DMPA-HHO: A Hybrid MPA-HHO optimizer with Dynamic Opposite Learning
Shoubao Su, Liukai Xu, Chishe Wang, Chao He · Research Square · 2022
Abstract A new hybrid algorithm is proposed by incorporating Harris Hawks Optimization with Marine Predators algorithm and dynamic Opposition-based learning, namely DMPA-HHO. In the algorithm, the problem is addressed that Harris Hawks Optimization (HHO) tends to fall into local optima and low accuracy of the solution. Dynamic Opposite Learning (DOL) improves the swarm diversity and swarm quality, and enhances the global search capability and search accuracy. HHO and the Marine Predators Algorithm (MPA) are blended to enhance the progressive rapid dives of the Harris hawk flock, effectively improving the algorithm's exploitation capabilities. DMPA-HHO uses the FADs’ effect of the MPA to increase the possibility of individuals escaping from the local optimum solution when the search falls into the local optimal solution. Compared with others on several benchmark functions, the DMPA-HHO algorithm has a better search accuracy and a stronger ability to avoid trapping in local optima.