Application of Sampling Methods for Constrained Space in Hull Form Optimization

Abstract

Use of approximation models instead of direct application of CFD tools plays a crucial role in hull form optimization to enhance efficiency. The selection of sample points directly impacts the accuracy and cost of approximation models, and the effectiveness of hull form optimization. This paper presents a sampling method based on constrained space. The distribution pattern of the constrained space is initially analyzed, and its boundary is subsequently extracted by the Support Vector Machine (SVM), providing guidance for
the subsequent sample selection. To ensure effective sampling within the constrained space, the maximum minimum distance criterion is employed. The proposed methodology is validated via a case study involving a 13,000DWT inland twin-screw bulk carrier. The Kriging approximation model is constructed to optimize the hull form while adhering to specific constraint conditions, thereby demonstrating the feasibility and efficacy of the proposed approach.