AI-Powered Control and Demand-Side Optimization in Decentralized Renewable Energy Systems: Architecture, Applications, and Insights

Decentralized renewable energy systems, such as photovoltaic (PV)-battery microgrids and hybrid plants, require control strategies to operate under uncertainty and communication constraints. This paper presents a technical review of AI-driven energy optimization in decentralized renewable systems across five main application areas: load forecasting and demand prediction, demand-side management and smart scheduling, predictive maintenance and fault detection, AI-based control architectures, and AI-IoT integration. For each area, representative artificial intelligence methods are mapped to key microgrid interfaces, which include PV inverters, storage units, and energy management controllers such as the Energy Management System (EMS) and Home Energy Management System (HEMS). The review also synthesizes reported outcomes through a task-method-result comparison, highlighting results such as about 51% cost reduction in smart-home scheduling and more than 90% accuracy in PV hotspot fault classification in representative studies. Based on this synthesis, the paper identifies major research gaps related to dataset standardization, edge-level validation, interoperability, and explainability. Finally, these insights are linked to the design of IGT Optima, showing how forecasting, demand-side management, maintenance analytics, and supervisory control agents can be coordinated within a deployable edge-cloud workflow.