Application of A Self-adaption Dual Population Genetic Algorithm in Multi-objective Optimization Problems

Abstract

Multi-objective evolutionary algorithm is a powerful tool in resolving multi-objective optimization problems. This algorithm inherits the advantages of parallel random search, strong global searching capability and the ability to solve highly-complicated non-linear problems of evolutionary algorithm and it is usually used in the optimization problems with multiple mutual conflicts. However, such algorithms are slow in convergence and easy to be trapped in local optimal solution. This paper proposes a multi-objective dual population genetic algorithm (MODPGA) and explores the improvement strategies of multi-objective genetic algorithm. The adoption of self-adaption and dual population strategy can guarantee that the algorithm of this paper can converge to Pareto solution set in a reliable and quick manner and it can perform more extensive search on the objective function space and conduct more samples on multi-objective functions so as to be closer to the approximate optimal solution set of global optimal solutions. This solution set also includes more optimal feasible points and provides reliable basis for the decision making.