Convolutional Neural Networks for Steady Flow Approximation
Xiaoxiao Guo*, University of Michigan; Wei Li, Autodesk Research; Francesco Iorio,
In aerodynamics related design, analysis and optimization problems, ﬂow ﬁelds are simulated using computational ﬂuid dynamics (CFD) solvers. However, CFD simulation is usually a computationally expensive, memory demanding and time consuming iterative process. These drawbacks of CFD limit opportunities for design space exploration and forbid interactive design. We propose a general and ﬂexible approximation model for real-time prediction of non-uniform steady laminar ﬂow in a 2D or 3D domain based on convolutional neural networks (CNNs). We explored alternatives for the geometry representation and the network architecture of CNNs. We show that convolutional neural networks can estimate the velocity ﬁeld two orders of magnitude faster than a GPU-accelerated CFD solver and four orders of magnitude faster than a CPU-based CFD solver at a cost of a low error rate. This approach can provide immediate feedback for real-time design iterations at the early stage of design. Compared with existing approximation models in the aero-dynamics domain, CNNs enable an eﬃcient estimation for the entire velocity ﬁeld. Furthermore, designers and engineers can directly apply the CNN approximation model in their design space exploration algorithms without training extra lower-dimensional surrogate models.
Filed under: Deep Learning