Research On Joint Estimation Of State Of Charge Of Lithium-ion Battery Based On Fusion Model

Lithium-ion batteries,which are widely used in new energy storage due to their long cycle life,high energy density,and no pollution,are an important boost to the realization of low-carbon economy.Accurate estimation of the state of charge(SOC)can assist the battery management system to develop a safe and efficient energy management strategy and extend the battery life.This thesis takes the lithium iron phosphate battery,which is widely used in new energy storage,as the research object,and studies its open circuit voltage(OCV)characteristics,modeling scheme and SOC estimation method.An accurate OCV-SOC curve is the basis to ensure the accuracy of battery modeling.This thesis analyzes the principle and process of OCV-SOC curve acquisition by incremental OCV test and low current OCV test.Through experiments,it is found that the incremental OCV test cannot describe the nonlinear characteristics of the nontesting point,and the low current OCV test has a certain polarization effect.To solve this problem,a high-precision OCV-SOC curve acquisition method is proposed in this thesis.This method takes the low current discharge curve fitted in piecewise form as the optimization object,designs constraints based on the incremental OCV test data and first-order RC equivalent circuit model,and uses the differential evolution method to find the optimal OCV-SOC curve.The experimental results show that the proposed method can accurately simulate the OCV characteristics of Lithium-ion batteries.As a concentrated mathematical expression of the external characteristics of lithium-ion batteries,the battery model is an important basis for SOC estimation.This thesis introduces four kinds of equivalent circuit models widely used at present,and the second-order RC equivalent circuit model with simple structure and high precision is selected as the research object through analysis and comparison.On this basis,the principle and process of recursive least square(RLS)identification of battery the model are analyzed.It is found through experiments that the equivalent circuit model of second-order RC link obtained by the RLS is difficult to distinguish RC link with large and small time constants effectively,and the model has certain limitations in accuracy and adaptability.To solve this problem,a fusion model construction scheme is proposed in this thesis.The weighted average optimization method is used to fuse multiple battery models to obtain large RC stages and ohm internal resistance parameters.Then the fusion parameters are used as constraint factors and the RLS is used to identify the remaining parameters of the battery model.The experimental results show that the proposed scheme can provide more accurate parameter identification results and improve the accuracy and adaptability of the battery model.In the SOC estimation part,this thesis focuses on the sliding mode observer(SMO)and particle filter(PF)and applies them to the SOC estimation of the battery.The results show that the PF has good anti-interference ability to the system noise,but it is affected by particle degradation and dilution and SMO is robust to model parameter disturbance but sensitive to system noise.Based on the above analysis,a joint SOC estimation method of sliding mode observer and particle filter with initial value compensation function is proposed in this thesis.In this method,a SOC initial compensation strategy is designed based on the observer theory to improve the convergence speed of the algorithm.Then,the particle filter is taken as the optimization object,the voltage observation value is introduced in the process of particle diffusion,and the sliding mode observer is simulated to guide the particles to move towards the high likelihood region to take into account the advantages of both.Experimental results show that the proposed method can accurately estimate SOC under normal environments,noise disturbance,model parameter disturbance,initial SOC error disturbance,and different test conditions,which greatly improves the accuracy and adaptability of the SOC estimation algorithm and has a certain practical value.