AI-Integrated Energy Management System for Renewable–Thermal Hybrid Power Systems: Forecasting, Uncertainty-Aware Scheduling and Operational Decision Support.

This research presents a structured artificial intelligence-integrated Energy Management System (AI-EMS) framework for renewable–thermal hybrid power systems operating under increasing renewable penetration, variable demand, forecast uncertainty, ramping conditions, and operational constraints. The proposed framework integrates data acquisition and preprocessing, renewable generation forecasting, load forecasting, net-load estimation, ramp prediction, uncertainty and scenario modeling, thermal generator scheduling, unit commitment, economic dispatch, ramp-constrained optimization, reserve management, renewable curtailment analysis, constraint validation, monitoring, and operator-oriented decision support. The architecture separates AI-based prediction from optimization-based operational decision-making, allowing forecast information to be translated into physically feasible generation schedules under generator, ramp-rate, storage, power-balance, reserve, and other operating constraints. A central component of the framework is the connection between renewable and load forecasting, net-load analysis, ramp-event detection, flexibility assessment, and thermal generation scheduling. The system is designed to identify operational conditions such as rapid demand increases, renewable-generation reductions, evening net-load ramps, sudden wind or solar variations, and combined renewable–demand fluctuations. These predictions can subsequently inform flexibility requirements, reserve assessment, generator commitment, dispatch decisions, and renewable curtailment management. The framework further incorporates uncertainty-aware operational planning through probabilistic and scenario-based representations. A rolling-horizon and closed-loop decision process enables the system to compare forecast, scheduled, and actual operating conditions and update operational decisions when significant deviations occur. This supports an adaptive energy-management workflow rather than a one-time forecasting process. The research framework is designed for systematic experimental evaluation under representative operating conditions, including high renewable penetration, demand variation, renewable ramp events, forecast errors, energy-storage operation, generator contingencies, abnormal operating conditions, and high-uncertainty scenarios. Evaluation is structured around measurable engineering indicators including forecasting accuracy, operating cost, renewable utilization, renewable curtailment, thermal generation, generator starts and stops, reserve margin, ramp adequacy, ramp shortfall, schedule feasibility, constraint violations, optimization execution time, and re-optimization requirements. The proposed work also defines validation procedures covering module-level, integration-level, system-level, stress, fault, performance, and end-to-end testing. Reproducibility is supported through controlled datasets, code and model versions, configuration parameters, optimization settings, evaluation periods, and other experimental conditions. Overall, the framework provides a research-oriented foundation for combining artificial intelligence forecasting with mathematical optimization and operational decision support for renewable–thermal hybrid power systems. It is intended to support transparent, constraint-aware, adaptive, and human-supervised energy management while providing a structured basis for future simulation, benchmarking, uncertainty analysis, energy-storage integration, security-aware scheduling, and scalable intelligent-grid applications.