船舶操纵性优化的约束多目标进化算法  

A constrained multi-objective evolutionary algorithm for ship maneuverability optimization

作　　者：刘冰洁 毕晓君 LIU Bingjie;BI Xiaojun(College of Information and Communication Engineering, Harbin Engineering University, Harbin 150001, China;Department of Information Engineering, Minzu University of China, Beijing 100081, China)

机构地区：[1]哈尔滨工程大学信息与通信工程学院,黑龙江哈尔滨150001 [2]中央民族大学信息工程学院,北京100081

出　　处：《哈尔滨工程大学学报》2020年第9期1391-1397,共7页Journal of Harbin Engineering University

基　　金：国家自然科学基金项目(51779050).

摘　　要：针对现有船舶操纵性优化算法收敛性不高的问题,将基于分解的多目标进化算法应用到船舶设计中,本文提出一种船舶操纵性设计的约束多目标进化算法。建立了以直线稳定性和相对回转直径为目标的优化模型,采用基于分解的多目标进化算法框架,结合优秀不可行解改进了差分算子;其次,充分利用优秀不可行解,设计了新的个体选择准则。将本文算法与另外3种船舶操纵性优化算法进行对比,该算法可以提供更多的设计方案,且设计方案收敛性更好。Maneuvering is an operation during which a vessel enters or exits coastal waters of a country,crosses several ships on the way,and proceeds toward or departs from a berth or jetty of a port.The existing algorithms are not able to completely solve the problem of low-level convergence zone(air)that affects the ship maneuverability.We propose a multi-objective evolutionary algorithm based on decomposition for the ship design,and a constrained multi-objective evolutionary algorithm for ship maneuverability optimization in this paper to solve the problem.Firstly,an optimization model is established in this paper,which selects straight line stability and relative gyration diameter as objects,and,multi-objective evolutionary algorithm based on decomposition(MOEA/D)is utilized as the framework,then,optimize the difference operator by dealing with the exceptional infeasible solutions.Second,a new individual selection criterion is obtained by taking full advantage of the exceptional infeasible solutions.Finally,this algorithm is compared with three other design algorithms of ship maneuverability.The compared results show that the proposed method provides more design solutions with better convergence.

关 键 词：船舶操纵性 船型参数 船舶主尺度 约束多目标优化 基于分解的多目标进化算法 差分进化 不可行解 ε约束 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]