Reinforced exploitation and exploration grey wolf optimizer for numerical and real-world optimization problems

摘要

Grey Wolf Optimizer (GWO) has been proposed recently. As GWO has superior performance, it has been employed to solve various numerical and engineering issues. However, it easily traps into stagnation when solving complex and multimodal problems. GWO mainly searches around the top three wolves and assigns the same weights to them, deteriorating the convergence and exploration. A reinforced exploitation and exploration GWO (REEGWO) is developed. In the proposed REEGWO algorithm, the top three wolves are given different weights on the basis of their knowledge about the location of the prey. Then, a random search based on the tournament selection is used to enhance the exploration. A well-designed mechanism is developed to balance exploration and exploitation. The experimental results have proved that REEGWO is perfect among GWO and its four recently top variants. Then, the proposed REEGWO is compared with the latest heuristic algorithms and their latest variants. The results have shown that REEGWO is competitive. Four real-world applications are also solved by six algorithms, and results have further validated the efficiency of REEGWO.