A data-driven fuzzy information granulation approach for battery state of health forecasting

This paper proposes an estimation method for battery state of health (SOH) based on fuzzy information granulation. The time interval of the equal charging current difference (TIECCD) in the constant voltage charging mode is extracted as a feature. Then, grey relational analysis is employed to find the optimal health indicators. After granulating fuzzy information on SOH and health indicators, the three parameters Low, R and Up are obtained to characterize the SOH range. In the implementation, the least squares support vector machine (LSSVM) is selected to construct the nonlinear regression model of the parameters and the granulated feature data to realize the prediction of the trend of battery health. Finally, one reference group is set as contrast, and the prediction results based on experimental data prove the superiority of the proposed method.