Research On State Estimation Algorithm And Its Application For Batteries In Electric Vehicles

The state estimation of Lithium-ion battery represented by SOC and SOH is the core function of BMS(Battery Management System).The accurate battery state is not only significant for efficient utilization of battery,but also estimating vehicle driving mileage,formulating energy management strategies and improving driving safety.In addition to the time-varying nonlinear characteristics of the battery,the accuracy of estimation algorithm is also affected by sampling noise,inaccurate model parameters and state coupling in practical applications.Focusing on solving problems above,some work about the joint estimation algorithm of SOC and SOH for Lithium-ion battery has been summarized as follows:(1)To achieve the balance between model complexity and computational efficiency,this thesis compares Thevenin model with DP(Dual Polarization)model which are combined with the feature fitting method to accomplish the offline parameter identification.The results show that the DP model has better prediction accuracy.Based on the offline parameters,the external influence factors of the model parameters and the effect of each impedance parameter on the model accuracy are studied.The results show that the ambient temperature has great significant influence on the model parameters.The impact of each parameter on the model accuracy is ranked from high to low as follows: ohmic resistance,long time scale RC network and short time scale RC network.(2)In terms of online parameters and SOC estimation,a multi-timescale joint estimation algorithm based on EKF-HIF(Extended Kalman Filter-H Infinity Filter)is proposed for the multi-timescale effect of DP model.The EKF algorithm is used to identify model parameters of different frequency characteristics with different timescales,which avoids the reduction of model accuracy caused by identifing all parameters simultaneously.As for the SOC estimation,the HIF algorithm is used instead of the traditional EKF algorithm to improve robustness of SOC estimation.Based on the experimental data of UDDS(Urban Dynamometer Driving Schedule),the proposed joint estimation algorithm is evaluated by three indicators: accuracy of model terminal voltage prediction,SOC estimation accuracy and algorithm convergence speed with initial error.The results show that the joint estimation algorithm has good estimation accuracy and robustness.(3)In terms of SOH estimation,capacity is selected as the indicator of SOH which is estimated based on RTLS(Recursive Total Least Square)algorithm.The accuracy and convergence of the RTLS using different timescales are studied.This thesis constructs the multi-timescale joint estimation algorithm of SOC and SOH by EKF-HIF and RTLS,and servel indicators are used to evaluate the performance of the algorithm based on the experimental data of aged battery under the UDDS test profile,such as the estimation accuracy of SOC and SOH,along with SOH convergence speed.The results show that the joint estimation algorithm can effectively improve the state estimation accuracy of aged battery.(4)In the application of the joint estimation algorithm,the real-time simulator based on PXIe-8135 processor board is used to construct the BMS hardware-in-the-loop simulation test platform,as well as BMS hardware based on MPC5744 P and MC33771 B chips.The joint estimation algorithm proposed is tested in the BMS with sampling noise,the results show that the algorithm can accurately estimate the battery state even the input contains sampling noise.