Machine Learning-Based Eco-Driving Level Evaluation System for Electric Bus Stop Entry and Exit Analysis
DOI:
https://doi.org/10.62643/Keywords:
Eco-Driving, Machine Learning, Random Forest, SVM, KNN, Electric Vehicles, Driving Behavior Analysis, Sustainable Transportation, Predictive Modeling, Smart MobilityAbstract
The rapid adoption of electric vehicles (EVs) in public transportation has created new opportunities for improving energy efficiency and sustainability. Among various operational factors, driving behavior plays a crucial role in determining energy consumption, battery lifespan, and passenger comfort. This project presents a Machine Learning (ML)-based Eco-Driving Level Evaluation System designed specifically for analyzing electric bus behavior during stop entry and exit phases. These phases are critical as they involve frequent acceleration and deceleration, significantly impacting energy usage. The proposed system integrates a user-friendly graphical interface developed using Tkinter with powerful ML algorithms such as Random Forest, Support Vector Machine (SVM), and K-Nearest Neighbors (KNN). The system allows users to load real-world datasets in CSV format, train selected ML models, and predict ecodriving levels based on input driving parameters. Key features considered for prediction include mean deceleration, peak deceleration, regenerative braking ratio, maximum jerk, acceleration variance, and energy consumption per departure. The workflow begins with dataset loading and preprocessing, followed by splitting the data into training and testing sets. The selected model is trained using historical driving data and evaluated based on accuracy metrics. Once trained, the system enables real-time prediction of eco-driving levels by taking manual inputs from users. The predicted output is categorized into five levels ranging from aggressive driving to excellent eco-driving behavior, providing meaningful insights into driver performance. The system’s GUI enhances usability by offering options for model selection, dataset upload, training, and prediction in a visually intuitive manner. Additionally, full-screen functionality improves user experience during demonstrations or academic presentations. This project contributes to sustainable transportation by promoting energy-efficient driving practices through data-driven insights. It can be used by transport authorities, fleet managers, and researchers to monitor and improve driving behavior, ultimately reducing energy consumption and environmental impact. The modular architecture also allows for future enhancements such as real-time sensor integration, cloud deployment,and advanced deep learning techniques. Overall, the proposed system demonstrates the effectiveness of machine learning in evaluating and improving eco-driving behavior in electric public transport systems, making it a valuable tool for smart mobility solutions.
Downloads
Published
Issue
Section
License

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.













