Enhanced Electric Two-Wheeler Performance with Machine Learning-Driven SoC Estimation

Electric vehicles (EVs) are becoming more and more popular because of their environmental commitment and energy efficiency. One of the most important aspects of operating an EV is accurately estimating the battery's state of charge (SoC), which indicates the amount of energy left for vehicle operation. Accurate SoC calculation is essential for effective battery management, maximizing vehicle range, preventing sudden battery discharge and enhanced user experience. In this paper, an approach for estimating an electric two- wheeler vehicle's SoC is presented which uses machine learning techniques to forecast the complex relationships between various vehicle attributes and the state of charge of the battery. The primary advantage of machine learning is its ability to identify patterns and produce accurate predictions based on the historical data. To develop the SoC model, a dataset including information on the battery voltage, current, temperature, speed, and other relevant parameters is collected. Machine-learning techniques, such as Support Vector Machines (SVM), K-Nearest Neighbors (k-NN), Ridge Regression, Lasso Regression, Gradient Boosting, Decision Tree, and Random Forest are used to train the model. It has been found that models with high prediction accuracy, such as Random Forest and Decision Tree, show better performance for estimating SoC.