It is challenging and valuable to estimate state of health (SOH) of lithium-ion batteries under random, dynamic, and real operation conditions. This work aims to estimate SOH of battery aged by field forklift load profile and research the influence of feature extraction and selection based on different parts of one charging segment on the estimation accuracy. The battery was aged by one synthetic forklift load profile first accompanied by limited reference performance tests (RPT). Then the features were extracted from one segment and its sub-segments. Filter type and wrapper type feature selection methods were combined to select the final features. Support vector regression (SVR) was adopted to estimate the SOH. Finally, the estimation accuracy was evaluated and compared. The results show that the proposed feature selection can improve SOH estimation accuracy effectively. Estimation accuracy is below 1%RMSE based on one sub-segment even the data in the later aging stage not involved in training.
.jpg "Xingjun Li, MA photo")