Deep-Learning Based Lithium-ion Battery SOH Estimation Using Multi-Channel Charging Profile and Discharge Capacity

초록

For safe and efficient use of lithium ion battery pack, it is important to monitor the states of battery. Among various states of battery, it is required to estimate SOH (State-of-Health), which represents the performance and life of battery. In this paper, we estimate SOH using various structures of artificial neural network (ANN). We use the measured voltage, current, and temperature of battery cell during charging process as a feature to estimate SOH. We also use the discharged capacity, measured by the coulomb counting method, of battery cell as the feature. We evaluate the performance of various structures of ANN such as feedforward neural network (FNN), convolutional neural network (CNN) and long short-term memory (LSTM) and confirm that the use of discharged capacity significantly improves the SOH estimation performance. © 2022, Korean Institute of Communications and Information Sciences. All rights reserved.

키워드