Hyperparameter Optimization for SOC Estimation by LSTM with Internal Resistance

Abstract

State of charge (SOC) is an important indicator to describe battery performance. However, most of the training input data for SOC estimation using deep learning algorithm do not consider the battery internal resistance. In this paper, a method for SOC estimation of lithium-ion batteries based on the LSTM neural network is proposed, and the battery voltage, current, temperature and internal resistance are taken as the training input data of the LSTM model. When the correlation function is determined, the influence of different hyperparameters of the LSTM model with battery internal resistance as the training data on the SOC estimation results is analyzed and compared to determine the optimal setting of the hyperparameters of the model. The experimental results show that the model with the battery resistance as the training input data has higher accuracy. On this basis, the LSTM model with appropriate hyperparameters has the advantages of high accuracy and fast convergence in SOC estimation.