Estimating the state of charge of lithium-ion batteries using different noise inputs

Abstract

State of charge estimation (SOC) is the most significant functionality of a vehicle's battery management system (BMS). The methods for this estimation are conventionally oriented towards model-based methods. As part of this paper, we introduce a first order equivalent circuit estimation approach known as the Thevenin model, along with an extended Kalman filter (EKF) approach to accurately estimate the SOC. We then deploy and simulate it in MATLAB by using a reference load profile from the new European driving cycle (NEDC). Afterwards, the simulation results are reviewed based on various initial noise values, and the results are compared to those of other EKF algorithms. According to the results, SOC estimation accuracy has significantly increased as a result of the improvements made. Specifically, the root-mean-square error decreased from 0.0068 to 0.0020.