Machine Learning For Fuel Consumption Prediction

In recent years, real-time fuel consumption prediction and driving profile classification have gained prominence due to their impact on vehicle efficiency and environmental sustainability. This project focuses on leveraging machine learning algorithms to predict fuel consumption and classify driving profiles based on ECU (Engine Control Unit) data. The existing system utilizes XGBoost, SVR (Support Vector Regression), and Ridge Regression. The proposed system aims to enhance predictive accuracy and profile classification by incorporating Random Forest, Logistic Regression, and Adaboost algorithms. Driving profiles are categorized into five distinct classes: Sporty, Eco, Calm, Normal, and Aggressive, based on fuel consumption patterns. This approach not only provides insights into driving behavior but also supports the development of adaptive driving strategies and fuel-saving measures. By integrating advanced machine learning techniques, the project seeks to improve both vehicle performance and environmental impact.

Keywords: Machine Learning, Fuel Consumption, Driving Profile Classification, ECU Data, XGBoost, SVR, Ridge Regression, Random Forest, Logistic Regression, Adaboost