point Kalman filter for mineused battery

 

HE Lingna, WANG Yunhong

(College of Information Engineering, Zhejiang University of Technology, Hangzhou 310023, China)

 

Abstract: In order to estimate the state of charge（SOC） for mineused battery exactly and realtime, the weighted statistical linear regression method was used to achieve the linearization of model function, and the technique of sampling points Kalman filter was applied to the state of charge estimation for mineused battery. Aiming at the limited resource of the battery management system, the nonlinear filtering algorithm that combined standard Kalman filter and sampling point Kalman filter was proposed based on the linear characteristic of the state equation and the nonlinear characteristic of the observation equation of the battery model. In order to achieve the strong tracing ability of forced condition and the robustness of the model inaccuracy, the singular value decomposition was introduced, and the error covariance matrix was replaced by the characteristic covariance matrix, and suboptimal fading factor was introduced based on the principle of strong tracing. The simulation results indicate that the state of charge estimation algorithm based on improved sampling point Kalman filter for mineused battery takes the filtering accuracy and the amount of computation into account, and has the strong tracing ability to deal with the forced condition and the robustness to deal with the inaccuracy of model, and is totally applicable to the state of charge estimation for mineused battery whose resource is limited. Therefore, the new algorithm has good practical value.

Key words: mineused battery; state of charge(SOC); sampling point Kalman filter; singular value decomposition; strong tracing filter