Solution of constrained optimization problems by multi-objective genetic algorithm

Abstract

This paper introduces a method for constrained optimization using a modified multi-objective algorithm. The algorithm treats the constraints as objective functions and handles them using the concept of Pareto dominance. The population members are ranked by two different ways: first ranking is based on objective function value and the second ranking is based on Pareto dominance of the population members. The maintenance of elite lists for both rankings facilitates preservation of potentially superior solutions. A range of problems including non-linear programming and mixed integer non-linear programming has been solved to test the efficacy of the proposed algorithm. The algorithm effectively handles constraints encountered in both small-scale and large-scale optimization problems. The performance of the algorithm compares favourably with existing evolutionary and heuristic approaches.