CN-122017631-A - Method for determining state of charge of battery

CN122017631ACN 122017631 ACN122017631 ACN 122017631ACN-122017631-A

Abstract

The application relates to the field of battery state detection, in particular to a method for determining a battery state of charge. The application collects the mixed power pulse data of different temperature intervals and constructs a corresponding state equation, thereby making up the neglect of a single normal temperature/high temperature model on a low temperature working condition, solving the problem of insufficient model precision under low temperature and high internal resistance, secondly, realizing the effective fusion of RC models of different temperatures by weighting and integrating a plurality of temperature state equations through temperature related weights, fully utilizing battery characteristic parameters under each temperature, then, replacing noise matrix optimization by weight optimization, avoiding the algorithm stability defect caused by noise uncertainty, and finally, carrying out Kalman filtering estimation on SOC based on a target equivalent state equation, thereby not only ensuring the estimation precision under the full temperature working condition, but also improving the algorithm stability, and effectively solving the technical pain point that the prior scheme cannot be used for both the precision and the stability of the full temperature range.

Inventors

LU ZETONG
JIA MEI
PAN DONGHUA

Assignees

固德威技术股份有限公司

Dates

Publication Date: 20260512
Application Date: 20251231

Claims (10)

1. A method of determining a state of charge of a battery, the method comprising: Collecting mixed power pulse data of the battery at different temperatures; Calculating battery characteristic parameters of the battery at different temperatures based on the mixed power pulse data at different temperatures to obtain state equations at different temperatures; The state equations at different temperatures are weighted and integrated through the temperature-related weights, so that an equivalent state equation is obtained; Obtaining a target weight, substituting the target weight into the equivalent state equation to obtain a target equivalent state equation, and performing Kalman filtering equation calculation based on the target equivalent state equation to obtain the state of charge of the battery, wherein the target weight is used for representing the proportion of the state equation corresponding to different temperatures in the equivalent state equation.
2. The method of claim 1, wherein calculating battery characteristic parameters of the battery at different temperatures based on the mixed power pulse data at different temperatures to obtain state equations at different temperatures comprises: calculating battery characteristic parameters of the battery at different temperatures based on the mixed power pulse data at different temperatures; Constructing an equation frame containing input and output relations according to battery characteristic parameters at different temperatures, and supplementing process noise and observation noise items at different temperatures to the equation frame to obtain an initial state equation at corresponding temperatures; And constructing a corresponding rule among input, output and noise in the initial state equation to obtain state equations at different temperatures.
3. The method of claim 1, wherein prior to acquiring the target weight, the method further comprises: Acquiring an actual measurement voltage corresponding to the battery and a predicted voltage of a battery model, and acquiring a theoretical weight corresponding to the current temperature, an estimated weight at the last moment and an adjusting strength; and constructing an objective function based on the actual measured voltage, the battery model predicted voltage, the theoretical weight, the estimated weight at the last moment and the adjusting strength.
4. A method according to claim 3, wherein the obtaining the target weight comprises: based on the objective function, determining a minimized solving direction taking the weight as an optimized variable to obtain a to-be-solved The problem of weight optimization of solutions; solving the weight optimization problem through a recursive least square method to obtain the target weight.
5. A method according to claim 3, characterized in that the expression of the objective function is as follows: ; Where J is the objective function, In order to actually measure the voltage, the voltage is, The battery model predicts the voltage of the battery, As the current weight is to be given, The weight of the theory is that, 、 The strength of the steel is regulated, the strength is regulated, As the weight at the i-th moment, Is the weight at time i-1.
6. The method of claim 1, wherein the performing a kalman filter equation calculation based on the target equivalent state equation to obtain the state of charge of the battery comprises: acquiring a historical state of charge, input parameters and an error covariance matrix at the previous moment; Predicting according to the historical state of charge, the input parameters and the target equivalent state equation to obtain a state predicted value at the current moment, and determining a priori error covariance matrix at the current moment by utilizing the error covariance matrix; calculating Kalman filtering gain according to the prior error covariance matrix and the matrix in the target equivalent state equation; Updating the state based on the Kalman filtering gain, the state pre-estimation value and the actual measurement value of the current moment to obtain a state optimal estimation value, and determining a posterior error covariance matrix of the current moment by using a prior error covariance matrix, wherein the posterior error covariance matrix is used as an error covariance matrix of the current moment when the state of charge is determined at the next moment; and extracting a first element in the state optimal estimated value to obtain the state of charge of the battery.
7. The method of claim 6, wherein predicting based on the historical state of charge, the input parameter, and the target equivalent state equation to obtain the state estimate at the current time comprises: extracting an equivalent matrix from the target equivalent state equation; determining a historical state predicted value of the last moment based on the historical state of charge; substituting the product between the equivalent matrix and the historical state predicted value and the input parameter into a state equation function to obtain the state predicted value at the current moment.
8. The method of claim 6, wherein the state pre-estimate is calculated as follows: ; ; In the formula, For the state estimate at time k-1, Is an input parameter at time k-1, A state estimate at the current time k, For the a priori error covariance matrix at time k-1, For the a priori error covariance matrix at the current k-time, For the purpose of the equivalent transfer matrix of the object, Is the transpose of the target equivalent transfer matrix, The noise adjustment coefficient is used to adjust the noise, The temperature at the present moment k, As a reference to the temperature of the liquid, Is a process noise, wherein the process noise is calculated according to a covariance matrix of the process noise, , Is the covariance matrix of the process noise.
9. The method of claim 6, wherein the kalman filter gain is calculated as: ; In the formula, For the kalman filter gain, For the last time (k Time 1) a posterior error covariance matrix, For the equivalent observation matrix of the object, Is the transpose of the observation matrix in the target equivalent state equation, In order to measure the temperature sensitivity coefficient of noise, As the current temperature is set to be the current temperature, As a reference to the temperature of the liquid, Is the measurement noise at the reference temperature, wherein the measurement noise is calculated according to a measurement noise covariance matrix, , Is the covariance matrix of the measured noise.
10. The method of claim 6, wherein updating the state based on the kalman filter gain, the state estimate, and the actual measurement at the current time to obtain the state optimal estimate comprises: Extracting an observation matrix from the target equivalent state equation, and multiplying the state predicted value by the observation matrix to obtain a predicted measurement value; obtaining a measurement deviation between the actual measurement value and the predicted measurement value; multiplying the Kalman filtering gain by the measurement deviation to obtain a correction quantity; and calculating the optimal predicted value based on the correction amount and the state predicted value.

