为了降低船舶多学科设计优化(MDO)模型的耦合度,从而间接提高优化效率,以一种基于联系信息流量的聚类算法为基础,提出了一种重组模型的建模技术。对聚类算法的基本假设和计算方法进行了改进,用指数函数取代了线性函数,并完善了编码。提出了针对聚类方案的评价指标—联系权重在聚类内部的百分比和BUS聚类规模,用于比较改进效果。最后以5400箱集装箱船方案设计初期的概念建模为例说明了基于改进聚类算法的建模技术,验证了该技术用于船舶建模的可行性。
In order to reduce the coupling degree of the MDO ( Multidisciplinary Design Optimization) model of ships, and thus to indirectly improve the optimization efficiency, a remodeling method for modeling is proposed based on a relation information flow-based clustering algorithm. In this algorithm, an exponential function is used as the substitution of the linear function to improve the theoretical hypothesis and the calculation method, and the co- ding of the algorithm is also modified. Moreover, two assessment indexes, namely the proportion of relation weight in the interior of the clustering and the BUS clustering scale, are proposed for comparison. Finally, a conceptual model of a 5400TEU container ship in the early conceptual design is used to illustrate the proposed modeling method.The results verify the feasibility of the method in ship modeling.