Lithium-Ion Battery SOC Estimation Based On STFDEKF

In recent years,with the increase of energy crisis and environmental pollution,all countries in the world are continuously increasing the research and development of new energy vehicles.The development of battery systems and battery management technologies are the top priorities for the development of electric vehicles,both domestically and abroad.The battery management system introduced by automobile enterprises or parts suppliers can basically achieve the most basic functions,but how to improve the accuracy of battery SOC estimation under various complicated driving conditions is still in urgent need of solution.From the perspective of this problem,this paper selects a power-type ternary Li-ion battery and conducts a series of tests,simulations,and simulations around how to more accurately estimate the SOC of the battery based on the strong tracking finite difference extended Kalman filter algorithm which has launched a wide range of research and discussion.Firstly,various types of power batteries and their applications are introduced.Lithium-ion batteries are,particularly suitable for a large number of applications in electric vehicles because of their relatively low manufacturing cost,high specific energy and cycle life,and lack of memory.Comparing with the existing SOC estimation methods,Kalman filter and its extended optimization algorithm are notable because of their insensitivity to initial values and online estimation.Then from the structural principle of lithium-ion batteries,three battery models are proposed.The equivalent circuit model uses a simple electronic component to build a circuit frame and describes the charge and discharge characteristics of the battery based on the Kirchhoff voltage and current law,which is convenient for the estimation of model parameters.With comprehensive accuracy and ease of calculation,this paper uses a second-order RC network to build the equivalent circuit model of the battery.Next,we need to carry out various tests for the selected power-type ternary lithium-ion batteries,including Calculating the battery capacity at different temperatures and charge and discharge rates,and obtain HPPC operating condition response curves for the battery,Analyzing the relationship between polynomial order and voltage curve fitting accuracy and use a fifth-order polynomial to fit the battery's OCV-SOC curve,obtaining the relationship between the life of the battery and the cycle number by charging and discharging the battery using 1C current rate at room temperature.Next,this article describes the different types and important concepts of artificialimmune algorithms.Artificial immune algorithm does not need to seek the derivative of the function.It does not require continuous adaptability and adaptability.It can automatically optimize in the global scope.It is widely used in multi-objective parameter optimization and parameter identification.This paper proposes a battery model parameter identification method using artificial immune algorithm and realizes the model parameter identification under HPPC operating conditions.The voltage error of the model is less than 0.035 V under the cyclic current mode.After the model is established and the parameter identification is complete,the selected method can be used to estimate the SOC of the battery.This paper analyzes the limitations of the extended Kalman filter algorithm because there are two main reasons: using the Taylor formula expansion that ignores high-order terms to approximate the linearization and the need to solve the Jacobian matrix,then points out that the strong tracking finite-difference extended Kalman filter uses a finite difference method instead of a first-order derivative,which is more simple and adaptable.At the same time,a strong tracking suboptimal fading factor is introduced to eliminate the accumulation of stale data and prevent saturation.Phenomenon,this method enhances the tracking performance of the algorithm and can improve the accuracy of the SOC estimation of the lithium-ion battery.In this paper,a battery system simulation model including the STFDEKF(Strong Tracking Finite Difference Extended Kalman Filter,STFDEKF)algorithm and the second-order RC equivalent circuit model is built by Matlab/Simulink software.Estimating the SOC of the battery using periodic constant current conditions,the error is within 4.1%.Finally,the battery was tested in two cycle conditions: NEDC and UDDS.The experimental data were collected to estimate the SOC of the battery.The errors were less than 2.1% and 4.3%,are better than the extended Kalman algorithm.