battery and the amplitude of the pulse current.

[0014] Secondly, the present invention provides a battery parameter identification system based on information content evaluation, comprising: The model building module is used to obtain battery parameters and establish an equivalent circuit model of the battery to be identified. The sensitivity information calculation module is used to calculate the sensitivity information of the model output to each model parameter based on the equivalent circuit model. The analysis module is used to construct parameter information evaluation index based on sensitivity information. It uses Fisher information matrix and CRLB to quantitatively analyze the data validity under different forms of current excitation, so as to evaluate the degree of influence of different excitation forms on the identification of parameters of each model. The optimization module is used to select a composite current excitation form containing periodic pulse components and sinusoidal components based on the evaluation results of parameter information, and to optimize the amplitude, frequency and period ratio of the composite current. The identification module is used to collect battery terminal voltage and current data under composite current excitation, and to identify the parameters of the battery equivalent circuit model based on the collected data.

[0015] Thirdly, the present invention provides a computer device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of a battery parameter identification method based on information content evaluation.

[0016] Fourthly, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the battery parameter identification method based on information content evaluation.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: by introducing an information quantity evaluation mechanism that reflects the identifiability of parameters, and using the Fisher information matrix and the Cramér-Rao lower bound (CRLB) of parameter estimation, the identifiability of model parameters under different current excitation conditions is quantitatively analyzed, avoiding blind selection of test conditions; and by optimizing the design of the amplitude, frequency and period of the excitation current, the identification accuracy of capacity parameters, ohmic internal resistance parameters and polarization parameters can be improved, further improving the accuracy and stability of overall multi-parameter collaborative identification. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the overall process of a battery parameter identification method based on information content evaluation according to an embodiment of the present invention.

[0020] Figure 2 This is a schematic diagram of a first-order RC equivalent circuit model structure in a battery parameter identification method based on information content evaluation according to an embodiment of the present invention.

[0021] Figure 3 This is a battery terminal voltage prediction result and corresponding prediction error diagram in a battery parameter identification method based on information content evaluation according to an embodiment of the present invention.

[0022] Figure 4 This is a waveform diagram of periodic pulse current, sinusoidal alternating current, and the composite current of the two in a battery parameter identification method based on information content evaluation according to an embodiment of the present invention. Detailed Implementation

[0023] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0024] Example 1, referring to Figure 1This invention provides a battery parameter identification method based on information content evaluation, illustrating the overall flowchart of the battery equivalent model parameter identification and excitation design method based on information content evaluation. The method includes: S100: Obtain battery parameters and establish an equivalent circuit model of the battery to be identified; In this embodiment of the application, the step of establishing the equivalent circuit model of the battery to be identified in S100 includes: A first-order resistor-capacitor equivalent circuit model is constructed. The model parameters include capacity parameters, ohmic internal resistance parameters, and polarization branch parameters used to describe the dynamic polarization characteristics of the battery. Based on the model parameters, the state vector and parameter vector are defined. The system input is the load current, and the system output is the battery terminal voltage.

[0025] Specifically, S100 uses a first-order RC equivalent circuit model, and its state and observation equations are as follows: in, For battery capacity, The voltage across the polarization branch, Terminal voltage, This is the load current, which is positive during charging. Open circuit voltage, For ohm resistance, Polarization resistor, It is a polarizing capacitor.

[0026] Furthermore, in S100, the state vector is defined as... The parameter vector is defined as The system input and output are defined as follows: and Then the equation can be expressed as: in, It is the mapping function of OCV-SOC. It measures noise.

[0027] S200: Based on the equivalent circuit model, calculate the sensitivity information of the model output to each model parameter; Specifically, the sensitivity information of the calculation model parameters can be expressed by the following formula: in, It is the sampling time; It is the index of the parameter to be identified in the parameter vector; Given parameters The predicted voltage value is calculated at that time.

[0028] It should be noted that the greater the parameter sensitivity, the more significant the impact of parameter changes on the system output, and the easier it is to accurately identify.

