Lithium Battery State Estimation Based On The Multi-innovation Kalman Method

The battery management system is the key to ensuring the safe and stable operation of lithium batteries.The battery state estimation is the most important role of the battery management system,including battery SOC estimation,battery SOH estimation,battery SOP estimation,etc.However,the power battery has a strong time-varying non-linearity and is affected by random factors such as working conditions and environment in the application of electric vehicles,so it is challenging to accurately estimate its state.Thesis studies how to use the improved gradient algorithm and the improved Kalman filter method to estimate the battery parameters and state respectively,and obtain some results.The main research content includes the following parts:First,based on the characteristics of lithium battery,the second-order RC equivalent circuit model and capacity model of lithium battery are studied and established,and the state space equation with the battery SOC and polarization resistance voltage as variables is constructed.In addition,based on the difference equation of battery,forgetting factor stochastic gradient algorithm(FFSG)was used to identify the model parameters of lithium battery.Further,by extending the number of new information of SG algorithm,a multiinnovation forgetting factor stochastic gradient algorithm(MI-FFSG)is proposed to improve the accuracy of parameter identification.Second,in order to improve the accuracy of the extended Kalman filter(EKF)algorithm,a multi-innovation extended Kalman filter(MI-EKF)algorithm for lithium battery SOC estimation is proposed,which expands a single innovation at the current moment to include current and historical moment innovations,the number of innovations is increased to obtain a more accurate SOC estimation value.By transforming the battery state space equation,the battery capacity model is established,so that the battery SOH and SOC are combined,and the multi-innovation dual extended Kalman filter(MI-DEKF)joint estimation algorithm is proposed for battery capacity estimation.Third,for the problem of battery peak power estimation,the results of continuous peak power under single constraint and multiple constraints are discussed.In order to solve the problem that the peak power estimation result under single constraint is too conservative or it is difficult to ensure the safe use of the battery,it is adopted multi-constraint continuous peak power estimation method.Constraints include: battery charge and discharge limit constraints,battery SOC constraints,and battery model constraints.Fourth,use the NEWARE battery test platform to conduct discharge experiments on18650 lithium batteries under pulse cycle conditions,DST cycle conditions,and UDDS cycle conditions,and collect experimental data.The effectiveness of the model and the accuracy of the parameter identification results are verified in MATLAB simulation using pulse discharge experimental data;the effectiveness and accuracy of the joint estimation algorithm for estimating battery SOC and SOH is verified in MATLAB simulation using experimental data of UDDS and DST cycle conditions.It is verified by MATLAB simulation that the continuous peak charging and discharging power of the battery can be accurately estimated under the conditions of multiple constraints,and the method of multiconstrained continuous peak power estimation is practical.