The Elastic Net Regularized Extreme Learning Machine for State of Charge Estimation

Abstract

The state of charge (SOC) is a crucial component of battery management, and the reliability of its assessment is crucial for predicting battery life. In this paper, an elastic net regularized extreme learning machine is developed for SOC estimation. Unlike traditional neural networks, the extreme learning machine does not require updating all the weights/parameters of the network to obtain accurate SOC estimates, thus, it has a simpler structure. In addition, the elastic net regularization combining the L1 regularization and L2 regularization, can better describe the relationship between current, voltage and SOC. The simulation results show the effectiveness of the proposed method.