Multi-objective Wolf Pack Algorithm Based on Random Scouting and Hierarchical Learning

Since the wolf pack algorithm has the advantages of fast convergence and good robustness in solving single-objective problems, we propose a multi-objective wolf pack algorithm based on random scouting and hierarchical learning (MOWPA-RH) for solving multi-objective optimization problems by combining the excellent hunting habits of wolves. Firstly, a random scouting strategy is used to help the population spread its search range and enhance the global search ability of the people. Secondly, the population is stratified by non-dominated sorting. The individuals in the first layer carry out the Levy-Flight strategy to enhance the ability of the algorithm to jump out of the local optimum, and the individuals in each layer except the first layer are guided by the dominant individuals in the former layer, which is conducive to searching for higher-quality solutions. Finally, the current population is merged with the previous generation, and then the population update is accomplished through the screening mechanism so that the algorithm has good convergence and distribution. Comparing MOWPA-RH with five multi-objective optimization algorithms on 12 different benchmarking problems, the experimental results validate the effectiveness of MOWPA-RH.