PREDICTIVE ANALYTICS MODEL FOR ENERGY OPTIMIZATION IN ELECTRIC CITY BUSES

Abstract

The increasing adoption of electric buses in urban transportation systems presents new challenges in energy management, route optimization, and battery utilization. To ensure sustainable and efficient operation, accurate prediction of energy consumption under varying driving conditions is essential. This paper proposes a Predictive Analytics Model based on machine learning techniques to optimize the energy economy of electric city buses. The proposed framework leverages real-time and historical operational data—including speed profiles, route topology, passenger load, ambient temperature, and driving behavior—to forecast energy demand and optimize power utilization strategies. The system employs a hybrid learning architecture combining regression-based and ensemble learning models such as Random Forest, Gradient Boosting, and Long Short-Term Memory (LSTM) networks to enhance prediction accuracy. Data preprocessing and feature selection techniques are applied to handle noisy data and identify the most influential parameters affecting energy consumption. The model’s predictive performance is evaluated using key metrics such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and R² score, demonstrating superior accuracy compared to traditional empirical and rule-based approaches. Experimental results show that the proposed model effectively reduces prediction errors and provides actionable insights for route planning, charging infrastructure management, and fleet operation scheduling. The integration of predictive analytics into the electric bus ecosystem supports data-driven decision-making, energy efficiency improvement, and cost reduction, contributing to the development of sustainable and intelligent public transportation systems.