Abstract: In order to solve more complex optimization problem, the original algorithm of coyote is analyzed, and a kind of multi-stage coyotes optimization algorithm (MCOA)is proposed in this paper. The algorithm can be divided into before, while and post stages. Firstly, the coyote is allowed to grow normally in the early stage, and then the rotation strategy is introduced to enhance the search range of its exploration process to avoid falling into the local optimal solution prematurely. Secondly, in the middle stage, the strategy of “taking advantage of the advantages and taking advantage of the disadvantages” is proposed, and then the new regeneration method of the head wolf is adopted to update the head wolf, so as to better balance the exploration and exploitation process. Finally, in the late stage, a new type of late growth mode is proposed for coyote growth, and the convergence factor is avoided to solve the problem of late convergence. CEC2014 test function set is used for simulation. Compared with other advanced algorithms with the newly proposed Coyote algorithm, the results show that MCOA has better search efficiency, convergence accuracy, optimization performance and stability.

Key words: coyote optimization algorithm, swarm intelligence algorithm, test function, multi-stage, strategy