Research On Key Technologies In Ship Parametric Automatic Optimization Design Based On SBD

With the development of computer technology and modern numerical simulation analysis methods,simulation-based design(SBD)has been proposed in the process of ship design.The main idea of SBD is that performances of a ship are evaluated by numerical simulation analysis methods during the optimization.Then,an optimal design of a ship that exhibits the best performace is obtained through an optimization technique.SBD integrates the advanced numerical simulation analysis codes,the optimization algorithms and the geometry modification mehods,which provides a new way for ship optimization design and attracts worldwide attention.From the perspective of optimization process,the essential techniques used in SBD based ship parametric optimization design includes geometry modification and reconstruction,optimization algorithms,numerical simulation analysis,metamodeling techniques of builing an approximation for numerical simulation alaysis,etc.This thesis focus on improvements of those key techniques,aiming at the problms to be solved in the existing key technology research and improving the efficiency of optimization.After a state of the art introduction of the key techniques mentioned aboved is provided,some of the key techniques are improved in this thesis.The main researches of this thesis are summarized as follows.(1)In terms of parametric expression and modeling,the parameterizations of hull shape and ship structure are studied.According to the characteristics of SWATH,two kinds of full parameterization methods of hull shape for SWATH are established.The parameterization of hull structure is discussed.In the aspect of grid reconstruction of numerical simulation analysis model,this thesis explores a local mesh reconstruction approach to automatic supplying grid models for optimization of ship structure.The shape and topology information of the finite elements are handled and controlled effectively by generating the dynamic node based data structure of the initial structure.The modeling template is created to guarantee the quality of the reconstructed elements.This method is suitable for the mesh reconstruction of ship finite element analysis,and provides an efficient and convenient technical means for the automatic reconstruction of mesh model.(2)For simplification technology of numerical calculation,the surrogate modeling technology is mainly studied in this thesis.One of the key steps to create a surrogate model is sample collection.An improved sequential sampling approach for surrogate modeling to surrogate computationally expensive simulations is proposed.Starting from an initial training set,a series of new samples are sequentially selected in important Voronoi cells based on the Voronoi diagram.The sampling results have shown the effectiveness of the proposed sampling method and the advantage of the surrogate model that achieve the required accuracy with relatively 40%less samples,which effectively reduces the computational time necessary for numerical simulation analysis and improves the efficiency of the optimization.(3)In terms of optimization methods and comprehensive integration,a simulation-driven automatic optimization design integration framework is constructed.The dynamic process intergrations of codes involved in ship optimization design are realized.Data transfer problems among different modules are also solved.Four ship engineering optimization examples are given to verify the feasibility and universality of the proposed integrated framework.The optimization results have shown the effectiveness of the proposed methods.