Abstract:In this work, we perform a systematic comparison of the effectiveness and efficiency of generative and non-generative models in constructing design spaces for novel and efficient design exploration and shape optimization. We apply these models in the case of airfoil/hydrofoil design and conduct the comparison on the resulting design spaces. A conventional Generative Adversarial Network (GAN) and a state-of-the-art generative model, the Performance-Augmented Diverse Generative Adversarial Network (PaDGAN), are juxtaposed with a linear non-generative model based on the coupling of the Karhunen-Loève Expansion and a physics-informed Shape Signature Vector (SSV-KLE). The comparison demonstrates that, with an appropriate shape encoding and a physics-augmented design space, non-generative models have the potential to cost-effectively generate high-performing valid designs with enhanced coverage of the design space. In this work, both approaches are applied to two large foil profile datasets comprising real-world and artificial designs generated through either a profile-generating parametric model or deep-learning approach. These datasets are further enriched with integral properties of their members' shapes as well as physics-informed parameters. Our results illustrate that the design spaces constructed by the non-generative model outperform the generative model in terms of design validity, generating robust latent spaces with none or significantly fewer invalid designs when compared to generative models. We aspire that these findings will aid the engineering design community in making informed decisions when constructing designs spaces for shape optimization, as we have show that under certain conditions computationally inexpensive approaches can closely match or even outperform state-of-the art generative models.

What problem does this paper attempt to address?

The paper aims to address the issue of comparing the effectiveness and efficiency of generative models versus non-generative models in engineering shape optimization, particularly in the application of aerodynamic/hydrodynamic airfoil/hydrofoil design. The study evaluates the performance of these two types of models in constructing efficient design spaces through comparative analysis. The paper mainly addresses the following issues: 1. **Comparison of Generative and Non-Generative Effectiveness**: The paper systematically compares generative models (such as Generative Adversarial Networks, GAN, and its improved version, Performance-Augmented Diverse Generative Adversarial Network, PaDGAN) with non-generative models (based on Karhunen-Loève expansion combined with physics-informed shape feature vectors, SSV-KLE) in terms of their effectiveness and efficiency in constructing design spaces. 2. **Importance of Shape Encoding and Physical Information**: The study emphasizes the importance of shape encoding methods and the integration of physical information into design descriptions to enhance the performance of non-generative models. This includes using shape feature vectors (SSV) combined with geometric invariant moments to improve the quality of the design space. 3. **Improvement of Design Space Quality**: By using physics-informed non-generative models, it is possible to generate high-performance and efficient designs while reducing the number of invalid designs. This is crucial for improving the overall quality and coverage of the design space. 4. **Dataset Generation and Processing**: The paper details the generation process of two large airfoil datasets (D1 and D2). These datasets not only include actual design samples but also a large number of artificial designs generated through parametric models or deep learning methods to enrich the design space. 5. **Model Performance and Diversity Evaluation**: The paper comprehensively evaluates the generated designs using multiple metrics, including design effectiveness, diversity, and performance. This helps to understand the capabilities of different models in complex design spaces. In summary, the core objective of this study is to empirically demonstrate that under specific conditions, non-generative models can achieve performance comparable to or even better than state-of-the-art generative models at a lower computational cost, especially in terms of design effectiveness and coverage. This finding has significant practical implications for the field of engineering design, helping designers make more informed choices and promoting the development of innovative designs.

Generative VS non-Generative Models in Engineering Shape Optimization

Generative vs. Non-Generative Models in Engineering Shape Optimization

Multiobjective Shape Optimization of A Hypersonic Lifting Body Using A Correlation-Based Transition Model

Aerodynamic Optimization Design of Airfoil Configurations Based on Cascade Feedforward Neural Network

Airfoil GAN: Encoding and Synthesizing Airfoils for Aerodynamic Shape Optimization

Airfoil Design Parameterization and Optimization using Bézier Generative Adversarial Networks

Physics-guided training of GAN to improve accuracy in airfoil design synthesis

Deep Generative Model for Efficient 3D Airfoil Parameterization and Generation

Diffusion Models Beat GANs on Topology Optimization

Deep Generative Design: Integration of Topology Optimization and Generative Models

MO-PaDGAN: Reparameterizing Engineering Designs for augmented multi-objective optimization

3D Design Using Generative Adversarial Networks and Physics-Based Validation

Generative Models for Anomaly Detection and Design-Space Dimensionality Reduction in Shape Optimization

Aerodynamic shape optimization using graph variational autoencoders and genetic algorithms

A generative design technique for exploring shape variations

Deep Learning Based Adaptive Deformation of Aerodynamic Shape for Ducted Propellers

Shape related constraints aware generation of Mechanical Designs through Deep Convolutional GAN

Tailoring Generative Adversarial Networks for Smooth Airfoil Design

A Unified Learning Model for Airfoil Geometry Fitting and Efficient Optimization

Generative Aerodynamic Design with Diffusion Probabilistic Models