Research On Estimation Method Of State Of Energy And Peak Power For Lithium Titanate Batteries

In recent years,the rapid development of my country’s electric vehicles,rail transit and other industries has put forward higher requirements for energy storage technology.Lithium titanate batteries are widely used in energy storage systems because of their long cycle life and high safety.In practical applications,the Battery Management System(BMS)is required to estimate the state of the battery.The state estimation is related to the efficient and safe operation of the energy storage system.Therefore,it is very important to study the battery state estimation algorithm.However,existing battery state estimation algorithms still have room for improvement in terms of accuracy and robustness.Therefore,this article focuses on the state of charge(SOC),State of Energy(SOE)and peak power three battery states to carry out state estimation algorithm research,the main research content is as follows:First,for the problem of how to accurately and quickly establish an equivalent model of a lithium titanate battery,after analyzing the advantages and disadvantages of the existing battery model,the first-order RC-Thevenin model with higher accuracy and low complexity is selected as the equivalent model.The recursive least squares(RLS)algorithm is used to identify the model parameters of lithium titanate battery.In order to solve the problems of the traditional RLS algorithm,such as the complexity of calculation and the large amount of calculation,a step change recursive least squares(SC-RLS)algorithm based on piecewise optimization strategy is proposed.According to the characteristics of the internal parameters of the battery,the appropriate calculation step is selected to give consideration to the calculation accuracy and timeconsuming.Experiments show that this scheme can accurately establish a battery model,and the amount of calculation is small.Then,in order to solve the problems of poor stability and accuracy in estimating battery SOC and SOE by traditional Kalman filter,a kind of singular value decomposition-adaptive volume Kalman filter(SVD-ACKF)algorithm is proposed.The algorithm uses singular value decomposition instead of Cholesky decomposition in the traditional CKF to improve the stability of the algorithm;the Sage-Husa method is used to update the noise covariance matrix in real time to improve the accuracy of the algorithm.Experimental results show that SVD-ACKF can track battery SOC and SOE accurately and stably.Finally,for the problem of how to accurately estimate the peak power of the battery under multi-factor conditions,based on the data-driven idea,the input feature quantity is first analyzed according to the statistical principle,and the battery SOC,temperature and cycle number are selected as the input feature quantity.The Gaussian process regression algorithm is used to estimate the battery peak power,and the gray wolf optimization algorithm(GWO)is used to optimize its hyperparameters.Select Support Vector Regression(SVR)as the comparison algorithm.Experiments show that the Gaussian process regression algorithm has higher accuracy than SVR.