Techno-economic optimization and hybrid deep learning-based net-load forecasting for intelligent microgrid energy management using HOMER Pro

Abstract The integration of renewable energy resources into electrical grids has increased the need for replicable and cost-effective microgrid design and operation strategies. This paper presents a microgrid planning and energy management framework that combines HOMER Pro-based techno-economic optimization with hybrid deep learning-based forecasting. First, three microgrid configurations comprising photovoltaic (PV) generation, battery energy storage, diesel generation, and grid interconnection are modeled and evaluated based on economic, operational, and environmental performance. The optimal configuration identified using HOMER Pro is subsequently used as the reference architecture for a forecasting framework that employs recurrent neural networks and hybrid deep learning models to predict short-term building load demand and photovoltaic generation. The predicted load and PV generation are combined to estimate the future net-load, providing information that can support battery scheduling and grid interaction. The operational improvements discussed in this study are indicative of the potential benefits of forecast-informed operation and are not derived from a fully integrated forecast-driven dispatch implementation. The results demonstrate that combining HOMER Pro-based techno-economic analysis with hybrid deep learning forecasting provides a practical framework for supporting intelligent microgrid planning and predictive energy management.