Abstract:  Aiming at the problem of slow convergence speed and insufficient global search ability of whale optimization algorithm (WOA) in solving high-dimensional complex problems, a  whale optimization algorithm based on hybrid reverse learning strategy (MWOA) is presented. Firstly, an adaptive inertial weight is introduced to adjust the step length in the early stage of the optimization and the population diversity in the later stage of the optimization. Secondly, a hybrid reverse learning strategy is proposed and integrated into WOA to improve the convergence accuracy of the algorithm. Finally, a nonlinear parameter attenuation strategy is introduced to improve its exploration and mining ability and convergence speed on high dimensions and complex problems. The optimization effects of MWOA, WOA, MS-WOA and IWOA on 10 benchmark functions are compared, and the result shows that MWOA improves the convergence speed and optimization accuracy compared with the other algorithms. Furthermore, the comparison among MWOA, CODE,CPSO,EGWO and DIHS shows that MWOA has better convergence accuracy.

Key words: whale optimization algorithm, hybrid reverse learning, nonlinear convergence factor, adaptive weight