[0029] S300: Based on sensitivity information, a parameter information content evaluation index is constructed. Fisher information matrix and CRLB are used to quantitatively analyze the data validity under different forms of current excitation, so as to evaluate the degree of influence of different excitation forms on the identification of parameters of each model.

[0030] In this embodiment of the application, S300 uses Fisher information matrix and CRLB to quantitatively analyze the validity of data under different forms of current excitation, including the following steps A1-A4: A1: Based on the principle of maximum likelihood estimation, assume the battery terminal voltage observation noise has zero mean and variance. The independent and identically distributed random noise is used to obtain the joint probability density function of the observed voltage sequence; Specifically, in A1, a parameter estimation model is constructed using the maximum likelihood estimation principle. It is assumed that the battery terminal voltage observation noise has zero mean and variance. Given independent and identically distributed random noise, the joint probability density function of the observed voltage sequence can be expressed as: in, It is the first The measured battery terminal voltage at each sampling time. This represents the total number of sampling points.

[0031] A2: Calculate the Fisher information matrix based on the joint probability density function. The Fisher information matrix is ​​defined as the statistical expectation of the log-likelihood function with respect to the parameter gradient, and is expressed as a weighted cumulative form of the model output with respect to the parameter sensitivity. Specifically, the statistical expectation in A2 can be expressed as: make Representative moment The observation noise can be obtained from the joint probability density function, which gives the corresponding log-likelihood function. Further, the observation noise of the th... Taking the partial derivative with respect to each parameter yields: Since the observation noise satisfies zero mean and variance is The independent and identically distributed condition is as follows: Substituting the above results into the definition of the Fisher information matrix, the Fisher information matrix can be expressed as a weighted cumulative form of the model output's sensitivity to the parameters: It should be noted that the larger the value of the diagonal element of the Fisher information matrix, the more sensitive the corresponding model parameter is to changes in battery terminal voltage under the current excitation conditions, and the easier it is for the parameter to be accurately identified.

[0032] A3: The optimal estimates of the model parameters are obtained by solving the extreme values ​​of the log-likelihood function; Specifically, the optimal estimate of A3 The calculation process can be represented as follows: A4: Based on the optimal estimate and the diagonal elements of the inverse of the Fisher information matrix, determine the Cramér-Rao lower bound of each model parameter under the unbiased estimation condition. The Cramér-Rao lower bound is used to characterize the theoretical minimum value that the variance of the estimation error of each parameter can reach. Specifically, in A4, the CRLB of the parameter estimation error is obtained from the inverse of the Fisher information matrix, and is expressed as: Under ideal conditions of unbiased estimation, CRLB represents the theoretical minimum value that the variance of parameter estimation error can reach, and can be used as an important indicator to evaluate the upper limit of parameter identification accuracy under current excitation conditions.

[0033] S400: Based on the parameter information evaluation results, select a composite current excitation form that includes periodic pulse components and sinusoidal components, and optimize the amplitude, frequency and period ratio of the composite current. In this embodiment, S400 selects a composite current excitation form containing periodic pulse components and sinusoidal components based on the parameter information evaluation results, and optimizes the amplitude, frequency, and period ratio of the composite current, including the following steps B1-B3: B1: Construct a composite current excitation formed by the superposition of periodic pulse components and sinusoidal components; Specifically, the composite current excitation in B1 can be expressed as: in, This refers to the amplitude of the pulse current. The amplitude of the sinusoidal current; The angular frequency of the sinusoidal current; Pulse period With sine period The proportionality coefficient between them satisfies .

[0034] It should be noted that the above parameters , , , As a variable to be optimized.

[0035] B2: Under periodic or continuous current excitation conditions, a time-averaged Fisher information matrix is ​​constructed by averaging the Fisher information matrix in the time domain to evaluate the comprehensive identifiability of model parameters throughout the entire excitation cycle. Specifically, the elements of the time-averaged Fisher information matrix in B2 can be calculated in the following way: B3: In response to the battery's open-circuit voltage-state-of-charge (OCV-SOC) curve changing at a rate lower than a preset threshold within the target operating range, the composite current excitation is repeated over multiple cycles, and the elements related to the battery capacity parameters in the time-averaged Fisher information matrix are corrected.

