Dynamic load scheduling based on virtual energy storage and solar power prediction in smart grid

The intermittent nature of renewable energy resources (RES) poses significant challenges in distributed energy networks. Hence, this study presents a dynamic load scheduling strategy primarily for smart buildings to deplete peak demand and enhance renewable energy utilization. The thermostatically controllable loads, such as refrigerators, are modelled analogously to electro-chemical batteries by exploiting their thermal storage capacity, thereby enabling flexible demand response in a virtual energy storage system (VESS) framework. The LSTM (Long Short-Term Memory) model predicts solar power outputs, addressing the variability and uncertainty inherent in renewable energy sources. These predictions are integrated with real-time thermal storage states to schedule controllable loads dynamically within a demand response framework. Thus, prediction capability is crucial, as both the thermal energy storage capacity of controllable loads and the availability of solar power are variable and can significantly impact energy management. The results demonstrate significant reductions up to 57.14% in grid energy consumption, 42.86% cost savings, and mitigation of generation-load imbalances, underlining the efficacy of integrating advanced predictive analytics and smart storage solutions in smart grid applications.