State of charge (SOC) estimation method of variable length sliding window by identifying battery parameters

Abstract

The invention relates to an SOC estimation method of a variable length sliding window by identifying battery parameters. The method comprises the steps that the battery parameters are identified in real time by a variable length sliding window least square method, then the battery parameters are used for conducting self-adaptive extended Kalman filter to estimate the battery SOC, and therefore the influence of the battery parameter change caused by battery ageing on estimation precision of the battery SOC is avoided. According to the method, the time-varying battery parameters can be tracked effectively, and no dependency exists to the selection of initial data; the self-adaptive extended Kalman filter is more suitable for the state estimation of a nonlinear time-varying system; a matrix inversion operation is converted into scalar division operation, and better real-time performance is achieved.

Description

technical field [0001] The invention relates to the technical field of electric vehicle power battery management systems, in particular to a method for estimating the state of charge of a battery using a variable-length sliding window to identify battery parameters. Background technique [0002] The power battery is a key component that affects the power, safety and economy of electric vehicles. Its efficient and reliable control is the key guarantee for the battery management system to realize the safe driving of the vehicle. The accurate estimation of the battery state of charge (SOC) is one of the key technologies of the battery management system. [0003] At present, SOC estimation methods include classic ampere-hour integral method, open circuit voltage method, Kalman filter method, neural network method and so on. The ampere-hour integration method will produce cumulative errors over time, and the errors will become larger and larger. Usually, it is necessary to use ...

AI Technical Summary

Problems solved by technology

① Offline parameter identification is to use the least squares method to identify all battery experimental data to obtain relevant battery parameters. In the process of estimating SOC using the Kalman filter algorithm, these parameters are calculated as constants. The disadvantage of this algorithm is As the battery life decays, these battery parameters will change. As the battery ages, if the initial battery parameters are still used to estimate the SOC, it will cause a relatively large estimation error.

② Online parameter identification uses the iterative least squares algorithm to identify battery parameters in real time. During the process of estimating SOC with the Kalman filter algorithm, these parameters are updated in real time. Although this method can track the changes of battery parameters in real time, the iterative least squares With the growth of data, the multiplication algorithm will experience data saturation, so that the recursive least squares algorithm will gradually lose its correction ability

The Kalman filter algorithm also has the following deficiencies in practical applications: ①The dependence on the initial parameters is relatively large. If the initial value that is closer to the actual value cannot be accurately given, a relatively large estimation error will be generated; ②The calculation of the filter The volume increases sharply with the cube of the state dimension, which cannot meet the real-time requirements in the process of vehicle operation

[0004] To sum up, the power battery is a nonlinear and time-varying system. The offline parameter identification algorithm always uses a fixed constant to estimate the state of charge of the battery. As the battery ages, the estimation error will become larger and larger, which cannot satisfy the overall The use requirements of the car; although the recursive least squares algorithm used in online parameter identification has the advantage of real-time tracking of parameter changes, data saturation will occur as the data grows, causing the algorithm to gradually lose its correction ability; The initial parameter dependence is relatively large. When the initial value close to the actual value cannot be accurately given, a relatively large estimation error will often be generated; in addition, the calculation amount of the Kalman filter algorithm is increased by the cube of the state dimension. Meet the real-time requirements during vehicle operation

Images