Research On Model Parameters And SOC Estimation Method Of Lithium-ion Battery Based On Cubature H Infinite Filter

Under the double pressure of environmental pollution and energy crisis,electric vehicles have developed rapidly and steadily increased in number.Among them,battery as its main power source has a profound impact on the safety of the vehicle.Battery Management System(BMS)monitors,manages,and maintains batteries to ensure battery safety.The estimation of State of Charge(SOC)of batteries is one of the important components of BMS.Accurate estimation of the battery SOC can prevent the battery from over charging-discharging,thus extending the battery life and making it easier for users to reasonably plan the path and predict the range of the battery.In order to obtain accurate simulation results of battery SOC estimation,the following work is done in this thesis.Firstly,in order to achieve the accurate estimation of battery SOC,a Thevenin battery equivalent circuit model is established and a battery test platform is built.The fitting function between Open Circuit Voltage(OCV)and SOC is obtained by the fitting test.Considering that battery parameters will have non-negligible changes with battery aging,thus affecting SOC estimation results,it is necessary to confirm battery model parameters in the current aging state in time.In order to obtain the changes of model parameters in the whole aging process of the battery,this thesis designed and completed the battery aging test,and extracted the changes of the battery capacity,resistance,polarization resistance and polarization capacitance along with the battery charging and discharging times.Voltage,current and other parameters,which are easy to measure,are extracted and used as Aging Characteristics(AC)to estimate the parameters of the battery model.In addition,parameters are extracted from battery test data from the Center for Advanced Life Cycle Engineering(CALCE)at the University of Maryland to provide supplementary data for subsequent method validation.Secondly,Cuckoo Search-Back Propagation(CS-BP)algorithm is used to estimate battery model parameters.BP algorithm has good applicability to nonlinear problems such as estimating battery model parameters,but in the process of application,improper selection of initial weights and thresholds may cause gradient explosion or gradient disappearance,resulting in large error of estimation results.In order to solve this problem,this thesis adopts Cuckoo Search(CS)algorithm to optimize the initial weights and thresholds,and adds momentum items related to the previous update into the weight updating process to improve the stability of the neural network.Meanwhile,under the same number of iterations,it also improves the accuracy of the algorithm.The algorithm is verified by using the extracted data of laboratory lithium manganate battery and lithium cobalt CALCE battery.The simulation results demonstrate the universality and effectiveness of the proposed method.By using the estimated battery model parameters,the battery equivalent circuit model is further improved,and an accurate battery model is established for the subsequent SOC estimation.Finally,based on the battery equivalent circuit model established above and the estimated model parameters,The Flower Pollination Algorithm-Cubature H Infinity Filter(FPA-CHIF)is used to estimate SOC.Considering that the battery model is a nonlinear model with complex internal reactions,the volume rule and the HIF algorithm are integrated in this thesis.On the basis of preserving the accuracy and robustness of the HIF algorithm,the adaptability of the HIF algorithm to the battery model and other nonlinear system is improved,and the influence of non-Gaussian noise is suppressed to the greatest extent.In addition,FPA is used to optimize the estimated battery model parameters,and the model parameters are updated in real time during the estimation process,so as to further improve the accuracy of the model and SOC estimation.Considering conversion probability as an important parameter of FPA,the ratio of global optimization and local optimization of the algorithm is adjusted.In this thesis,adaptive conversion probability is used to replace the original fixed conversion probability,which improves the optimization effect of the algorithm.Finally,due to the limited battery data provided by CALCE,only the laboratory battery data are used in this thesis to design the short-term SOC estimation simulation under different aging degrees of batteries,and the long-term SOC estimation simulation with alternating aging degrees of batteries.The simulation results show that the improved algorithm can achieve accurate SOC estimation in both cases.Moreover,the accuracy of SOC estimation is significantly improved compared with the unoptimized HIF algorithm and CHIF algorithm without considering the change of model parameters.