An Improvement Evolutionary Algorithm Based on Decomposition and Grid-based Pareto Dominance for Many-objective Optimization

In this paper, an better evolutionary algorithm based on decomposition and grid-based Pareto dominance (MOEA/DG) is proposed to work out many-objective optimization problems. The main goal is to heighten the convergence and diversity by generating good offspring with a good selection strategy. To be specific, a selection method based on decomposition and grid-based Pareto dominance is given to equilibrate exploration and exploitation. Moreover, the non-dominated solutions are kept on file external and a new diversity strategy is used to maintain diversity. The MOEA/DG compares with several advanced and excellent algorithms on many-objective benchmark functions, the experimental data verify that the algorithm put forward has good effects on most issues.