抄録
There are a lot of efficient mathematical programming methods that can solve convex problem even when we have a large number of design variables. However, in the real engineering problem, most of them becomes mixed variable problem, and therefore the objective function we need to treat becomes peaky problem, and in the most case, it is not convex problem. In such a case we need to select some parameters suitable for each methodology to keep its performance. These days there are a lot of new methods especially in heuristic search as GA, EA, SA, PSO, DE and so on. Most of these methods do not sensitivity in searching. Actually, they are very powerful in the peaky problems, but their effectiveness is limited in the relatively small number of design variables. In this paper, we use some bench mark test problem, and show the effectiveness of the conventional mathematical programming method, PSO and the proposed method with respect to the number of design variables, and show the effectiveness of the proposed method.
