The integration of large-scale renewable energy (RE) poses complex challenges for day-ahead scheduling, characterized by high-dimensional, nonlinear, and tightly constrained optimization problems. Traditional optimization methods often fail to efficiently handle such complexity, leading to suboptimal solutions and high computational costs. Inspired by the growing potential of artificial intelligence (AI) in clean energy systems, this paper introduces an AI-powered scheduling framework based on the Rime Optimization Algorithm (RIME) to address these issues. We develop a high-fidelity Mixed-Integer Nonlinear Programming (MINLP) model that incorporates wind, photovoltaic, hydro, thermal, nuclear, and pumped storage units, with dual objectives of minimizing operating costs and maximizing RE utilization. Leveraging RIME’s bio-inspired mechanisms—soft rime search for adaptive exploration and hard rime piercing for escaping local optima—the algorithm demonstrates superior performance in balancing exploration and exploitation. Extensive simulations on a real-world regional energy base (comprising 470+ units) under four operational modes show that RIME achieves a 6%–10% reduction in operating costs, 100% RE utilization, and 10%–15% faster computation compared to benchmark methods. These results highlight RIME’s potential as an efficient AI-driven tool for the smart and sustainable management of complex renewable-rich power systems.
Zeyi et al. (Thu,) studied this question.