基于分解的进化算法和多变量分析技术在船型参数设计中应用  

Application of Decomposition-Based Evolutionary Algorithm and Multivariate Analysis in Ship Form Parameter Research

作　　者：杨路春[1,2] 杨晨俊[1] 汪志强[2] 李学斌[2] 

机构地区：[1]上海交通大学高新船舶与深海开发装备协同创新中心,海洋工程国家重点实验室,上海200240 [2]武汉第二船舶设计研究所,武汉430000

出　　处：《上海交通大学学报》2016年第8期1200-1206,共7页Journal of Shanghai Jiaotong University

基　　金：“十二五”预研项目(15GFZ-JZ11132)资助

摘　　要：引入优化和决策技术讨论船型参数设计的多目标优化论证,基于分解的进化算法(DBEA)将船型参数设计的多个目标优化问题分解为一定数量的单目标优化子问题,采用进化算法同时求解这些单目标优化子问题.DBEA算法的种群由法向边界相交方法(NBI)构建,子问题的优化通过和邻近个体的进化操作完成.采用熵权和灰色关联方法对DBEA算法得到的船型Pareto解集进行综合评价,给出每个设计方案的定量指标排序.基于多变量分析技术讨论了船舶设计变量的层次聚类属性,给出了设计变量间的类别特性.采用多维标度方法(MDS)给出了这些变量在二维平面里的映射图形,结合聚类树形图可以加深对船舶参数设计模型的认识.对一艘3万t油船进行船型参数设计,算例分析表明,DBEA算法能够快速获得分布均匀的Pareto解,灰色关联方法的决策合理可行.The study of hull parameters of ship is an important design task for ship designers.This task can be treated as a multi-objective optimization project. A decomposition-based evolutionary algorithm（DBEA）was adopted for approximating the Pareto set of multi objective optimization problem.In this way,the multi objective optimization problem was explicitly decomposed into a series of scalar optimization problems.The evolutionary algorithm was used to find solutions to these single-objective optimization problems simultaneously.The gray relational analysis was employed for comprehensive evaluation of Pareto set.The ranking of each solution were obtained after entropy-weights of attributes were provided.The hierarchical clustering of design variables and attribute variables were also investigated.The dendrogram plots were conducted for these variables.The projections of variables on 2-dimensional space were given by using the multidimensional scaling（MDS）skill.The relations of variables were studied together with dendrogram.An example of main dimensions design of tanker with 30 000 DWT was provided for demonstration of the present hybrid method.

关 键 词：船型参数设计 基于分解的进化算法 熵权-灰色关联分析 多变量分析 综合评价 

分 类 号：U661[交通运输工程—船舶及航道工程]