A design withuf reeboard =101%, meanwhile, would be just about infea-sible. Such a quantification would not be possible withoutnormalization.Finally, the relationship between constraints and freevariables ought to be looked at. Fig. 9 and Fig. 10 dis-play scatter diagrams of the normalized constraints ofdraft and freeboard vs. the normalized free variables ˜ x1,x1 being the ferry’s depth. (It should be noted that the or-dinates of Fig. 9 and Fig. 10 were the abscissas of Fig. 7and Fig. 8, respectively.) Of course, all feasible designslie below the linegk(−→ x )σk= 0. Moreover, the general trendis that for higher ˜ x1 the designs’ mean get closer to theconstraints. Moreover, for small ˜ x1 neither the draft northe freeboard constraint is crucial. In Fig. 10 it is foundthat the frontier of the feasible designs bend away fromthe linegk(−→ x )σk= 0 for small ˜ x1. In this region one or sev-eral other constraints are active. If a region of a free vari-able did not yield any feasible designs at all it could aswell be discarded so as to subsequently reduce the searchdomain.Fig. 11 and Fig. 12 represent similar plots for another freevariable, i.e., ˜ x3, x3 being the parallel extent of a chineclose to the design waterline. Again trends are noticeablewhich assist the design team in the constraint manage-ment.ConclusionTangible improvement of the design process can be re-alized if as many constraints as possible are managed di-rectly by the designer. This includes simultaneous screen-ing of constraints, systematic assessment of interdepen-dencies and support for decision taking. The theoreticalbackground of constraint management stems from formaloptimization. Several assessment instruments are pro-posed such as indices, frequency distributions and scatterplots.A hydrodynamic design example was used to illustratethe constraint management approach taken within theconsultancy work of the authors. A Sobol search strat-egy was employed in which many thousand variantshad been automatically generated and evaluated evenbefore time consuming further analyses were launched.The FRIENDSHIP-Modeler served to generate the hullforms and a very comprehensive picture could be drawnabout the design task at hand. Focus was given on themanagement of geometric constraints. Nevertheless, non-geometric constraints can be treated equivalently.Future work will focus on the further analysisand systematic interpretation of the feasible domainand the multitude of data generated when under- taking both (manual) design and (formal) optimiza-tion.ReferencesAbt, C., Bade, S., Birk, L. and Harries, S. (2001): Para-metric Hull Form Design - A Step Towards One WeekShip Design, 8th International Sypmposium on Prac-tical Design of Ships and Other Floating Structures,PRADS 2001Abt, C., Harries, S., Heimann, J. and Winter, H.(2003): From Redesign to Optimal Hull Lines by meansof Parametric Modeling, 2nd International Conferenceon Computer Applications and Information Technol-ogy in the Maritime Industries, HamburgBirk, L. and Harries, S. (Editors) (2003): OPTIMISTIC– Optimization in Marine Design, Mensch & BuchVerlag, Berlin, ISBN ISBN 3-89820-514-2Harries, S. (1998): Parametric design and hydrodynamic 

摘要

船舶系统的设计经常受到许多约束的限制，这些约束涉及到贯穿产品整个使用周期的许多对立方面。上述言论源自自动优化理论，其中约束通过正式统一的方式处理。一种基于参数化设计原则的模拟设计方法被提出用于系统的设计领域探索。这种模拟设计方法允许综合研究约束，监测评估其带来的深远价值。这种约束管理以一典型船体设计工程举例说明。

关键词

CAD；约束；几何模型设计；船体设计；优化组合；船舶设计

0.引言

设计一个相当复杂的物体意着相当多的约束要满足相互对立的目标应该最大化（或最小化）。这样才能设计出具有创造性又有趣的东西，如果很多约束都需考虑在内这也是一个挑战。限制条件和目标通常都有跨领域的特点，同时可能受到流体动力性、构造、生产、运作、经济性等影响。 船舶设计应用中的约束管理英文文献和中文翻译(4):http://www.751com.cn/jixie/lunwen_63336.html