[0036] Specifically, the number of cycles in a multi-cycle process can be determined based on the ratio of the maximum allowable change in state of charge to the change in state of charge within a single excitation cycle, and the maximum integer number of cycles not exceeding this ratio is taken.

[0037] For example, in B3, when the rate of change of the battery's open-circuit voltage-state-of-charge (OCV-SOC) curve within the target operating range is lower than a preset threshold, it means that the OCV-SOC curve is flat. The preset threshold needs to be set according to different battery characteristics. For instance, in specific implementations, the preset threshold can be set according to the battery type and operating range. For lithium iron phosphate batteries, a flat OCV-SOC curve slope of less than 0.1V / 1%SOC can be considered flat; for ternary lithium batteries, this threshold can be adjusted accordingly. Those skilled in the art can set this threshold themselves based on the battery's OCV-SOC characteristic curve and the required identification accuracy.

[0038] It should be noted that the correction object in B3 is the element in the time-averaged Fisher information matrix that is related to the capacity parameter θ1 (see Table 3 in Example 2 for an exemplary case), and its purpose is to enhance the amount of information corresponding to the capacity parameter.

[0039] In this embodiment of the application, S400 selects a composite current excitation form containing periodic pulse components and sinusoidal components based on the parameter information evaluation results, and optimizes the amplitude, frequency, and period ratio of the composite current. It also includes the following steps B4-B5: B4: Normalize the time-averaged Fisher information matrix; Specifically, in B4, to ensure that the Fisher information matrix can comprehensively measure the uncertainty of parameters of different orders of magnitude during the identification process, the time-averaged Fisher information matrix is ​​normalized, and its matrix elements are as follows: B5: The optimization objective is to maximize the minimum eigenvalue of the normalized average Fisher information matrix, and to construct an optimization problem for the submatrix corresponding to the ohmic internal resistance parameter and the polarization branch parameter, thereby optimizing the pulse current amplitude and the ratio coefficient of the pulse period to the sine period. The larger the minimum eigenvalue, the stronger the overall identifiability of the model parameters under the current excitation conditions, and the smaller the lower bound of the parameter estimation error.

[0040] Specifically, since the CRLB and Fisher information matrix are inverses of each other, the larger the minimum eigenvalue of the normalized Fisher information matrix, the smaller the lower bound of the parameter estimation error, and the stronger the overall identifiability of the parameters. Therefore, the excitation parameters are optimized by maximizing the minimum eigenvalue of the normalized Fisher information matrix. Based on this, an optimization problem is constructed using the submatrices corresponding to the ohmic resistance parameters and polarization branch parameters in the normalized Fisher information matrix to optimize the pulse current amplitude. and cycle ratio Optimize: in, The maximum allowable current of the battery. The maximum permissible change in state of charge. For the battery's rated capacity, The polarization branch time constant is used to ensure that the instantaneous amplitude of the excitation current does not exceed the maximum allowable current of the battery, the cumulative change in the state of charge of the multi-cycle composite excitation does not exceed the preset maximum change in the state of charge, and the sinusoidal current angular frequency is subject to the combined constraints of the polarization branch time constant and the sampling frequency of the battery test system.

[0041] In this embodiment of the application, S400 further includes step B6: finding an optimal solution for the optimization objective; The solution is obtained under the premise of satisfying the preset constraints. Among them, the instantaneous amplitude of the composite excitation current shall not exceed the maximum allowable current of the battery; The sinusoidal current angular frequency satisfies the excitation requirements of the dynamic response characteristics of the polarization branch and is limited by the sampling frequency of the battery testing system; For example, the upper limit of the angular frequency of a sinusoidal current. The value should be determined based on the sampling frequency, and should not exceed 1 / 10 to 1 / 5 of the sampling frequency to ensure that only limited information loss occurs during the sampling process. A suitable value can be selected. , or Considering the time constant of the RC network There are also predictable reference values, therefore the lower limit... Set as .

[0042] During multi-cycle composite excitation, the cumulative change in the battery's state of charge does not exceed the preset maximum change in state of charge.

