State-of-charge estimation method of lithium-ion batteries based on long-short term memory network

Abstract

This paper presents a method to estimate the state-of-charge (SOC) of lithium-ion batteries based on long-short term memory network (LSTM). The method is mainly composed of two parts: (1) A linear neural network is used to identify the parameters of second-order equivalent circuit model. (2) A LSTM network is built to estimate the SOC of lithium-ion battery. The linear neural network is trained using the American Dynamic Stress Test Condition (DST) dataset of battery (1st cycle), while the LSTM network is trained using the Chinese Standard Operating Condition (QCT) datasets of battery (1st, 10th and 20th cycle). After that, the trained LSTM network is tested using DST datasets of the batteries and QCT datasets under different temperatures. The results show that the LSTM network can accurately estimate the SOC of lithium battery under different temperatures, different working conditions and lifespan.