A Survey on Data-Driven Methods for State of Charge Estimation of Battery

Abstract

Battery Management System (BMS) is a type of an electronic regulator used to protect and monitor the battery, estimating the state and finally indicating it to the users. Precise valuation of the State of Charge (SOC) of the battery is an essential parameter in BMS, where SOC denotes the charge left in the battery. A decent estimation of SOC can lead to the longevity of the battery and avoid hazardous battery failures. Factors like overcharging or over-discharging can be eliminated by using a BMS. Various types of methods are present for accurate estimation of SOC. In this paper, a few Data-Driven methods are surveyed that are used for SOC estimation. Data-Driven methods apply complex algorithms on a huge amount of battery data to learn the behavior of the battery. Hence it outperforms other methods in case the system is not fully understood. A comparative study on the commonly used Data-Driven methods is conducted. Methods like Support Vector Machine (SVM), Fuzzy Logic (FL), Artificial Neural Network (ANN) and Deep Learning have been reviewed. These methods are been compared based on various parameters like computation cost, accuracy, amount of storage error percentage and many more parameters.