[0043] In this embodiment of the application, B6 of S400 further includes: The genetic algorithm is used to find the optimal solution for the variable to be optimized. After obtaining the optimal values ​​of the pulse current amplitude and the ratio of the pulse period to the sine period, the change in the state of charge within a single excitation cycle is calculated. The number of repetition cycles of the composite current excitation is determined based on the maximum allowable change in state of charge, and the amplitude of the sinusoidal current is determined based on the difference between the maximum allowable current of the battery and the amplitude of the pulse current.

[0044] Specifically, in B6, the genetic algorithm solves the problem. and After finding the optimal value, the change in state of charge over one cycle can be calculated. Then calculate the number of excitation cycles. , .

[0045] S500: Acquires battery terminal voltage and current data under composite current excitation, and identifies battery equivalent circuit model parameters based on the acquired data.

[0046] After obtaining the optimal combination of excitation parameters, the composite current excitation is applied to the battery model, and the model parameter identification process is simulated to verify the effectiveness of the excitation design method in improving the accuracy and stability of parameter identification.

[0047] Example 2, based on the above implementation method, referring to Figures 2-4 The above embodiment is described in further detail as one example of the present invention. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0048] like Figure 2As shown, based on the method in Example 1, this example establishes a first-order RC equivalent circuit model to describe the dynamic response characteristics of the battery terminal voltage. The model includes an open-circuit voltage source, an ohmic internal resistance, and a polarization branch composed of a polarization resistor and a polarization capacitor.

[0049] In this embodiment, the least squares method is used to identify the battery parameters offline under both DST and PCT operating conditions. The predicted terminal voltage results and corresponding prediction errors are as follows: Figure 3 As shown in Table 1, the parameter identification results and corresponding time constants under different charging states are shown in Table 2, and the model voltage fitting error index is shown in Table 2.

[0050] Table 1: Parameter identification results and corresponding time constants under different states of charge.

[0051] Table 2: Model Voltage Fitting Error Indicators

[0052] Since the parameters in the 40%–60% SOC range are relatively close, the RMSE and MAE are lower, and the slope of the OCV-SoC curve is also larger, it indicates that the parameters of the first-order ECM are more stable in this range, which is conducive to the identification of battery capacity-related parameters. Therefore, the equivalent circuit model in the 40%–60% SOC range is selected as the benchmark model for subsequent current excitation information evaluation and excitation trajectory optimization.

[0053] The baseline parameters of the model are selected as the average values ​​of the PCT condition identification results under 40% SOC and 60% SOC conditions, i.e. , , Within the SOC range, the relationship between open-circuit voltage and state of charge is fitted using a linear function, the expression of which is: Based on this, the parameter information content under different forms of current excitation is evaluated. Specifically, given the model structure and parameters, the response characteristics of the battery terminal voltage to changes in various model parameters are analyzed, and the Fisher information matrix of the current is calculated.

[0054] In this embodiment, periodic pulse current, single-frequency sinusoidal current, and composite current excitation formed by the superposition of pulse current and sinusoidal current are constructed respectively, and the relevant parameter information contained therein is analyzed.

[0055] For a periodic pulse current, its corresponding time-averaged Fisher information matrix can be expressed in the following form: in, It is the linearized slope of the OCV-SOC curve within the selected state of charge range; This is the time constant of the polarization branch. From the structure of this information matrix, it can be seen that as the pulse period T increases, the parameters related to the polarization resistance... The corresponding amount of information gradually increases, while the parameters related to polarization capacitance... Then the amount of information decreases.

[0056] For a sinusoidal current, its corresponding time-averaged Fisher information matrix can be expressed in the following form: As can be seen from the matrix, different model parameters exhibit different sensitivity characteristics to current frequency. Among them, the capacitance parameter and polarization resistance parameter can obtain more information at lower frequency conditions, while the identifiability of the ohmic internal resistance parameter is basically unchanged with frequency, and the polarization capacitance parameter reaches the maximum information value near a specific frequency related to the RC time constant.

[0057] Based on the above analysis, it can be inferred that using only pulsed current is insufficient to provide sufficient information for multi-parameter identification. Other forms of current excitation need to be introduced to improve the identification effect of polarization resistance and polarization capacitance parameters.

