HYBRID METHOD OF CHAOTIC GENETIC ALGORITHM AND BOUNDARY SIMULATION FOR CONSTRAINED OPTIMIZATION

Many real-world problems are usually subject to some constraints and are posed as constrained optimization problems. In this paper, we present a hybrid method for constrained optimization problems, which combines the improved boundary simulation method, chaotic initialization method and genetic algorithm (GA). A novel boundary simulation method based on the backward binary search technique is proposed to get the boundary of the feasible region. In order to maintain diversity among initial solutions, a chaotic method is proposed to generate initial population from the boundary of the feasible region. Sonic self-adaptive parameters are designed in crossover and mutation to generate more valid feasible solutions, and a simple repair method is used to update infeasible solutions. The proposed hybrid method is tested on six benchmark functions and three engineering design problems. The results compared with those of some optimization algorithms show the competitive advantage of our algorithm.