Coyote Optimization Algorithm with Linear Convergence for Global Numerical Optimization

In particular, the popularity of computational intelligence has accelerated the study of optimization. Coyote Optimization Algorithm (COA) is a new meta heuristic optimization. It is pays attention to the social structure and experience exchange of coyotes. In this paper, the coyote optimization algorithm with linear convergence (COALC) is proposed. In order to explore a huge search space in the pre-optimization stage and to avoid premature convergence, the convergence factor is also involved. Thus, the COALC will explore a huge search space in the early optimization stage to avoid premature convergence. Also, the small area is adopted in the later optimization stage to effectively refine the final solution, while simulating a coyote killed by a hunter in the environment. It can avoid the influence of bad solutions. In experiments, ten IEEE CEC2019 test functions is adopted. The results show that the proposed method has rapid convergence, and a better solution can be obtained in a limited time, so it has advantages compared with other related methods.