[0058] Compared to the exponentially decaying response caused by step current, sinusoidal current can continuously excite the dynamic behavior of polarization branches under steady-state conditions, thereby helping to enhance the identifiability of relevant parameters. At the same time, the generation method of sinusoidal current is simple, and it can meet the excitation requirements of the time constant of the actual battery RC network in a low frequency range.

[0059] Therefore, this embodiment constructs a composite current excitation form formed by the superposition of periodic pulse current and sinusoidal current. (Refer to...) Figure 4 The diagram illustrates the waveforms of periodic pulse current, sinusoidal alternating current, and their combined current. Since the equivalent circuit model can be approximated as a linear system within the operating range, the voltage response under combined current excitation can be considered as the superposition of individual excitation responses, thus reducing the complexity of parametric excitation analysis.

[0060] Considering that capacity parameters are mainly reflected through the OCV-SOC relationship, their identifiability depends on the cumulative effect of state-of-charge changes during excitation. When the OCV-SOC curve is flat, the state-of-charge changes within a single excitation cycle are insufficient to produce a significant OCV difference. Therefore, multi-cycle composite current excitation can be used to enhance capacity parameters. The corresponding amount of information. In this case, the Fisher information matrix used in the parameter information content evaluation is no longer limited to a single period interval. Instead of averaging over time, it extends to... The time average is performed over the interval, where This represents the total number of excitation cycles. When a sinusoidal current is applied, the steady-state component of each element in the parameter sensitivity vector is also a sinusoidal function. The number of current cycles remains constant. However, the ohmic internal resistance parameters and polarization branch parameters are also periodic; therefore, only the current cycle number needs to be adjusted. The coefficients related to the capacity parameters were corrected, and the correction results are shown in Table 3.

[0061] Table 3: Coefficient Correction Results

[0062] Furthermore, to eliminate the impact of differences in the dimensions and orders of magnitude of different parameters on the evaluation results of parameter identifiability, the time-averaged Fisher information matrix is ​​normalized.

[0063] When multiple parameters are identified simultaneously, due to the coupling relationship between the parameters, the diagonal elements of the Fisher information matrix can no longer accurately quantify the actual identifiability of the corresponding parameters. Since the CRLB and the Fisher information matrix are inverses of each other, the larger the minimum eigenvalue of the normalized Fisher information matrix, the smaller the lower bound of the corresponding parameter estimation error, and the stronger the overall identifiability of the parameters.

[0064] Therefore, the minimum eigenvalue of the normalized Fisher information matrix is ​​used as the overall identifiability evaluation index. The larger the minimum eigenvalue, the richer the effective identifiability information in the test data.

[0065] During the optimization of excitation parameters, constraints are applied to the composite current excitation: First, the pulse current amplitude and sinusoidal current amplitude The sum cannot exceed the battery's maximum allowable current. Secondly, the angular frequency of the sinusoidal current is limited by the sampling frequency range of the test system, with the upper limit of the angular frequency being... The value should be determined based on the sampling frequency, and should not exceed 1 / 10 to 1 / 5 of the sampling frequency to ensure that only limited information loss occurs during the sampling process. A suitable value can be selected. , or Considering the time constant of the RC network There are also predictable reference values, therefore the lower limit... Set as This ensures that an effective optimal solution exists within the frequency range. Furthermore, the cumulative change in the state of charge of multi-cycle composite excitation does not exceed the preset maximum change in the state of charge.

[0066] According to the expression for the time-averaged Fisher information matrix under sinusoidal current excitation, when the angular frequency... Polarization capacitance parameters at a specific frequency The corresponding information content reaches its maximum value. Therefore, for the identification of polarization branch parameters, the optimal sinusoidal excitation angular frequency satisfies: because The identifiability is only affected by the cumulative capacity change under load current excitation, therefore it can be initially determined based on the submatrices of the normalized Fisher information matrix. Finish and The optimization problem described above is a mixed integer optimization problem involving both continuous and discrete variables, which can be solved using a genetic algorithm. The optimal pulse current amplitude is obtained. and cycle ratio Then, the amplitude of the sinusoidal current is determined based on the maximum allowable current constraint. The number of excitation cycles is determined based on the allowable range of changes in the state of charge. .

