Generative Design of Ship Propellers using Conditional Flow Matching

TL;DR

This paper introduces a generative AI approach using conditional flow matching to create diverse ship propeller designs that meet specific performance targets, supported by simulation data and data augmentation techniques.

Contribution

It presents a novel application of conditional flow matching for bidirectional design-performance mapping in ship propeller generation, with data augmentation strategies to improve model accuracy.

Findings

Generated multiple valid designs for the same performance targets.

Data augmentation with pseudo-labels enhances model performance.

Examples show diverse geometries with similar performance characteristics.

Abstract

In this paper, we explore the use of generative artificial intelligence (GenAI) for ship propeller design. While traditional forward machine learning models predict the performance of mechanical components based on given design parameters, GenAI models aim to generate designs that achieve specified performance targets. In particular, we employ conditional flow matching to establish a bidirectional mapping between design parameters and simulated noise that is conditioned on performance labels. This approach enables the generation of multiple valid designs corresponding to the same performance targets by sampling over the noise vector. To support model training, we generate data using a vortex lattice method for numerical simulation and analyze the trade-off between model accuracy and the amount of available data. We further propose data augmentation using pseudo-labels derived from less data-intensive forward surrogate models, which can often improve overall model performance. Finally, we present examples of distinct propeller geometries that exhibit nearly identical performance characteristics, illustrating the versatility and potential of GenAI in engineering design.