Research On Modeling And SOC Estimation Of Lithium Battery For Power Grid Energy Storage

With the increase of uncertainties of power supply and load in power grid and the reduction of lithium battery cost,more and more battery energy storage systems have been applied in power grid.Battery Management System(BMS),as the key equipment of power grid energy storage system,is quite different from the battery and its management system for electric vehicles,which has been studied extensively.This thesis chooses a lithium iron phosphate battery for grid energy storage as the research object,designs the experimental scheme,establishes the circuit model,and optimizes the state of charge(SOC)estimation scheme to provide a reference for the construction of the battery management system for grid energy storage.The specific work is as follows:(1)According to the performance index and possible actual operating environment of the storage battery,the experimental equipment is selected,the experimental procedure and content are designed,the experimental data are recorded,and the data are cleaned and normalized,which provides data support for the modeling and verification of the charging state estimation scheme.(2)Based on the experimental data,the circuit model of the battery is optimized,the parameters of the circuit model are estimated offline,and the specific circuit model of the specific battery is generated,which can be embedded in the battery management system to provide support for the efficient management of the battery management system.(3)Based on the established circuit model and simulating the actual operation data as input,the selection and estimation accuracy of SOC estimation algorithm are studied,and the estimation effect of KF series filtering algorithm and BP algorithm based on neural network model on SOC of this kind of grid energy storage battery are compared.The comparison results show that the two estimation methods are feasible for the lithium iron phosphate batteries studied in this paper.The KF series filtering algorithms have better performance and can be used in battery management system.The modeling method and SOC estimation of grid storage batteries can be applied to batteries of other specifications and have certain reference value.