A Study On Lithium-ion Batteries SOC Initial Value Tracking Based On Suboptimal Solutions Of Bayesian Filter

The automobile exhaust has become the most serious cause of the air pollution in the present world.Because the number of cars is still rapidly increasing,the air pollution which caused by automobile exhaust is becoming more serious.At the same time,the traffic oil consumption in the total oil consumption is growing more and more only to put heavy pressure on oil resources.Countries around the world reached a consensus for this problem: the development of new energy vehicles is imperative.A core issue in the development of new energy vehicles is the research of battery management system.After years of development,the battery management system is quite mature,but there are still some problems which have not been solved yet.How to get the the accurate estimation of the state of charge is the first problem to be solved.In the theoretical study on the estimation of the state of charge by using loop algorithm,the known initial value is usually set beforehand,which does not conform to the practical application.So this method can't fully demonstrate the performance indexes of the algorithm.This topic's study of the SOC estimation problem aims at the problem about uncertain initial assignment and studies its convergence to the true value through Bayesian filtering and the supplemented experiment of MATLAB simulation.The research into this topic is of important significance to the test of the robustness of the closed-loop algorithm.This paper firstly introduces the applying background,analyzes the basic functions of battery management system,focuses on the research into the estimation of SOC under the condition of the uncertainty of the initial assignment,and introduces the structure and research significance of this thesis.Then several power batteries are introduced.Combining the present power battery selecting criteria,I select the power battery used in this topic-lithium iron phosphate battery.I intensively study the open circuit voltage characteristic of this battery,and define the voltage plateau.There are different ways to build the power battery model.According to the external characteristic of lithium iron phosphate battery,the SOC dynamic observation model is established in this thesis and the least square method to obtain the observed model parameters is determined.The article goes on to introduce the algorithm of the SOC estimation-no model SOC algorithm and SOC algorithm based on the model.This topic focuses on the three second-best solutions of Bayesianalgorithm-extended Kalman filter,limitless Kalman filter and particle filter.I use three algorithms respectively for the researching different initial value assignments and different real SOC initial value,and obtain the results of using the MATLAB simulation.According to the simulation results,I find that when we want to know the initial value of the voltage plateau's models,it's best to use limitless Kalman filter and the particle filter is best to be used in tracking the initial value in other fields.So my thesis puts forward a method of tracking the initial value combining limitless Kalman filter and the particle filter and gives the flow chart of this method.