[0067] Furthermore, after obtaining the optimal combination of excitation parameters, the composite current excitation can be applied to the battery model, and the model parameter identification process can be simulated and verified to validate the effectiveness of the excitation design method in improving the accuracy and stability of parameter identification.

[0068] Example 3 illustrates a schematic scheme for a battery parameter identification method based on information content evaluation. It should be noted that the technical solution of this battery parameter identification system based on information content evaluation belongs to the same concept as the technical solution of the battery parameter identification method based on information content evaluation described above. Details not described in detail in this embodiment can be found in the description of the technical solution of the battery parameter identification method based on information content evaluation described above.

[0069] This embodiment provides a battery parameter identification system based on information content evaluation, including: The model building module is used to obtain battery parameters and establish an equivalent circuit model of the battery to be identified. The sensitivity information calculation module is used to calculate the sensitivity information of the model output to each model parameter based on the equivalent circuit model. The analysis module is used to construct parameter information evaluation index based on sensitivity information. It uses Fisher information matrix and CRLB to quantitatively analyze the data validity under different forms of current excitation, so as to evaluate the degree of influence of different excitation forms on the identification of parameters of each model. The optimization module is used to select a composite current excitation form containing periodic pulse components and sinusoidal components based on the evaluation results of parameter information, and to optimize the amplitude, frequency and period ratio of the composite current. The identification module is used to collect battery terminal voltage and current data under composite current excitation, and to identify the parameters of the battery equivalent circuit model based on the collected data.

[0070] This embodiment also provides a computer device applicable to a battery parameter identification method based on information content evaluation, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement a battery parameter identification method based on information content evaluation as proposed in the above embodiment.

[0071] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a battery parameter identification method based on information content evaluation as proposed in the above embodiments.

[0072] The storage medium proposed in this embodiment and the battery parameter identification method based on information content evaluation proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0073] From the above description of the implementation methods, those skilled in the art will clearly understand that the present invention can be implemented using software and necessary general-purpose hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0074] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A battery parameter identification method based on information content evaluation, characterized in that, include: Obtain battery parameters and establish an equivalent circuit model of the battery to be identified; Based on the equivalent circuit model, calculate the sensitivity information of the model output to each model parameter; Based on sensitivity information, a parameter information content evaluation index was constructed. Fisher information matrix and CRLB were used to quantitatively analyze the data validity under different forms of current excitation, so as to evaluate the degree of influence of different excitation forms on the identification of parameters of each model. Based on the parameter information evaluation results, a composite current excitation form containing periodic pulse components and sinusoidal components is selected, and the amplitude, frequency and period ratio of the composite current are optimized. Battery terminal voltage and current data are collected under the combined current excitation, and the parameters of the battery equivalent circuit model are identified based on the collected data.

2. The battery parameter identification method based on information content evaluation as described in claim 1, characterized in that, The process of establishing an equivalent circuit model of the battery to be identified includes: A first-order resistor-capacitor equivalent circuit model is constructed. The model parameters include capacity parameters, ohmic internal resistance parameters, and polarization branch parameters used to describe the dynamic polarization characteristics of the battery. Based on the model parameters, the state vector and parameter vector are defined. The system input is the load current, and the system output is the battery terminal voltage.

3. The battery parameter identification method based on information content evaluation as described in claim 2, characterized in that, Fisher information matrix and CRLB were used to quantitatively analyze the validity of data under different forms of current excitation, including: Based on the principle of maximum likelihood estimation, let the observed noise of the battery terminal voltage have zero mean and variance. The independent and identically distributed random noise is used to obtain the joint probability density function of the observed voltage sequence; The Fisher information matrix is ​​calculated based on the joint probability density function. The Fisher information matrix is ​​defined as the statistical expectation of the log-likelihood function with respect to the parameter gradient, and is expressed as a weighted cumulative form of the model output with respect to the parameter sensitivity. The optimal estimates of the model parameters are obtained by solving the extrema of the log-likelihood function; Based on the optimal estimate and the diagonal elements of the inverse of the Fisher information matrix, the Cramér-Rao lower bounds of each model parameter under unbiased estimation conditions are determined. The Cramér-Rao lower bounds are used to characterize the theoretical minimum value that the variance of the estimation error of each parameter can reach.

