Mathematical optimization versus Metaheuristic techniques: A performance comparison for reconfiguration of distribution systems

• Autor(a) principal: Silveira, Christoffer L. Bezão [UNESP]
• Data de Publicação: 2021
• Outros Autores: Tabares, Alejandra [UNESP], Faria, Lucas Teles [UNESP], Franco, John F. [UNESP]
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1016/j.epsr.2021.107272; http://hdl.handle.net/11449/206297
• topic: Classical optimization; Distribution systems; Minimization of power losses; Performance comparison; Reconfiguration; Soft computing

Reconfiguration is a complex combinatorial problem in which the topology of distribution systems is modified by the opening/closing of interconnection switches aiming techno-economic benefits (e.g., minimization of losses). Numerous optimization methods have been developed to solve the reconfiguration problem, although a comparative analysis of their performances is still a challenging task due to the nature of the methods, differences in their implementation, and used computational equipment. To fulfill that gap, this paper assesses classical models along with metaheuristics already applied in the specialized literature considering the reported losses and computational effort. To eliminate differences due to implementation and equipment, two proposed metrics are assessed using a reference specialized power flow: ‘equivalent time’ and ‘equivalent number of power flows’. The quality of the solutions was compared for standard test systems (33, 136, and 417 buses) and a ranking of the methods was produced. It was concluded that linear and conic programming models find the optimal solution for low and medium-size systems; moreover, the linear model requires lower computational effort than the conic and the nonlinear programming formulations. On the other hand, it was verified that metaheuristics need lower computational effort and provide better solutions for large-size systems compared to classical optimization.