A multi-stage time-backtracking grey wolf optimizer introducing a new hierarchy mechanism

Abstract The Grey Wolf Optimizer (GWO) is a swarm intelligence algorithm that mimics natural creatures. It is characterized by fast convergence, few parameters and high accuracy. However, after extensive research, it is found that GWO is easily converged too early in the process of the algorithm, which leads to the reduction of population diversity and the results are not accurate enough, in addition, GWO has a problem that the results tend to converge to 0 easily, which means that the algorithm obtains exceptionally good results on the test function with the optimal point of 0, but not in other cases. To solve the main problems of GWO, this paper proposes to introduce a new hierarchical multi-stage adaptive search grey wolf optimization algorithm called MTBGWO. In this paper, we compare MTBGWO with other excellent algorithms on 23 benchmark standard test sets before and after panning and CEC2013 conference test sets, and the results show that MTBGWO can largely improve the above mentioned problems and improve the performance of the algorithm. To further test the usability of the algorithm on real-world problems, the algorithm is used in this paper on three classical engineering problems and the results show that the algorithm has good applicability and accuracy on real-world problems.