4. The battery parameter identification method based on information content evaluation as described in claim 3, characterized in that, Based on the parameter information evaluation results, a composite current excitation form containing periodic pulse components and sinusoidal components is selected, and the amplitude, frequency, and period ratio of the composite current are optimized, including: Construct a composite current excitation formed by the superposition of periodic pulse components and sinusoidal components; Under periodic or continuous current excitation conditions, a time-averaged Fisher information matrix is ​​constructed by averaging the Fisher information matrix in the time domain to evaluate the comprehensive identifiability of model parameters throughout the entire excitation cycle. When the rate of change of the battery's open-circuit voltage-state-of-charge curve is lower than a preset threshold within the target operating range, the composite current excitation is repeated over multiple cycles, and the elements related to the battery capacity parameters in the time-averaged Fisher information matrix are corrected.

5. A battery parameter identification method based on information content evaluation as described in claim 1 or 4, characterized in that, Based on the parameter information evaluation results, a composite current excitation form containing periodic pulse components and sinusoidal components is selected, and the amplitude, frequency, and period ratio of the composite current are optimized. This also includes: Normalize the time-averaged Fisher information matrix; The optimization objective is to maximize the minimum eigenvalue of the normalized average Fisher information matrix. An optimization problem is constructed for the submatrices corresponding to the ohmic internal resistance parameters and polarization branch parameters. The pulse current amplitude and the ratio coefficient of the pulse period to the sine period are optimized.

6. The battery parameter identification method based on information content evaluation as described in claim 5, characterized in that, It also includes: finding the optimal solution for the optimization objective; The solution is obtained under the premise of satisfying the preset constraints. Among them, the instantaneous amplitude of the composite excitation current shall not exceed the maximum allowable current of the battery; The sinusoidal current angular frequency satisfies the excitation requirements of the dynamic response characteristics of the polarization branch and is limited by the sampling frequency of the battery testing system; During multi-cycle composite excitation, the cumulative change in the battery's state of charge does not exceed the preset maximum change in state of charge.

7. The battery parameter identification method based on information content evaluation as described in claim 6, characterized in that, The optimization objective also includes: The genetic algorithm is used to find the optimal solution for the variable to be optimized. After obtaining the optimal values ​​of the pulse current amplitude and the ratio of the pulse period to the sine period, the change in the state of charge within a single excitation cycle is calculated. The number of repetition cycles of the composite current excitation is determined based on the maximum allowable change in state of charge, and the amplitude of the sinusoidal current is determined based on the difference between the maximum allowable current of the battery and the amplitude of the pulse current.

8. A battery parameter identification system based on information content evaluation, using the method described in any one of claims 1-7, characterized in that, include: The model building module is used to obtain battery parameters and establish an equivalent circuit model of the battery to be identified. The sensitivity information calculation module is used to calculate the sensitivity information of the model output to each model parameter based on the equivalent circuit model. The analysis module is used to construct parameter information evaluation index based on sensitivity information. It uses Fisher information matrix and CRLB to quantitatively analyze the data validity under different forms of current excitation, so as to evaluate the degree of influence of different excitation forms on the identification of parameters of each model. The optimization module is used to select a composite current excitation form containing periodic pulse components and sinusoidal components based on the evaluation results of parameter information, and to optimize the amplitude, frequency and period ratio of the composite current. The identification module is used to collect battery terminal voltage and current data under composite current excitation, and to identify the parameters of the battery equivalent circuit model based on the collected data.

9. A computer device, characterized in that, include: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the battery parameter identification method based on information content evaluation as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores computer-executable instructions that, when executed by a processor, implement the steps of the battery parameter identification method based on information content evaluation as described in any one of claims 1 to 7.

Images