A Dynamic High-Order Equivalent Modeling of Lithium-Ion Batteries for the State-of-Charge Prediction Based on Reduced-Order Extended Kalman Filtering Algorithm

Abstract

Detection of battery power has always been the core of the battery management system of electric vehicles, and the fast and accurate estimation of charged state can guarantee the safe operation of electric vehicles. The key to improving accurate state-of-charge estimation is an appropriate model establishment coupled with a suitable estimation algorithm. This research seeks to adopt and accomplish a lithium-ion battery state-of-charge estimation based on the Gaussian function to build up the open-circuit voltage algorithm. A reduced-order extended Kalman filtering algorithm is proposed with hybrid pulse power characterization parameter identification to estimate the battery characterization state-of-charge. The model’s parameters in different state-of-charge points are calculated through the lithium-ion battery’s charge and discharge process; the 2RC modeling correction method and Reduced-order extended Kalman filter method are used separately based on the High-order equivalent 2RC modeling. The Experimental results show that the above method can achieve state-of-charge estimation more accurately and conveniently, providing a certain reference value for the rational management and distribution of power lithium-ion batteries. The maximum error of state-of-charge estimation based on the established high-order equivalent 2RC model using the Reduced-order extended Kalman filtering algorithm is less than 1.85%. The REKF algorithm achieved a maximum voltage error of 0.0409V and an average error of 0.0299V and therefore can satisfy the accuracy of the battery management system application needs.

Keywords: Lithium-ion battery; state-of-charge; high-order equivalent 2RC modeling; open-circuit voltage; parameter identification; reduced-order extended Kalman filtering algorithm

DOI: 10.7176/JETP/11-3-03

Publication date:June 30^th 2021