Abstract
The hunger games search (HGS) algorithm is a recently proposed population-based optimization algorithm that mimics a common phenomenon of animals searching for food due to hunger stimuli and has a simple and easy-to- understand structure. However, the original HGS still suffers from shortcomings, such as low population diversity and the tendency to fall into local optima. To remedy these shortcomings, an improved HGS, called OCBHGS, is proposed, which introduces three main strategies, namely the chaotic initialization strategy, the Gaussian barebone mechanism, and the orthogonal learning strategy. Firstly, chaotic mapping is used for initialization to improve the quality of the initialized population. Secondly, the embedding of the Gaussian barebone mechanism effectively improves the diversity of the population, facilitates the communication between members, and helps the population avoid falling into local optima. Finally, the orthogonal learning strategy can extend the domain exploration and improve the solution accuracy of the algorithm. We conducted extensive experiments in the CEC2014 competition benchmark function, comparing OCBHGS with nine other metaheuristics and 12 improved algorithms. Also, the experimental results were evaluated using Wilcoxon signed-rank tests to analyze the experimental results comprehensively. In addition, OCBHGS was used to solve three constrained real-world engineering problems. The experimental results show that OCBHGS has a significant advantage in convergence speed and accuracy. As a result, OCBHGS ranks first in overall performance compared to other optimizers.