Accurate Surface Temperature Estimation of Lithium-Ion Batteries Using Feedforward and Recurrent Artificial Neural Networks

Abstract

Lithium-ion batteries are an essential component in electric vehicles. A robust battery management system (BMS) must be able to estimate the battery states including state of charge (SOC), state of health (SOH), and, ideally, battery temperature as well. The cells in the pack may experience significant temperature differences during operation, and this would typically be monitored by a multitude of temperature sensors. A surface temperature estimation model can be used to reduce the number of sensors necessary for a pack, which has the side benefit of reducing cost and potentially increasing reliability. In this paper, two data-driven models are proposed to estimate the surface temperature of Li-ion batteries. The first model is based on a feed-forward neural network (FNN), while the second model is based on a recurrent neural network (RNN) with long short-term memory (LSTM). These models are trained and tested using cylindrical cell drive cycle data at a range of temperatures. The LSTM model is shown to be capable of estimating temperature with no more than a few degrees Celsius of error, even for challenging low temperature and varying temperature conditions.