Description

Method for determining state of charge of battery Technical Field The invention relates to the field of battery state detection, in particular to a method for determining a battery state of charge. Background The State of Charge (SOC) estimation is used as a core function of the battery management system, and the estimation accuracy directly determines the energy utilization efficiency and the operation stability of the energy storage system. The accuracy of the SOC estimation is highly dependent on the accuracy of the battery model, but the construction of the battery model is very susceptible to temperature, and the Resistor-Capacitor (RC) parameters of the battery in different temperature environments may have significant differences, which may further cause the output characteristics of the battery model to deviate from the actual values, and reduce the accuracy of the SOC estimation. At present, two technical schemes are mainly adopted in the industry for battery modeling and SOC estimation, namely, a single RC model is built only under normal temperature or high temperature working conditions, the characteristic that the internal resistance of a battery is obviously increased under a low temperature environment is not considered in the scheme, the model precision is greatly reduced under the low temperature working conditions, and the accuracy of SOC estimation is difficult to ensure, and the RC model is built at different temperatures respectively, and the temperature is used as an input parameter to optimize a noise matrix in a Kalman filtering algorithm, so that the accuracy of SOC estimation is improved. However, the scheme has obvious defects that on one hand, noise has strong uncertainty, stability of an algorithm is difficult to ensure in a mode of optimizing a noise matrix through temperature, on the other hand, RC models at different temperatures are not effectively fused, battery characteristic parameters under a multi-temperature working condition cannot be fully utilized, and accordingly adaptability and estimation accuracy of the models still have a large improvement space. Disclosure of Invention In view of the above, the embodiment of the invention provides a method, a device, equipment and a storage medium for determining a battery state of charge, which aim to solve the technical problems of the existing battery modeling and state of charge estimation scheme, namely, firstly, a part of scheme does not fully consider the influence of extreme working conditions such as low temperature on the dynamic characteristics of a battery, and directly causes the estimation accuracy of the model to not meet the actual application requirements, and secondly, the other part of scheme improves the estimation effect by introducing temperature related parameters or optimizing a noise matrix, but still has the defects of poor model stability, excessive dependence on artificial experience in noise parameter adjustment, incapability of effectively fusing equivalent circuit models under the condition of multiple temperatures, and the like, and is difficult to consider the estimation accuracy and stability of the state of charge in the whole temperature range. Aiming at the problems, the application introduces a self-adaptive fusion mechanism and a stable state estimation strategy of the multi-temperature equivalent circuit model, and finally realizes high-precision and strong-robustness estimation of the battery state of charge under the full-temperature working condition. In a first aspect, an embodiment of the present invention provides a method for determining a state of charge of a battery, where the method includes: Collecting mixed power pulse data of the battery at different temperatures; Calculating battery characteristic parameters of the battery at different temperatures based on the mixed power pulse data at different temperatures to obtain state equations at different temperatures; The state equations at different temperatures are weighted and integrated through the temperature-related weights, so that an equivalent state equation is obtained; Obtaining a target weight, substituting the target weight into the equivalent state equation to obtain a target equivalent state equation, and performing Kalman filtering equation calculation based on the target equivalent state equation to obtain the state of charge of the battery, wherein the target weight is used for representing the proportion of the state equation corresponding to different temperatures in the equivalent state equation. Further, the calculating the battery characteristic parameters of the battery at different temperatures based on the mixed power pulse data at different temperatures to obtain state equations at different temperatures includes: calculating battery characteristic parameters of the battery at different temperatures based on the mixed power pulse data at different temperatures; Constructing an equation frame containing i