Redução do espaço de busca para o problema de unit commitment térmico através de uma matriz de relevância

Abstract

The growing increase in global energy and electrical demand implies the need to increase energy generation capacity. Therefore, the use of non-polluting and renewable sources has intensified, becoming a major field of study in engineering. Despite the increase in generation capacity from these sources, some countries resort to generation through thermoelectric plants, which, despite being more expensive, provide greater reliability in meeting demand. The independence of climatic conditions makes the use of generation from thermoelectric plants attractive during periods of water scarcity or sudden increase in demand. The high number of thermoelectric generation projects and the inherent complexity in their operation are the focuses of this work, where the objective was to obtain a new methodology to support the resolution of the problem known, in the literature, as Thermal Unit Commitment (TUC). The combinatorial explosion inherent to the TUC, referring to the operating decisions of the Generating Units (GU) and the temporal coupling between these units, makes the daily planning of the operation of thermoelectric units a difficult problem to solve. Given this scenario, the main contribution of this work is the proposition of a Relevance Matrix (RM) that aims to reduce the search space for solutions by indicating the relevance of activating each generating unit throughout the entire planning period under study. This reduction provides a reduction in the number of decision variables, a reduction in processing time and even guarantees convergence in some situations. The RM is obtained through a constructive heuristic procedure, considering sensitivity indexes widespread in the literature for forming Hybrid Priority Lists (HPLs). The HPLs are responsible for indicating which generating units are most and least relevant to the daily planning operation. The developed methodology was validated and compared using thermoelectric generation systems widely disseminated in specialized literature. These generation systems consider, among other points: generation ramp, penetration of renewable sources and consideration of transmission lines. The results obtained demonstrated the efficiency of using MR by drastically reducing the computational effort in addition to obtaining reduced planning costs in a simple-to-implement hybrid algorithm.