Research On Robustness State Of Charge Estimation Of Lithium-ion Power Battery For Electric Vehicles

Nowadays,as for electric vechiles,battery technology is still the primary bottleneck that hinders its developmet.Multiple fuctions of the battery management system(BMS)are accomplished depending on the accurate state estimation of batteries.However,due to the complex physical and chemical reactions occurring inside the batteries during the working operations,the strong nonlinearity and time-variant property make it impossible to obtain the precise state of batteries via simple measurements.Moreover,many external factors of uncertainty will put great impact on the performance of the batteries,such as time-varying amibient temperature or the demanding power,etc.This raises higher requirements for the state estimation method for batteries,of which not only the accuracy should be guaranteed,but also the robustness should be regarded.Some work is studied with respect to the Lithium-ion battery,which has been listed as follows:1)Establishing the test bench for Lithium-ion batteries.Many experiments are carried out under different temperatures for distinct structures of batteries.The collected data are used for verify the effectiveness of algorithms.Based on the test data,the variations of open circuit voltage(OCV)and electrochemical impedance spectroscopy(EIS)influenced by temperature or ageing process are analyzed.The results demonstate that the shift of OCV is vulnearable because of ageing process and it is obvious that the change of ohimic resistance since the different temperature and level of ageing.2)According to the analysis that the variations of equivalent circuit models cause the difference of SOC estimation results,the multi-model probabilities state of charge fusion estimation method is proposed.This approach improves the redundancy of estimation methodology and declines the failure probability of the BMS once the SOC estimation based on single model is not reliable.The simulation results show that adopting the distributed fusion estimation framework with feedback mechanism is beneficial to elevate the accuracy of SOC estimation.After fusion process,the comprehensive SOC estimation values are optimized and the credibility is promoted.This algorithm are also verified under different temperatures and shows its robustness.3)Since the relationship of OCV-SOC is sensitive to the capacity degradation,an offline available capacity estimation method based on the optimized response surface is presented.According to the test data,the alteration of the relationship of OCV-SOC is studied and the three dimentional response surface model about OCV,SOC,and Capacity is constructed.Combining the ampere-hour integration method and parameters identification technique,the available capacity can be treated as one of model parameters and identified by some intelligent optimization method.The results indicate even if under various cycles or level of ageing,the estimation error of the available capacity can be kept within 5%.4)Considering the multiple states joint estimation problem of Lithium-ion battery,the dual H infinity filters based SOC and available capacity joint estimation approach is investigated.Due to the time-varying feature,the dual filters based framework is employed to estimate the model parameters and SOC concurrently.Besides,since the strong dependency on the accurate model for common equivalent circuit models based methods,and meanwhile the model error(unconstructed part)always exists,it offers strict request on the robustness of state estimation methods.Moreover,the traditional Kalman filter approach confines the statistical feature of noise items,which incurs some conservation.In order to overcome the shortages mentioned above,the H inifinity filter technique is adopted,and the three dimentional response surface model on OCV is introduced as well to speed up the convergence of capacity.Eventually,the simulation results show the validation of this method.5)For the application and validation issue of multiple states joint estimation,the x PC target based hardware in the loop simulation bench is set up.The experimental results indicate this method is effective to estimate the SOC and available capacity of the Lithium-ion battery with satisfactory accuracy and simultaneously it shows the good robustness to the initial errors of states.To achieve more accurate,reliable SOC estimation for BMS,the robust state estimation methods for lithium-ion battery applied in electric vehicles are studied.Aiming at realizing this goal,the multi-model probabilities based fusion estimation approach is presented to deal with the unreliable SOC estimation based on single model.To overcome the estimated error of SOC resulting from uncertain capacity during the normal ageing process,an offline capacity estimation method based on the response surface model of OCV is studied.The dual H infinity filters based joint estimation approach is proposed to tackle the problem that SOC and available capacity of battery are estimated collaboratively and improve the robustness of estimation method to the noise and model error,which owns applicable value of importance in engineering.