Multi-stage design space reduction technology based on SOM and rough sets, and its application to hull form optimization

Abstract

Hull form design optimization is a highly computationally intensive and complex engineering problem. It has the following characteristics. (1) There are many design parameters that express hull form and the optimization space is large, making it a typical high-dimensional problem. (2) The design performance space is complex and optimization algorithms are difficult to develop. (3) The number of iterations is large and the computational load of computational fluid dynamics (CFD) numerical simulation is huge. These characteristics directly lead to the poor efficiency of hull form optimization technology based on CFD and it is difficult to obtain a global optimal solution. Based on the above analysis, this paper proposes a multi-stage space reduction technique combining the characteristics and advantages of a Self-Organizing Map (SOM) and the rough set theory. Knowledge discovery and data mining was conducted on sample simulation data and the obtained knowledge was used to guide design optimization and locate a subspace worthy of attention, which can significantly improve optimization efficiency. The proposed method was applied to the optimization of the 46,000 DWT oil tanker shape line.