An improved adaptive NSGA-II with multi-population algorithm

The NSGA-II algorithm uses a single population single crossover operator, which limits the search performance of the algorithm to a certain extent. This paper presents an improved version of the NSGA-II algorithm, named adaptive multi-population NSGA-II (AMP-NSGA-II) that divides the original population into multiple populations and assigns a different crossover operator to each subspecies. It introduces an excellent set of solutions (EXS), which can make the individuals in the EXS set close to the Pareto front and improve the convergence performance of the algorithm. And based on the analysis of the EXS set, the size of each subpopulation can be dynamically adjusted, which can improve the adaptability for different problems. Finally, the computation results on benchmark multi-objective problems show that the proposed AMP-NSGA-II algorithm is effective and is competitive to some state-of-the-art multi-objective evolutionary algorithms in the literatureis.