New operators for multicriteria optimization using the modified distance method

Distance method is a very efficient method to solve multicriteria optimization problems using genetic algorithms. This article propose two modifications to distance method. The first modification is the choice of a selection operator that gives better results than the classical roulette wheel when used with distance method. The second one is the introduction of an original elistist strategy, called Pareto elitism, specially suited for multicriteria optimization. The two modifications have been tested on a numerical example and have shown to improve both the quality of the solutions generated and the stability of the results from one run of the algorithm to another.