State of Charge and State of Health Estimation using Neural Networks

Abstract

State of Charge and State of Health Estimation using Neural Networks

1.INTRODUCTION

There is a spiking rise in pollution caused by the adverse effects of carbon-emitting fuel vehicles. Such impacts of pollution cause harm to the environment as well as human health and lifestyle. With this said, there is a booming demand and production of E-vehicles in the market.

new era of vehicles began with the introduction of electric vehicles. Just like a fuel gauge in fueloperated vehicles, battery-operated vehicles or E-vehicles, you may come across the word “SoC” many times in case of your Battery functionality and capacity. SoC stands for State of Charge.

State of Charge

State of charge is the level of charge of an electric battery relative to its capacity. They are measured in units of percentage points. They are used to give you an accurate estimation of how much percentage of charge is left in your battery. Estimation of the SoC of a battery is not an easy task. It varies based on the battery type and the type of application that is using the battery. That being said, SoC estimation is done conveniently using neural networks which not only provides accurate but faster results.

Dataset used and model

The dataset used is of a Panasonic 18650PF lithium-ion (Li-ion) battery. We have used a Multi Layer Perceptron, commonly known as MLP model to predict SoC. MLPs are known to provide better and accurate results compared to traditional linear regression for a battery dataset.