论文部分内容阅读
Based on the ratio of the size of the feasible region of constraints to the size of the feasible region of a constrained optimization problem,we propose a new constraint handling approach to improve the efficiency of heuristic search methods in solving the constrained optimization problems.In the traditional classification of a solution candidate,it is either a feasible or an infeasible solution.To refine this classification,a new concept about the relative feasibility degree of a solution candidate is proposed to represent the amount by which the ’feasibility’ of the solution candidate exceeds that of another candidate.Relative feasibility degree based selection rules are also proposed to enable evolutionary computation techniques to accelerate the search process of reaching a feasible region.In addition,a relative feasibility degree based differential evolution algorithm is derived to solve constraint optimization problems.The proposed approach is tested with nine benchmark problems.Results indicate that our approach is very competitive compared with four existing state-of-the-art techniques,though still sensitive to the intervals of control parameters of the differential evolution.
Based on the ratio of the size of the feasible region of constraints to the size of the feasible region of a constrained optimization problem, we propose a new constraint handling approach to improve the efficiency of heuristic search methods in solving the constrained optimization problems.In the traditional classification of a solution candidate, it is either feasible or an infeasible solution.To refine this classification, a new concept about the relative feasibility degree of a solution candidate is proposed to represent the amount by which the ’feasibility’ of the solution candidate exceeds that of another candidate.Relative feasibility degree based selection rules are also proposed to enable evolutionary computation techniques to accelerate the search process of reaching a feasible region. In addition, a relative feasibility degree based differential evolution algorithm is derived to solve constraint optimization problems. The proposed approach is tested with nine benchmark problems. Results indicate that that approach is very competitive compared with four existing state-of-the-art techniques, though still sensitive to the intervals of control parameters of the differential evolution.