AI-Driven Energy Cost Optimization for Mixed-Use Buildings Using Machine Learning and Renewable Integration

This paper presents an AI-enabled energy management framework for a mixed-use residential-commercial building, designed to minimize grid electricity costs while maximizing the utilization of on-site renewables. A modified IEEE-69 distribution model is adapted to represent the case study, integrating aggregated domestic and commercial loads with solar photovoltaic (PV), biogas generation, and a battery system at the point of common coupling (PCC). Short-term forecasts of cost drivers are obtained using machine learning models, among which the Random Forest regressor achieved the highest accuracy with a Mean Absolute Error (MAE) of $4.11, Root Mean Squared Error (RMSE) of $5.83, and an R² score of 0.97. These forecasts guide an optimization strategy that aligns flexible demand with renewable availability and tariff variations. The monthly energy cost was reduced from $10,742.51 to $8,102.63, yielding a 24.6% saving. The framework demonstrates how predictive analytics combined with cost-aware scheduling can achieve meaningful economic and sustainability gains, offering a compact and transferable solution for future smart city applications to achieve affordable and clean Energy.