Health factor analysis and remaining useful life prediction for batteries based on a cross-cycle health factor clustering framework

Accurate remaining useful life (RUL) is critical for battery management systems. A clustering prediction framework based on the K-means algorithm is proposed to improve the accuracy of predicting RUL of lithium-ion batteries (LIBs). The model addresses the battery capacity recovery characteristics. The influence of resting time on battery life is analyzed on a time scale. Data are differentiated using the K-means algorithm through two cross-cycle health factor analyses. According to the data distinguished by different characteristics, support vector regression (SVR) based on time series and the radial basis function (RBF) are used for prediction. The combination of prediction methods can be used to predict global degradation in the case of insufficient information. The LIBs use cross-cycle health factors to analyze their recovery capacity. The approach can effectively obtain the inflection point information of battery capacity recovery and improve the prediction accuracy. At present, it is blank to improve the prediction algorithm according to the characteristics of capacity recovery. The combination algorithm significantly improves the accuracy of predicting the remaining life of LIBs. Comparative studies confirm that the prediction model using the clustering framework is more accurate than other machine learning models with limited data samples. Compared with the unimproved SVR algorithm, the average root mean square error of the four groups of battery data after the improvement is reduced from 1.32e-2 to 1.19e-2. This model ensures the accuracy and reliability of prediction.