Optimizing Forecasting of Rooftop Solar PV Energy Generation for Peak-Load Management

To ensure a stable and sustainable power grid, it is imperative to implement sophisticated management strategies that effectively address the inherent variability and unpredictability of renewable energy generation. This research focuses on developing an advanced energy management system for buildings with photovoltaic (PV) systems, integrating continuous energy monitoring and forecasting intelligence. Through the examination of past data, meteorological trends, and current energy consumption patterns, the system forecasts future energy needs, identifies peak load periods, and optimizes energy consumption to maximize on-site solar generation. The next-day forecasting approach for commercial photovoltaic systems is explored through the integration of weather prediction models and technical specifications of PV plants to predict power output, incorporating battery storage and optimal control algorithms to address over-generation and reduce reliance on conventional power sources. Performance is validated using RMSE and MAE metrics. The study also examines energy consumption prediction in buildings with a com-bination of statistical analysis, time-series modelling, and machine learning methods applied to a five-year dataset (2020– 2024), utilizing ARMA, ARIMA, and SARIMA models to generate prediction of Solar Irradiance. Additionally, it highlights the role of PV systems in office buildings for peak shaving, improving economic efficiency and grid stability. For residential applications, a novel residential energy management system (REMS) is proposed to manage power fluctuations, leveraging Python for effective control and optimization.