Voltage variation with hybrid pulse as a novel indicator for lifetime estimation of Li-ion battery using in smart grid

Abstract

Batteries used in smart grid have a wider lifetime compared with electric vehicles (EVs) applications. Thus, accurate online lifetime estimation is significant for these low-capacity batteries because of the faster aging process. The dispersion of the batteries rises with aging, leading to a decrease in the robustness of some life-time estimators. In this paper, we introduce a novel battery lifetime estimator, which contains six points extracted from voltage variation with a hybrid pulse. The BP Neural Network (BPNN) is used to train the relationship between the points and lifetime to reduce the dispersion of different batteries. Several Li-ion batteries are discharged to the 50% SOH through profiles with different depth of discharge (DOD) and mean state of charge (SOC), to verify the accuracy and robust of the proposed method. The estimation errors of the tested batteries are all less than 1%, which shows a good performance on accuracy and robustness.