Reinforced cuckoo search algorithm-based multimodal optimization

Abstract

This work proposes a Reinforced Cuckoo Search Algorithm (RCSA) for multimodal optimization, which comprises three different strategies: modified selection strategy, Patron-Prophet concept, and self-adaptive strategy. The modified selection strategy has been proposed for efficient selection of next generation individuals instead of choosing a random set of individuals, which is predominantly followed in a standard Cuckoo Search (CS). The Patron-Prophet concept is based on a donor-acceptor concept where a donor donates information and the acceptor makes use of it. In the RCSA, the deviated information of abandoned solutions from selected solutions will be calculated and subsequently used by the newly generated solutions. A self-adaptive step size has been introduced to achieve multimodality in the RCSA. Experimental results using benchmark problems show that the RCSA performs well in terms of multimodality when compared with other existing algorithms found in the literature. This proposed RCSA is also implemented in three different engineering design problems for performance evaluation.