Research On SOC Estimation Of ALS-UKF Algorithm Based On Adaptive Battery Model

With the rapid development of the automobile industry,the strong support of national policies and the high attention of many enterprises,the development of resource-saving and environment-friendly electric vehicles has become an important goal of the global automotive technology revolution.The key issues of power battery power density and cycle life are the bottlenecks restricting the industrialization of electric vehicles.In addition to the battery materials themselves,accurate battery models and state of charge(SOC)estimates are the basis of improving the above questions.The traditional Extended Kalman Filter(EKF)omits the high-order term of the Taylor expansion term and assumes that the measurement noise is Gaussian white noise with a mean of zero,introducing errors.In this paper,lithium iron phosphate battery is used as the research object.The error of EKF estimation SOC is reduced by using Unscented Kalman Filter(UKF)combined with Autocovariance Least Squares(ALS).In this paper,the working principle and basic characteristics of Lithium iron phosphate(LiFePO4)battery are studied.The factors affecting the accuracy of SOC estimation are analyzed from the definition of SOC.The accurate modeling of lithium iron phosphate battery is given below and lays the foundation for SOC estimates.The advantages and disadvantages of the commonly used battery equivalent circuit model are compared and analyzed.Combined with the basic characteristics of lithium iron phosphate battery,a suitable first-order equivalent circuit model is selected for modeling.In order to be able to adaptively adjust the model parameters,an artificial neural network(ANN)is combined to establish an adaptive ANN controller model to best adapt to the real battery terminal voltage.The experimental results show that the adaptive ANN control battery model has strong prediction ability for the model,and the model error is within 10 mV.Then,based on the established battery model,the UKF is combined with the ALS algorithm to estimate the battery SOC.The UKF uses its unscented transformation to avoid the EKF omitting the high-order terms of the Taylor expansion and introducing errors.The ALS estimates the measurement noise covariance by calculating the correlation in the measurement update,so the ALS-UKF algorithm improves the traditional EKF estimating SOC.The experiment compares the SOC estimation results of the four estimation methods of EKF,EKF + ALS,UKF and UKF + ALS from two aspects of constant current discharge test and DST condition test.The UKF + ALS algorithm estimates the average absolute error of SOC is lower than that of EKF 0.02.Finally,the SOC estimation platform of lithium-ion battery pack was built,and the main hardware selection and design process,as well as the main software design flow,were carried out.The information measurement test and SOC estimation test were carried out for the established SOC estimation platform.The results show that the proposed SOC estimation method effectively improves the estimation accuracy and has certain practical application value.