Research On SOH Estimation Of Lithium Battery Based On IMM Fusion Model

In order to solve the energy crisis and environmental pollution,the development and utilization of new energy has become a new way to promote economic development.As one of the new energy sources,lithium battery has been vigorously developed because of its environmental protection and energy saving characteristics.Lithium batteries are not only widely used in electronic products accessible to the touch,but also widely used in automotive and power industry as power energy.Electric vehicles are the most representative applications.For electric vehicles,accurately predicting the remaining service life of batteries is one of the core issues for electric vehicles to gain competitive advantage with fuel vehicles.Accurate SOH estimation can provide accurate battery health status,provide security for vehicle driving safety,and has important significance for the use and timely replacement of battery packs.In this dissertation,we study a new algorithm of interactive multi-model to interact with multiple single empirical models,and use the unscented particle filter to improve the accuracy of estimation results.It has the advantages of simple model operation,high accuracy of estimation results and better stability of the model.The contents are as follows:(1)The working principle of lithium batteries and the causes of capacity decay are introduced from the electrochemical point of view,and the effects on SOH of lithium batteries are analyzed in detail,including charge-discharge ratio,charge-discharge depth,ambient temperature,inconsistency and cycle times.In view of these factors,researchers have developed a series of capacity decay models for lithium batteries,including equivalent circuit modes.The advantages and disadvantages of these models are analyzed from the aspects of complexity,accuracy and applicability of model parameter determination.Finally,the empirical model is taken as the basic model of this dissertation.(2)In order to overcome the shortcomings of empirical models,which are mostly applied to lithium batteries under specific operating conditions,and the estimation error of the models is large when the battery type changes,this dissertation uses the interactive multi-model algorithm to integrate the single empirical models,so as to improve the applicability and stability of the models.Because Kalman filter is used in the traditional interactive multi-model,there is a great error in dealing with the non-linear dynamic system such as lithium batteries,so this dissertation combines the trackless particle filter and the interactive multi-model algorithm to produce a new algorithm model,i.e.the interactive multi-model trackless particle filter algorithm.(3)The initial parameters of each single factor model are determined by MATLAB curve fitting,and the algorithm simulation program is compiled.The trackless particle filter algorithm is used for each single factor model and the interactive multi-model trackless particle filter algorithm is used for the three models respectively.The simulation results show that the estimation results of the interactive multi-model trackless particle filter algorithm are more accurate than those of the single empirical model.It has good performance in degree and stability.