A Novel Multi-Objective Evolutionary Algorithm Based on Linear Programming

Abstract

It is the goal of scholars in the field of multiobjective optimization to find wide distributive and uniform Pareto solution set over Pareto front. The reason is that the solutions of multi-objective optimization problem is a set of Pareto solutions which are non-dominated each other, and the obtained Pareto solutions are often not well distributed and cannot satisfy the needs of decision makers. It may be the case that decision-makers expect to have a solution in an area to assist them for making decisions. In this paper, we propose a local search strategy based on linear programming and construct a multi-objective evolutionary algorithm based on linear programming (MOEA/LP). MOEA/LP algorithm makes up for the large “gap” in Pareto front, and makes Pareto optimal solutions over Pareto front more uniform and more extensive. Thereby, the decision makers use MOEA/LP algorithm to make more effective choice. Experiment results show the proposed algorithm has better performance according to some measure indices such as running time, hypervolue and C metric, etc.