HS 최적화 알고리즘 매개변수 결정을 위한 해석적 방법

Abstract

Harmony search (HS) is a relatively recently developed meta-heuristic optimization method imitating the music improvisation process. The algorithm parameter, HMCR, play a very important role in HS. The value of HMCR must be determined differently depending on the dimensions of a problem and the number of optimum values. But there is no analytic approach to determine the parameter value and the rule of thump is applied. To solve the problem, the relation between HMCR and new harmony generation is studied. As a result, new harmonies are classified into two parts of ‘diversification’ and ‘intensification’ and the generation probabilities of the two group harmonies are analysed. In this paper, HMCR, unlike the conventional method, is determined based on the DIVR which is the probability of diversification harmony. By applying new method, the value of HMCR can be customized for a given optimization problem. The proposed method is applied to the test optimization functions and the results shows that the proposed method works very well.