A Meta-VAE for Multi-component Industrial Systems Generation

Fouad Oubari, Raphaël Meunier, Rodrigue Décatoire, Mathilde Mougeot

Abstract: Generative design is an increasingly important tool in the industrial world. It allows the designers and engineers to easily explore vast ranges of design options, providing a cheaper and faster alternative to the trial and failure approaches. Thanks to the flexibility they offer, Deep Generative Models are gaining popularity amongst Generative Design technologies. However, developing and evaluating these models can be challenging. A notable gap in the field is the absence of standardized benchmarks for objectively evaluating and comparing various Deep Generative Models architectures from a manufacturing perspective. Additionally, standard Deep Generative Models seem to struggle to accurately produce multi-component industrial systems governed by latent design constraints. In response to both challenges, our work introduces a two-fold solution. Firstly, we present an industry-inspired use case that integrates real-world characteristics of industrial systems, facilitating rapid generation for benchmarking purposes. Secondly, we introduce a Meta-VAE, designed to generate complex multi-component industrial systems, and demonstrate its efficacy using the proposed use case.