Abstract In the reference vector based many-objective evolutionary algorithms, two parents for reproduction are randomly selected in the current population, which probably slows down the speed of convergence. Furthermore, the lack of individuals assigned to some reference vectors will weaken the diversity of population. In order to solve these problems, a new Improved Many-objective Evolutionary Algorithm based on Decomposition (IMaOEA/D) was proposed. Firstly, when a reference vector was assigned at least two individuals in the framework of decomposition strategy, the parent individual was selected for reproduction according to its distance from the ideal point for the individual assigned to the reference vector, so as to speed up the search speed. Then, for the reference vector that could not been assigned at least two solutions, the point with the smallest distance from the ideal point along the reference vector was selected from all individuals, so that at least two individuals of the reference vector were associated with it. Meanwhile, at least one individual was related to it for each reference vector after environmental selection, so as to ensure the diversity of the population. The proposed method has been tested and compared to other four high-dimensional many-objective optimization algorithms based on decomposition on the MaF test problem sets with 10 and 15 objectives. The experimental results show that, the proposed algorithm has the better optimization ability for high-dimensional multi-objective optimization problems. its optimization results of 14 test problems in 30 test problems are better than the other four comparison algorithms. Especially, the proposed algorithm has the better optimization effect on the degradation problem.
Received: 04 December 2020
Published: 16 September 2021