State of Charge Estimation of Low-speed Electric Vehicle Battery using Back Propagation Neural Network

Abstract

This paper introduces a back propagation (BP) neural network for estimating the state of charge (SoC) of the battery pack in low-speed electric vehicles (LSEV), such as electric bicycles, tricycles, electric motorcycles, and so on. From the perspective of test and maintenance, the SoC estimation with machine learning techniques provides a method of predicting the health status and life-span of the battery in LSEV. Typically, SoC is associated with the factors of open circuit voltage (OCV), charging and discharging current, temperature and charging cycle. Data set collection have been finished by a smart electric bicycle charging platform provided by POWER+WWW. In this paper, the BP neural network theory including the training process is introduced briefly. The improvement of prediction accuracy and the model optimization are discussed. In the end, the experimental results in comparison with the real test value are presented.