文档介绍：IEEE TRANSACTIONS ON PUTATION, VOL. 8, NO. 4, AUGUST 2004 365
Hybrid Taguchi-ic Algorithm for Global
Numerical Optimization
Jinn-Tsong Tsai, Tung-Kuan Liu, and Jyh-Horng Chou, Senior Member, IEEE
Abstract—In this paper, a hybrid Taguchi-ic algorithm to solve global optimization problems with 30 or more dimen-
(HTGA) is proposed to solve global numerical optimization sions, where the improvements in the GA are to seek the op-
problems with continuous variables. The bines the timal breeding conditions (process of forming new trial chromo-
traditional ic algorithm (TGA), which has a powerful global
exploration capability, with the Taguchi method, which can exploit somes at each epoch). Here, it should be noted that, among these
the optimum offspring. The Taguchi method is inserted between proposed improved-GAs, it can be seen that, for 15 benchmark
crossover and mutation operations of a TGA. Then, the systematic problems of global optimization with 30 or 100 dimensions and
reasoning ability of the Taguchi method is incorporated in the very large numbers of local minima, the algorithm presented by
crossover operations to select the better genes to achieve crossover, Leung and Wang [12], which is named the orthogonal ic
and consequently, enhance the ic algorithm. Therefore,
the HTGA can be more robust, statistically sound, and quickly algorithm with quantization (OGA/Q), not only can find optimal
convergent. The proposed HTGA is effectively applied to solve or close-to-optimal solutions but also can give more robust and
15 benchmark problems of global optimization with 30 or 100 significantly better results than those other improved-GAs pre-
dimensions and very large numbers of local minima. - sented by Renders and Bersini [17], Michalewicz [14], Gen and
putational experiments show that the proposed HTGA not only Cheng [7], Yen and Lee [22], Chellapilla [3], Wang [20], An-
can find optimal or close-to-optimal solutions but also can obtain
both better