MO-PaDGAN: Generating Diverse Designs with Multivariate Performance Enhancement

TL;DR

MO-PaDGAN is a novel generative model that enhances diversity and performance in engineering design synthesis, effectively exploring high-performance regions beyond the training data.

Contribution

It introduces a Determinantal Point Processes based loss function to explicitly improve diversity and performance in generative design models.

Findings

Expands design space towards high-performance regions

Generates highly diverse and high-performing designs

Outperforms existing models in design quality

Abstract

Deep generative models have proven useful for automatic design synthesis and design space exploration. However, they face three challenges when applied to engineering design: 1) generated designs lack diversity, 2) it is difficult to explicitly improve all the performance measures of generated designs, and 3) existing models generally do not generate high-performance novel designs, outside the domain of the training data. To address these challenges, we propose MO-PaDGAN, which contains a new Determinantal Point Processes based loss function for probabilistic modeling of diversity and performances. Through a real-world airfoil design example, we demonstrate that MO-PaDGAN expands the existing boundary of the design space towards high-performance regions and generates new designs with high diversity and performances exceeding training data.