An Online State of Health Estimation Method for Lithium-ion Battery Based on ICA and TPA-LSTM

Abstract

Health management of lithium battery is one of the core functions of Battery Management System (BMS). In order to improve the estimation accuracy of existing SOH estimation method, an online SOH estimation framework based on incremental capacity analysis (ICA) and Time Pattern Attention Mechanism Long Short-Term Memory (TPA-LSTM) network is proposed. Firstly, the aging experiment of lithium battery is carried out, and the smooth IC curve is drawn through voltage local reconstruction and Gaussian Filtering method. Then, a series of IC values within specific voltage range are regarded as health indicator sequences (HIs). The effectiveness of all health indicators is proved by grey relation analysis. Finally, TPA-LSTM network is built to receive HIs and output SOH to realize the numerical mapping from HIs to SOH. The simulation results based on NASA lithium-ion battery aging dataset show that the proposed method has a mean absolute error of less than 0.7%, and the mean absolute error of Hardware-In-the-Loop test results is less than 0.2%.