Diversity-guided generalized extremal optimization for transformer design problem

Market competitiveness and pressure for rational use of resources led manufacturers to look for optimized electromagnetic devices. The large number of variables involving the design of this technology, as well as the compromise between performance and efficiency, have provide the opportunity to adopt optimization techniques in their design stage. Recently, a wide variety of both traditional and nontraditional mathematical algorithms have been introduced to solve challenging and practical optimization problems related to electromagnetic equipment. Evolutionary metaheuristics are general-purpose optimization algorithms inspired by natural self-organized systems and by biological evolution. An optimization method, called Extremal Optimization (EO), which is based on a simplified model of natural selection, was developed to show the emergence of self-organized criticality based on Bak-Sneppen model in ecosystems. Recently, a generalized version of EO method, called Generalized Extremal Optimization (GEO), was proposed. This technique can be applied to a broad class of design optimization problems. In order to enhance and to improve the search performance and efficiency of GEO, a novel Diversity-guided GEO (DGEO) metaheuristic algorithm, with focus in global continuous optimization, is proposed and investigated in this paper considering the design stage of an electromagnetic device. The analysis was carried for a singlephase transformer and revealed that DGEO provided considerable better performance in convergence speed and solution quality than standard GEO.