The Study On Lithium-ion Battery Model And SOC Estimation

It is a hot spot that how to quickly and accurately estimate the current performance of battery for BMS(Battery Management System)and how to reliably predict the future state for BMS.In the paper,we have carried out an in-depth study on how to quickly and accurately estimate SoC for Li FePO4 battery,and have mainly done the following aspects:First,This paper analyzes the advantages and disadvantages of the non-circuit model and equivalent circuit model.A new power battery model is introduced with high degree of approximation,simple structure and convenient to identify parameters,combined with the external characteristics of lithium iron phosphate battery.According to the identified parameters of the model,the simulation results show that the model has high precision and stability by simulink.Second,BMS is affected by various kinds of noise in the actual operating mode.In order to effectively filter the interference of colored observation noise and colored state noise to battery state estimation,this chapter applies the covariance matrix adaptive method to add the framework of Kalman filtering algorithm to achieve the estimation of noise covariance matrix.Based on the Unscented Kalman filter with the colored observation noise,the adaptive unscented Kalman filter with the colored observation noise is proposed.Based on the Kalman filter with the colored noise,the adaptive Extended Kalman filter with the colored noise is proposed.The simulation results show that this algorithm improves accuracy and stability of the SoC estimation.At the same time,It is not sensitive to initial values of the noise covariance matrix,has the adaptive ability for the unknown statistics of the system noise,and overcomes the problem of noise covariance matrix initial value setting lack of standards.Thirdly,lithium-ion battery is a typical dynamic,non-linear electrochemical system,and the common battery model can't accurately describe the characteristics of the dynamic changes,nonlinear and strong coupling.The paper builds lithium-ion battery model based on the online support vector regression machine.We select operating voltage and temperature as the input variables,SoC as the output variable.The simulation results show that the algorithm can online estimate the SoC in real time,and has high accuracy and algorithm stability in SoC estimation.Fourth,based on lithium-ion battery,which is a complex electrochemical system with characteristic of multiple-time scale,The paper proposes a multi-time scale battery model based on fusion model.At the macro time scale,the battery performance is slowly decline,so online support vector regression estimate curve for EMF-SoC;At the micro time scale,the battery status quickly change,and It is easy to be interfered by various kinds of noise.The paper puts forward the adaptive Kalman filtering algorithm with colored noise to estimate SoC.The simulation results show that the multi time scale battery model based on the fusion model can accurately estimate SoC with the slow change of battery parameters.