Hybrid advanced player selection strategy based population search for global optimization

Abstract

Motivation: Finding the optimum solution of real world complex optimization problem needs efficient solver system. Meta-heuristic like swarm, evolutionary or physics based methods essentially form an expert system which may strive for the search of optimum solution. However, these methods have their own merits and demerits. To supersede the performance of existing systems in context of real time and advanced problems and to propose new intelligent method having an optimal balance between exploration and exploitation phase have become a center of focus of the expert system design. Therefore, being inspired from such a scenario, a hybrid advanced player selection strategy based population search (HAPS-PS) technique is proposed and evaluated in this paper in context of modern applications.Methods: Sport is a multi-functional activity where different learning methods like cooperative, competitive and self-learning variants along with interactive environmental features help to improve the performance of the players. The properties of playing activity and the thematic learning process of sports, along with some features like, chaotic map based initialization, pool based topological structure selection, integration of blended laplacian operator etc. contributes in the design of HAPS-PS. Also a novel mechanism of team reformation through elimination of comparatively poor players and recruitment of new talents is employed. It provides diversification and regulates the search agents in a promising direction.Results: We have comprehensively evaluated the performance of the HAPS-PS by applying it on standard benchmark problems and on various shifted rotated, hybrid and composite problems of CEC2013 and CEC2014. Also some of the competitive problems of CEC 2017 are tested using the proposed method and they are also compared with other state-of-the-art algorithm to validate the efficacy of the system. Further, tests are carried out on some real-world problems from CEC2011 which includes applications related to parameter optimization of frequency-modulated (FM) sound wave, minimization of molecular potential energy of Lennard-Jones cluster, controlling the chemical reactors, and evaluation of the inter atomic tersoff potentials for covalent systems of silicon. The proposed method is also applied on spread spectrum radar polyphase code design problem as a part of real world application. The comparative results validate the effectiveness and efficiency of HAPS-PS.