船型优化中样本点选取方法对近似模型精度的影响研究

摘要

将近似技术应用于船体型线优化中，构建便于求解的近似模型，用以代替CFD数值求解，能有效提高船型优化的效率，缩短优化时间。如何构建高精度的近似模型关系到优化结果是否可靠，而样本点的选取方式是影响近似模型精度的重要因素之一。论文首先阐述了拉丁超立方设计的选点原理。其次，针对均匀设计存在的难点，提出了大试验次数均匀设计表的构造方法。然后，以数学测试函数为例，论述了两种选点方法的适用性。最后，以国际标模KCS为例，建立了船舶阻力性能的近似模型，并进行了精度检验。%An approximation technology is applied to hull form optimization for constructing approximate model that facilitates to computation and can be used to replace CFD tools. This technology can significantly improve the efficiency of hull form optimization and shorten the time period of optimization. High precision of the approximation model determines a reliable optimization result, while selection of sample points is one of the important factors affecting the accuracy of the model. In this paper, firstly the principle of Latin hypercube sampling is expounded. Secondly, the method of uniform design for large sample points is proposed. Then, the applicability of these two methods is discussed based on mathematical function. Finally, an approximate model of resistance performance for the international standard ship model is established, and the accuracy of this approximate model is examined.