Volume 27: Closing Carbon Cycles – A Transformation Process Involving Technology, Economy, and Society: Part II

Generalized Regression Neural Network Based State of Charge Estimation for Lithium-Ion Battery with Ambient Temperature Consideration Gaoqi Lian, Min Ye, Qiao Wang, Meng Wei, Xinxin Xu

Abstract

Obtaining an accurate mapping relationship between lithium-ion battery open-circuit voltage (OCV) with the state of charge (SOC) at different ambient temperatures is the basis for its accurate SOC estimation in the whole ambient temperature range. However, the experimental test of the OCV-SOC correspondence takes a lot of time; and it is obviously impossible to perform the test at all temperatures. To achieve accurate SOC estimation at different ambient temperatures with a lower experimental cost, a model-based SOC estimation method is proposed in this paper. First, based on generalized regression neural network (GRNN), an OCV-SOC mapping model for the whole ambient temperature range is established. Second, a new diagonalization of matrix adaptive cubature Kalman filter (DMACKF) is proposed, which enhances the filtering stability and realizes the adaptive update of noises in the recursive process. Finally, combined with the forgetting factor recursive least squares (FFRLS) algorithm, the proposed SOC estimation method is verified under the DST conditions at three temperatures. The root mean square errors (RMSEs) of SOC estimation results are within 0.4% at each temperature.

Keywords lithium-ion batteries, state of charge, different ambient temperature, generalized regression neural network