Research On Evaluation Method Of State Of Charge For Lithium Ion Battery In Electric Vehicles

In order to alleviate the increasingly harsh global environment,in recent years,there has been an upsurge in electric vehicles in the automotive industry,and electric vehicles have become the protagonist of the current automobile industry due to their remarkable advantages of energy saving and environmental protection.It is also one of the important ways to alleviate the current energy shortage and environmental pollution problems.With the continuous development of electric vehicles,lithium-ion batteries,as the main energy supply system and energy carrier of electric vehicles,are also the core part of electric vehicles,and it has gradually become the focus of research by many scholars.Battery Management System(BMS)is an extremely important part of battery technology research,and battery residual capacity(SOC)estimation is the core research content of battery management system(BMS).In this paper,the lithium-ion battery for vehicles is used as the research object.The BP neural network estimation method,the genetic algorithm optimization BP neural network estimation method and the cross-optimal BP neural network estimation method are used to calculate the remaining capacity(SOC)of the vehicle lithium ion battery.Detailed estimation study.The main research work done in this paper is as follows:(1)Firstly,the research status of electric vehicles and power batteries is described.The definition of traditional classic battery SOC and the definition of electric vehicle industry in practical application are briefly described.The existing methods of estimating SOC of various batteries are summarized.And they illustrate their advantages and disadvantages from different angles,confirm the research object and the battery SOC estimation method used,which lays a solid foundation for the later research work.Then the structure,basic working principle and common equivalent circuit model of lithium ion battery are briefly described.Secondly,the characteristics of the lithium-ion battery are elaborated and analyzed through experiments to summarize the main factors affecting the SOC of the battery.(2)The principle and design of BP neural network estimation battery SOC model are analyzed and studied in detail.Through the experiment,the large amount of data needed for the research in this paper is obtained and the data is effectively screened,and then the data is normalized to obtain the sample data,and the training performance of the estimated network can also be improved.The general structure of the model is analyzed in detail to determine the general structure of the model.Finally,the Matlab simulation software is used tosimulate and construct the network model.However,the traditional BP neural network has been difficult to determine the initial weight and threshold,and the initial weight.The threshold value has a great influence on the prediction error of the network and the convergence speed,which greatly affects the battery SOC estimation result.In order to solve this problem,the genetic algorithm is used to optimize the weight and threshold of BP neural network,then the genetic algorithm is used to optimize BP neural network to estimate the battery SOC,and the estimation result is compared with the traditional BP neural network to estimate the battery SOC.By comparing and analyzing,we can know the GA-BP neural network optimization algorithm.In the whole estimation process,the absolute error is basically in the range of 0-3%,which shows that the prediction accuracy of GA-BP neural network is more accurate.The high and estimation errors are also lower,but in the latter stage,there is still a certain error,but it can be said that the GA-BP neural network estimation method used in this chapter has a good effect in the whole battery SOC estimation process.(3)In order to further accurately estimate the SOC of the battery,the paper uses the cross-section algorithm to optimize the BP neural network to estimate the SOC of the lithium-ion battery.This method uses a two-way search mechanism,that is,the vertical cross and the horizontal cross Combined,a competitive mechanism is generated,which forms the basic search behavior of the cross-section algorithm.In short,it is a one-way search mechanism superior to the genetic algorithm.In the simulation experiment,the results of comparative analysis show that the absolute error is basically in the range of 0-1% in the whole estimation process,which shows that the prediction accuracy of CSO-BP neural network is higher and the estimation error is also Lower,basically there is no "horse tail" phenomenon in the late stage of estimation.This method can estimate the SOC value of the battery more accurately,and verifies the feasibility and effectiveness of the estimation method of the cross-section algorithm BP neural network in the field of battery SOC estimation.