Abstract
ln this study, an efficient global optill1ization method called the “hybrid optimization method” is proposed. This method cOll1bines a conventional genetic algorithm (GA) as a global optimizer and the method of feasible directions (MFD) as a local optimizer. The gradient information (or design sensitivity) obtained during global iterations by the GA is used for an efficient crossover. After the final global iteration by the GA, local optimization performed by the MFD may be used to further improve upon the final global iteration solution. A comparative test of the proposed hybrid optimization method is performed to evaluate its efficiency and accuracy. Finally, the proposed metnod is applied to a structural design optimization problem and ship design optill1ization problems. These applications showed that the proposed method is very efficient, robust, and reliable.