Artificial Intelligence for Sustainable Applications

16Review of Non-Recurrent Neural Networks for State of Charge Estimation of Batteries of Electric Vehicles

R. Arun Chendhuran¹ and J. Senthil Kumar²*

¹Department of Electronics and Instrumentation, Bannari Amman Institute of Technology, Erode, Sathyamangalam

²Department of Electrical and Electronics, Bannari Amman Institute of Technology, Erode, Sathyamangalam

Abstract

Advancements in artificial intelligence and machine learning (ML) has created paradigm-shifts in estimation of state-of-charge of batteries. Machine learning is a sub-set of artificial intelligence that facilitates computers to learn from data through algorithms, without explicitly programming it. Machine learning has evolved to model energy storage devices to determine their state of charge. Accessibility to battery-data and reduced computation time has made data-driven models based on machine learning more attractive. Non-recurrent SoC estimation techniques such as Feed-forward Neural Networks (FNNs), Radial Basis Functions (RBF), Extreme Learning Machine (ELM) and Support Vector Machine (SVM) are reviewed in this paper. It is recommended that SoC Estimation Techniques under comparison should preferably share common data-sets (both training and testing) and learnable parameters, else the comparison may be biased.

Keywords: Battery, SoC, electrochemical, equivalent circuit, impedance

16.1 Introduction

Global warming has demanded electrification of transport in many countries. Electric vehicles have superior ...