Self-Adaptive Genetic Algorithm for Constrained Optimization

Abstract

This study proposes a self-adaptive penalty function algorithm for solving constrained optimization problems using genetic algorithm (GA). Constrained optimization is a practically relevant and challenging field that deals with optimization of real world problems that involve complex constraints that make them difficult to tackle. GA is a stochastic search method based on the evolutionary ideas of natural selection and genetic. In GA candidate solutions to a certain problem, called individuals, will evolve from generation to generation toward finding better solutions. In this research GA based constraint handling algorithm is proposed that combines the merits of previously designed algorithms. In the proposed method a new fitness value, called distance value, and two penalties are applied to infeasible individuals that violate the constraints. The algorithm aims to encourage infeasible individuals with better objective function value and low constraint violation. The number of feasible individuals in the population is used to guide the search process either toward finding the optimum solution or toward finding more feasible solutions.The performance of the algorithm is tested on 22 benchmark functions in the literature. The results show that the approach is able to find very good solutions comparable to other state-of-the-art designs. Furthermore it is able to find feasible solutions in every run for all of the benchmark functions.