An improved sardine feast metaheuristic optimization based on Lévy flight

Abstract

The recently proposed Sardine Feast Metaheuristic Optimization (SFMO) is a population-based metaheuristic optimization algorithm inspired by the commensal behavior of various predators while preying on sardines at sea, which is commonly known as a sardine feast. This algorithm is a behavior imitation of dolphins and sea birds (blue-footed boobies and brown pelican) preying on schools of sardines. SFMO suffers from premature convergence since it relies on the normal random function to calculate predators' movement or step size during exploration and exploitation. The proposed improved SFMO (SFMO-Lévy) aims to enhance the ability of predators to explore divergent areas using Lévy flight in the step size calculation. The performance of the SFMO-Lévy is investigated using several predefined benchmark functions for global optimization problems. The outcomes of the tests are then compared with those generated by the standard SFMO algorithm. The SFMO-Lévy outperforms the SFMO by providing an average of 23.44% fewer function evaluations. The results reveal that the proposed algorithm can solve the benchmark functions better than the standard SFMO algorithm.