Online SOC Estimation of Lithium-ion Battery Based on Improved Adaptive H Infinity Extended Kalman Filter

TL;DR

This paper presents a real-time online State of Charge estimation method for lithium-ion batteries in electric vehicles, combining a Thevenin model, recursive least squares, and an improved adaptive H Infinity Extended Kalman Filter, demonstrating superior accuracy.

Contribution

It introduces a novel joint algorithm integrating recursive least squares and an improved adaptive H Infinity Extended Kalman Filter for online SOC estimation, enhancing accuracy and stability.

Findings

The proposed method achieves the lowest estimation errors among compared methods.

It maintains high accuracy under different driving conditions.

The algorithm effectively estimates model parameters and SOC in real-time.

Abstract

For the battery management system of electric vehicle, accurate estimation of the State of Charge of Lithium-ion battery can effectively avoid structural damage caused by overcharge or over discharge inside the battery. Considering that the lithium-ion battery is a time-varying nonlinear system, which needs real-time State of Charge estimation, a joint algorithm of forgetting factor recursive least squares and improved adaptive H Infinity Extended Kalman Filter is proposed for online estimation of model parameters and state of charge. Firstly, Thevenin equivalent circuit model is built in Simulink of MATLAB R2021b, and the model parameters are estimated by forgetting factor recursive least square in real time. Secondly, the improved adaptive H Infinity Extended Kalman Filter is used to estimate State of Charge in true time. Finally, the feasibility of the algorithm is verified by two different lithium-ion battery conditions. The experimental results show that improved adaptive H Infinity Extended Kalman Filter has the highest and most stable State of Charge estimation accuracy than the other three comparison methods. The Root Mean Square Error and Mean Absolute Error are 0.6008 % and 0.3578 % under the Dynamic Stress Test condition, and 1.0068 % and 0.8721 % under the Federal Urban Driving Schedule condition respectively