SoC Estimation for an Electric Two-Wheeler Using Machine Learning Techniques

Due to the energy efficiency and commitment to the environment, electric vehicles (EVs) have significantly increased in popularity. Accurately estimating the battery’s state of charge (SoC), which indicates the remaining energy available for vehicle operation, is a crucial component of EV operation. For efficient battery management, maximum vehicle range, and avoiding sudden battery discharge, accurate SoC calculation is crucial. In this abstract, a simple method for calculating the SoC of an electric two-wheeler vehicle is presented. The suggested methodology makes use of machine learning methods to predict the intricate connections between several vehicle characteristics and the battery’s SoC. Machine learning’s capacity to recognize patterns and generate precise predictions based on past data is its main benefit. An extensive dataset encompassing data on the battery voltage, current, temperature, speed, and other pertinent factors is gathered in order to create the SoC estimation model. The SoC estimation model is trained using a variety of machine-learning approaches, such as linear regression and logistic regression. It has been found that the linear regression model provides more accurate predictions as compared to the logistic regression model.

I. Introduction

Amid the rising popularity of eco-friendly EVs, precise SoC estimation is essential for efficient battery management, extended driving range, and reliable operation. This paper focuses on optimizing SoC estimation to enhance EV performance and sustainability [1]. SoC estimation is a critical aspect of battery management, as it reflects the remaining energy in the battery, developing an accurate and reliable SoC estimation method, enabling efficient battery utilization, enhanced driving range, and improved overall performance of electric two-wheelers [2]. The accurate estimation of the SoC is crucial for the efficient management of lithium-ion batteries in EVs which includes estimating model parameters developing an effective battery management system and enhancing the performance and range of 2-wheeler EVs to ensure reliability and sustainable transportation [3]. Accurate SoC estimation ensures optimal battery utilization, extends battery life, and enhances EV performance. Using advanced smart computing techniques SoC estimation of li-ion battery can be obtained. The method used leverages machine learning algorithms to model complex battery behavior and achieves improved SoC estimation accuracy. Real-world data from Li-Ion batteries in EVs is used to validate the effectiveness of the proposed approach, showcasing its potential in advancing battery management systems for sustainable and efficient EV operation [4]. Developing effective battery management systems for optimizing battery performance in various applications, including electric vehicles and renewable energy storage. Traditional approaches to parameter estimation may not adequately capture the complex interactions between battery parameters, limiting the accuracy and diversity of synthetic datasets used for battery modeling and simulation [5]. The SoC estimation model uses techniques like SoC interpolation based regression and driving accumulate based classification, which are cloud-based driving data, as well as a traditional machine learning approach with Support Vector Regression (SVR) to predict SoC while taking into account variables like driving style, temperature, and cell age that affect the advancement of EV technology and performance [6]. In order to improve battery health, safety, and performance, multistage converters using an auto regressive integrated moving average (ARIMA) model were used for the SoC estimate of lithium-ion batteries in EVs. The enhanced Random Forest Regression (RFR) technique uses the differential search algorithm (DSA) to optimize the RFR parameters and to demonstrate the method’s efficacy under various testing scenarios, which enhances EV safety and dependability [9]. For predicting the EV range, state of charge, and state of health of Li-ion batteries, some techniques are used, such as Cascaded Feed forward Neural Network (CFNN) along with the Feed forward Back Propagation Neural Network (FBPNN) and AutoML[10]-[11]. A back-propagation neural network method is used to determine the SoC of low-speed EV batteries [12]. For the best battery management and vehicle performance, the SoC calculation must be accurate. This study shows how machine learning methods can forecast SoC, improving the performance of electric two-wheelers and advancing EV technology. Though there are few papers available in the literature, not many authors have used this particular method to obtain the SoC of a two-wheeler battery. Therefore, this paper expresses the possible forms of linear regression and logistic regression and gives a simple solution for SoC estimation.