MULTIOBJECTIVE DESIGN OF A MARINE VEHICLE WITH AID OF DESIGN KNOWLEDGE

To generate the Pareto optimal set efficiently in multiobjective optim ization, a hybrid optimizer is developed by coupling the genetic algor ithm and the direct search method. This method determines a candidate region around the global optimum point by using the genetic algorithm, then searches the global optimum point by the direct search method co ncentrating in this region, thus reducing calculation time and increas ing search efficiency. Although the hybrid optimizer provides cost-eff ectiveness, the design optimization process involves a number of tasks which require human expertise and experience. Therefore, methods of o ptimization and associated programs have been used mostly by experts i n the real design world. Hence, this hybrid optimizer incorporates a k nowledge-based system with heuristic and analytic knowledge, thereby n arrowing the feasible space of the objective function. Some domain kno wledge is retrieved from database and design experts. The obtained kno wledge is stored in the knowledge base. The results of this paper, thr ough application to marine vehicle design with multiobjective optimiza tion, show that the hybrid optimizer with aid of design knowledge can be a useful tool for multiobjective optimum design. (C) 1997 by John W iley & Sons, Ltd.