Fast Battery SOH Estimation Based on Response Characteristics of Load Surges and GA-ELM

The wide use of Li-ion batteries in various industrial applications has led to a focus on their health status. This paper proposes a novel fast SOH estimation method by combining the response characteristics and the Genetic Algorithm-Extreme Learning Machine (GA-ELM) model. Based on the theoretical manipulations, the dynamic response due to the inrush current has been derived and shows that the response curve due to the inrush current contains information related to the battery SOH. Since there is no explicit relation, GA-ELM with the wavelet transform method is used to extract effective dynamic response voltage data features to estimate SOH without manually adjusting parameters. Experiment results validate that the fast estimation of the battery SOH can be achieved by using the proposed method with a resolution accuracy of 3% in any state of charge (SOC). Compared with other models by the crossover analysis, the method is efficient for feature extraction while providing better performance. This paper aims to highlight the use of load surges response characteristics-to estimate SOH with no additional sensors. It shows the application has a good prospect in the electric vehicles (EVs).