Research On Fractional Modeling And SOC Estimation Algorithms Of Lithium-ion Battery For Vehicles

As the energy source of electric vehicle,the performance of power battery directly affects the power,economy and safety of electric vehicle.Limited by material technology,the energy density,power density and service life of ternary batteries and lithium-iron phosphate batteries widely used in electric vehicles are not ideal,and they still cannot meet the application requirements of electric vehicles.Therefore,the battery management system is very important to ensure efficient,safe and reliable operation of electric vehicles,and it is also key research direction of power battery at present.The estimation of battery state of charge(SOC)is the most basic and important function of battery management system.The estimation accuracy of battery SOC directly affects the economy,safety and life of electric vehicles.However,time-varying non-linearity of power battery and variability of application conditions bring great challenges to the accurate estimation of battery SOC.Aiming at the estimation of lithium-ion power battery SOC,the ternary battery is taken as the research object in this paper.Based on modeling method of complex system and modern filtering technology,this paper improves estimation accuracy of battery SOC from two ways: firstly an estimation method is proposed based on unscented transformation strong tracking filter algorithm,secondly another estimation method is studied based on fractional order model.The main research work of this paper is as follows:The mechanism of electrochemical reaction of lithium ion power battery and the experimental platform for battery testing are introduced.According to the charging and discharging experiments of lithium-ion battery,the external characteristics and its influencing factors(including temperature,internal resistance,capacity,etc.)are analyzed.Six equivalent circuit models are established according to the number of RC circuits of equivalent circuit model and whether hysteretic voltage is considered.These models are compared and analyzed under various operating conditions.On this basis,a fractional second-order RC equivalent circuit model is established based on the principle of fractional calculus,and the fractional second-order state space equation is described.The parameters of the model are identified by the designed pulse power test experiment,and the fractional order of the fractional order model is identified by genetic algorithm.The accuracy of fractional order RC model is verified by experiments under various conditions.The results show that fractional order RC model has higher accuracy than integer order RC model.Based on integer second order RC model,the calculation process and shortcomings of traditional Kalman filtering algorithm and extended Kalman filtering algorithm are analyzed.Aiming at the problem of insufficient precision of Jacobian matrix and Taylor expansion in the extended Kalman filtering,the unscented Kalman filtering algorithm is introduced to estimate the SOC of power battery.Aiming at the problem of insufficient precision of the integer order model,the unscented transformation is proposed.The tracking filter estimates battery SOC,introduces fading factor,adjusts error covariance in real time,and reduces the impact of model misalignment.Finally,the accuracy and robustness of the algorithm are verified under constant current discharge condition,compound pulse power test and operation condition in Federal Cities of the United States.The results show that the trackless transform strong tracking filter has high estimation accuracy under various conditions.Based on fractional second-order RC model,a fractional-order central differential Kalman filter algorithm is designed to estimate battery SOC.This algorithm does not need to calculate Jacobian matrix.The nonlinear system is linearized by Stirling's firstorder interpolation formula.The linearization process is more accurate than Taylor's expansion.The accuracy and robustness of the fractional Kalman filter algorithm are verified under various operating conditions.The results show that the algorithm has high accuracy under various operating conditions.On this basis,an adaptive fractional central differential Kalman filter(AFCDKF)algorithm for estimating SOC of batteries is derived to solve the problem that noise characteristics are difficult to obtain in practice and are constantly changing.The adaptive algorithm can estimate both the noise state and the system state vector simultaneously.The accuracy and robustness of SOC estimation are verified by experiments under various conditions.The results show that the algorithm can track the noise characteristics well and reduce the impact of noise interference.In order to improve the accuracy of battery model and SOC estimation,the common equivalent circuit models are compared and analyzed.A fractional-order model modeling method is proposed.The algorithm of traceless transform strong tracking filter and adaptive fractional-order central difference Kalman filter are proposed to estimate battery SOC.It provides a better modeling and SOC estimation algorithm for battery management system.On the one hand,it can be used for electric vehicles.On the other hand,it can prevent overcharge and overdischarge of batteries and maintain safe and efficient operation of batteries.