Research On The Method Of Estimating The State Of Charge For Lithium-ion Battery Based On Electrochemical Model

The estimation of state of charge of lithium-ion battery is the core function of battery management system.Due to improving the accuracy of SOC estimation,the service efficiency of lithium-ion battery pac can be improved,the service life can be prolonged and the reliability and security of the lithium-ion battery can also be guaranteed.However,accurate battery modeling is the basis of accurately estimating the state of charge of battery.In this paper,we study the model of extension single particle and realize the SOC online-estimation on the base of the Pseudo two dimensional model and the single particle model.The main contents are following:(1)Based on the pseudo two dimensional model,an approximate solution method of lithium ion concentration at the boundary between positive electrode-collector and negative-collector is proposed.First of all,on the basis of P2 D model,we analyze the relationships between the extreme value of electrolyte concentration in the boundary and the discharge rate;Then,we adopt linear fitting method to get the function relation between the extreme value of electrolyte concentration and the discharge rate.Once electrolyte concentration reaches threshold,we keep it unchanged;Finally,we adopt BP neural network to train the electrolyte concentration at negative electrode current collector interface and positive electrode current collector interface and current density in the present moment and the relaition of the electrolyte concentration at negative electrode current collector interface and positive electrode current collector interface in the last moment,building the foundation for the subsequent extension of the single particle model.(2)Due to the great error of single particle model under the working condition of high-rate electric discharge,the expansion of single particle model with higher precision is proposed.Firstly,the analyzed rule of electrolyte concentration change is used to approximately solve liquid diffusion overpotential with the introduction of approximate solution of solid phase diffusion and enriching ohm polarization process,ensuring the precise of the extension of the single particle model is higher.Then,we analyze the sensitivity of the related electrochemical parameters and utilize genetic algorithm on the related parameters for the off-line identification.(3)Based on the the extension single particle model,online SOC estimation algorithm is studied.According to the accuracy analysis of single particle model,we find that the time-varying of correction parameters can improve the precision of the extension of the single particle model.So,in this paper,we adopt the extended kalman filtering estimator and the correction parameters are included in the state variables with SOC estimation.Considering the pure simulation experiments in this paperand noise in practical application,we may use the method of Covariance Matching to improve EKF algorithm,further improving the SOC estimation precision.Finally,in order to reduce the impact of aging on SOC estimation,we analyze the relationship between ?U-?Q and the SOH based on NASA data and propose a correction method of battery capacity and compensate the aging of the SOC,improving the accuracy of the SOC estimation.