Variable decomposition approach applied to multi-objective optimization for minimum powering of commercial ships
Variable decomposition approach applied to multi-objective optimization for minimum powering of commercial ships
复制标题
DOI：
10.1007/s00773-018-0551-5
复制
发表时间：
2018-04
影响因子：
2.6
通讯作者：
Y. Tahara;Yasuo Ichinose;Azumi Kaneko;Y. Kasahara
中科院分区：
文献类型：
--
作者：
Y. Tahara;Yasuo Ichinose;Azumi Kaneko;Y. Kasahara
随着计算流体动力学的成熟，它被广泛接受为船舶船体造型设计的关键工具，基于仿真的设计（SBD）的发展在过去的几十年中一直受到强烈的推动。虽然SBD的许多成功的示范，大多数情况下只是处理与单目标优化问题的制定总阻力最小化。一旦关注的是最小化船舶尺度传递功率或有效功率，与模拟的准确性相关的问题在许多情况下显得至关重要，这对船体形状设计者产生不令人信服的结果。我们在本文中提出的方法旨在克服这些问题。本文首先引入变量分解方法将目标船性能函数分解为包含嵌入参数的项，然后建立多目标优化问题（MOOP），而不是仅仅依靠仿真得到的预测功率来求解单目标优化问题。任何解决MOOP的方案都可以应用。在下文中，给出了本方法的概述，并通过与可用的实验流体动力学数据的比较以及对流动和积分参数的详细分析来呈现和讨论结果。本文还讨论了该方法的有效性。
As computational fluid dynamics has matured to the point where it is widely accepted as a key tool for ship hull form design, development of simulation-based design (SBD) has been strongly motivated in the past decades. Although many successful demonstrations of SBD were presented, most cases just deal with minimization of total resistance with a formulation of single-objective optimization problem. Once the interest is in minimization of ship-scale delivered power or effective power, issue related to accuracy of the simulation appears critical in many cases, which yield unconvincing results to hull form designers. The method we propose in this paper aims at overcoming the issues. Instead of just counting on predicted power from the simulation and solve a single-objective optimization problem, we first introduce variable decomposition approach to decompose a target ship performance function into terms including embedded parameters, then formulate and solve a multi-objective optimization problem (MOOP). Any scheme to solve MOOP can be applied. In the following, an overview of the present approach is given and results are presented and discussed through comparison with available experimental fluid dynamics data and detailed analysis of flow and integral parameters. The effectiveness of the present approach is also discussed.