FluidZero: Mastering Diverse Tasks in Fluid Systems through a Single Generative Model

Abstract Fluid-Structure Interaction (FSI) is one of the important cornerstones of science and engineering, such as spacecraft, submersibles, and biomedicine, which are important to understand and optimize, involving four key tasks throughout history: prediction, parameter identification, design, and control. Although each task has made significant strides individually, current approaches remain fragmented; existing models are limited to their specific domains and lack the capability to generalize across different tasks. To overcome this issue while utilizing the correlation between tasks to improve the performance on each task, we propose FluidZero , a unified deep generative model that unifies four different tasks into a joint probabilistic distribution modeling task. The key advantage of FluidZero is that it facilitates cross-modal and cross-task interactions, enabling effective learning of causality laws between physical variables, and then enhancing performance on diverse tasks in the Fluid-Structure Interaction (FSI) system. We evaluate FluidZero in a FSI system, including simulation data and real-world measured data obtained through Particle Image Velocimetry (PIV). Moreover, the designed foil is directly transferred to real-world experiments through 3D printing. FluidZero shows better generalization capabilities, achieving superior performance even in Out-Of-Bound (OOB) situations and real-world applications.