Multistrategy boosted multicolony whale virtual parallel optimization approaches

Abstract

A multistrategy boosted multicolony whale optimization algorithm (MSMCWOA) is proposed. First, humpback whales are divided into different subcolonies based on various behavioral mechanisms of whale predation. Each subcolony evolves independently to ensure the diversity of the algorithm population. At the same time, local knowledge or experience generated from each subcolony is periodically provided to the global storage space. When the diversity of colonies decreases sharply, a hierarchical decision-making model based on information sharing is stimulated. Global storage space applies optimal information to guide subcolonies to jump out of potential local optimization. In addition, individuals in different subcolonies choose different strategies to communicate with each other with certain probability. The ability of knowledge acquisition and dissemination and the distance between individuals are considered to improve the effectiveness of information exchange and space search. Twenty-three unimodal and multimodal benchmark functions are tested and compared with state-of-the-art algorithms, and the experimental results show that the MSMCWOA has a better performance in terms of convergence rate and stability. The MSMCWOA is also successfully applied to the optimization design of the pressure vessel and tension/compression spring, and the result shows the superior performance of the MSMCWOA in solving project optimization problems.