Abordagem para otimização do despacho hidrelétrico em usinas hidrelétricas baseada em processos decisórios de Markov e otimização por enxame de partículas

Abstract

This work is related to the research project "Optimization of Hydropower Plant Operation Through Minimization of Losses in the Generation Process" conducted under the ANEEL R&D framework, developed by UNIFEI (Federal University of Itajubá) in partnership with Jirau Energia - the concessionaire of the Jirau Hydroelectric Plant, and presents an innovative methodology for optimizing energy dispatch in hydroelectric plants. Thus, the study is particularly focused on UHE Jirau, one of the largest in Brazil, located in the Madeira River basin. The main objective of this research is to develop and implement an effective approach that integrates Markov Decision Processes (MDP) and Particle Swarm Optimization (PSO) to optimize energy dispatch, minimizing losses and maximizing the performance of the generating units (GU). The need for this approach is emphasized by the inherent complexity of managing multiple generating units and the load losses caused by sediment accumulation at their water intakes. The proposed methodology is divided into several phases, starting with the modeling of the operational characteristics of the generating units and the development of an objective function that includes maximizing energy efficiency and energy generation with the same water consumption. The integration of MDPs allows for real-time decision-making, considering multiple scenarios and operational states of the units, while PSO is used to explore optimal solutions in an expanded search space, thus ensuring the attainment of a globally optimal solution. The partial results demonstrate that the application of the proposed methodology is viable, allowing for increased efficiency in the operations of UHE Jirau. Compared with historical data and traditional dispatch methods, there was a reduction in the turbinated flow, accompanied by increases in the dispatched power and overall yield obtained for the plant. The analysis of the results also suggests that the developed strategy is robust and adaptable to different operational conditions, providing a reliable framework for optimal energy dispatch. This work not only meets the specific needs of UHE Jirau but can also significantly contribute to the body of knowledge in electrical engineering, with potential applications in other hydroelectric plants facing similar challenges. Furthermore, the methodological advances and solutions developed promote the efficient use of natural resources, aligning with the goals of sustainability and energy efficiency.