Deep Learning Convective Flow Using Conditional Generative Adversarial Networks

We developed a general deep learning framework, FluidGAN, datasets. In physics and engineering community, deep learning capable of learning and predicting time-dependent has introduced transformative solutions across diverse convective flow coupled with energy transport. However, is thoroughly data-driven with high speed and accuracy and most works are usually task-specific and still rely on satisfies the physics of fluid without any prior knowledge of understanding underlying physical rules. FluidGAN propose a FluidGAN model capable of inferring underlying also learns the coupling between velocity, pressure, and temperature physics and could directly predict stationary and timedependent fields. Our framework helps understand deterministic multi-physical phenomena using certain boundary multiphysics phenomena where the underlying physical conditions and initial conditions with both high accuracy model is complex or unknown.