BF-GAN: Development of an AI-driven Bubbly Flow Image Generation Model Using Generative Adversarial Networks

2025-02-08

AI Summary

High-quality bubble flow image acquisition and annotation are costly and labor-intensive in two-phase flow research. To address this, we propose BF-GAN, a physics-driven generative adversarial network. Trained on 140,000 experimentally acquired images—each labeled with gas and liquid superficial velocities (j₉, j_f)—BF-GAN integrates physics-informed conditional embedding and a multi-scale physics-aware loss function, incorporating both flow regime parameter mismatch penalties and pixel-level constraints. It is the first model to generate bubble flow images whose key regime characteristics—including bubble size and spatial density—match measured data and empirical correlations with <8% error. Compared to conventional GANs, BF-GAN enables high-fidelity, controllable synthesis across arbitrary (j₉, j_f) combinations, improving PSNR by 3.2 dB and SSIM by 0.07. The generated dataset is publicly released, serving as a reliable benchmark for bubble detection and segmentation, significantly enhancing training efficiency and model generalizability.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Natural Language Processing: Generation
• Machine Learning: Multimodal Learning

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Web Mining and Content Analysis: Web data generation and simulation
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

A generative AI architecture called bubbly flow generative adversarial networks (BF-GAN) is developed, designed to generate realistic and high-quality bubbly flow images through physically conditioned inputs, jg and jf. Initially, 52 sets of bubbly flow experiments under varying conditions are conducted to collect 140,000 bubbly flow images with physical labels of jg and jf for training data. A multi-scale loss function is then developed, incorporating mismatch loss and pixel loss to enhance the generative performance of BF-GAN further. Regarding evaluative metrics of generative AI, the BF-GAN has surpassed conventional GAN. Physically, key parameters of bubbly flow generated by BF-GAN are extracted and compared with measurement values and empirical correlations, validating BF-GAN's generative performance. The comparative analysis demonstrate that the BF-GAN can generate realistic and high-quality bubbly flow images with any given jg and jf within the research scope. BF-GAN offers a generative AI solution for two-phase flow research, substantially lowering the time and cost required to obtain high-quality data. In addition, it can function as a benchmark dataset generator for bubbly flow detection and segmentation algorithms, enhancing overall productivity in this research domain. The BF-GAN model is available online (https://github.com/zhouzhouwen/BF-GAN).

Problem

• Develops BF-GAN for bubbly flow image generation
• Enhances image quality with multi-scale loss function
• Reduces time and cost in two-phase flow research

Innovation

• Generative Adversarial Networks
• Physically conditioned inputs
• Multi-scale loss function