Optimizing solar energy utilization in facilities using machine learning-based scheduling techniques: A case study

This study introduces an approach to improving the utilization of solar energy in facilities by integrating advanced machine learning (ML) techniques into solar power scheduling. Traditional methods, often constrained by static schedules, fail to adequately adapt to the inherently dynamic and intermittent nature of solar energy. Our approach overcomes these limitations by employing ML algorithms to accurately predict solar generation patterns, enabling more efficient scheduling of electrical appliances. This methodology was applied to a facility equipped with a 5kW photovoltaic system, resulting in a significant reduction in grid dependency by more than 26%. This marked decrease in grid imports underscores the effectiveness of our approach in optimizing solar energy use, particularly in settings where traditional scheduling methods fall short. The study demonstrates the practical benefits of ML in managing solar energy resources to reduce dependence on conventional power grids, thus contributing to more sustainable energy practices. The findings of this research have far-reaching implications, suggesting a notable advancement in solar energy management towards more adaptive, data-driven solutions and paving the way for broader applications in various sectors seeking to maximize renewable energy use. • The implemented ML-based scheduling techniques optimized solar energy utilization. • The scheduling algorithm reduced grid electricity imports by over 26%. • The ML model predicted solar generation with ANMAE slightly above 45%. • Achieved significant savings even with limited flexibility in schedulable loads. • Enhanced economic value by aligning energy consumption with solar generation.