Study On The State Of Charge Estimation Of Lithium-ion Battery Based On The Adaptive Robust Unscented Kalman Filter Algorithm

The state of charge(SOC) of lithium-ion batteries, which are used on electric vehicle,is one of the important parameters for electric vehicle,which can be estimated for theremaining mileage and provided the necessary data to energy management,and alsoeffectively prevents the lithium battery from overcharge and overdischarge,so its accurateestimation has become one of the key technology of the electric vehicle. Aiming toimprove the accurary of the SOC estimation,this paper do some research about the SOCestimation algorithm,the main work of this paper are as fllows:(1) The commonly used Thevenin battery model and PNGV model are introducedin this paper.And the Thevenin battery model and PNGV model are established for thesingle LiFePO4battery and the battery pack in room temperature. The precision ofbattery is checked by using the battry terminal voltage. The simulation results show thatThevenin model and PNGV model have higher modeling precision.(2) The traditional Kalman Filter algorithm which includes the Extended KalmanFilter (EKF) algorithm and the Unscented Kalman Filter (UKF) algorithm is introducedin this paper. The EKF algorithm and UKF algorithm are respectively designed for singlebattery and battery pack based on the established battery model.Under the constantdischarging current,the simulation results shows that the UKF algorithm has higheraccuracy than EKF on the lithium-ion battery estimation and the reason is analized. In thereaserch,the affection factors that inclued the process noise covariance initial value,theobservation noise variance and the SOC initial value are analyzed and the simulationresults show that they have significant influnence in the stability and precision of theSOC estimation.(3) The priori estimation data,which are uesd in the UKF algorithm,that its samevalue during the estimation process will bring estimation error is pointed out in thispaper. The Adaptive Unscented Kalman Filter(AUKF)can be obtained by introducing theadaptive control method to the UKF algorithm.The posteriori estimation of noise can beachieved by the noise parameter adaptive adjustment.The simulation results show that theAUKF algorithm has higher pricision than UKF algorithm. The filter pricison decreasesand even lead to divergence under the colored noise through further study. Aiming at theproblem of AUKF,the Adaptive Robust Unscented Kalman Filter are firstly used in SOCestimation followed by its successful use in GPS navigation and positioning. Thesimulation results show that the ARUKF algorithm has higher pricision than AUKF algorithm.(4) In order to verify the SOC estimation validity and accuracy of ARUKFalgorithm under the harsh working condition,the complete vehicle performancesimulation is conducted in the ADVISOR environment under the Urban DynamometerDriving Schedule working situation.The SOC can be obtained by ARUKF algorithmaccording to the simulationg data. The simulation results show that the ARUKFalgorithm still matain the higher pricision under the harsh working condition.