Recently, extensive studies have focused on analyzing aerodynamic performance due to its important impact on aircraft design. Most of these works compute the aerodynamic coefficient of the airfoil through computational fluid dynamics (CFD) simulation, which is too time-consuming. To reduce the computational time required, some intelligence-based methods have been presented. However, these methods also suffer from certain issues. First, most of them directly implement existing machine learning methods used to predict the aerodynamic coefficient without adding any improvements. Second, some methods convert the airfoil shape and aerodynamic curves into images, which may lead to curve distortion and the introduction of noise. Third, some methods learn the relationship between the airfoil shape and aerodynamic coefficients but ignore the influence of initial inflow conditions. Accordingly, to address these issues, we propose an intelligent method for predicting the pressure coefficients (Cp) of airfoil based on a conditional generative adversarial network (cGAN). More specifically, we first present a two-step data augmentation strategy designed to expand the original airfoil dataset. Subsequently, we design a novel cGAN-based neural network to predict the Cp curve. To the best of our knowledge, this is the first work to apply generative adversarial network (GAN) to aerodynamic coefficient prediction. Moreover, we design a new loss function to train our network. Extensive experimental results demonstrate that the Cp curve predicted by our method is very close to that generated via CFD simulation. More importantly, our method achieves a speedup close to 1000x compared with CFD simulation.

Due to the complex topological relationship, poor data locality, and data racing problems in unstructured CFD computing, how to parallelize the finite volume method algorithms in shared memory to efficiently explore the hardware capabilities of many-core GPUs has become a significant challenge. Based on a production-level unstructured CFD software, three shared memory parallel programming strategies, atomic operation, colouring, and reduction were designed and implemented by deeply analysing its computing behaviour and memory access mode. Several data locality optimization methods—grid reordering, loop fusion, and multi-level memory access—were proposed. Aimed at the sequential attribute of LU-SGS solution, two methods based on cell colouring and hyperplane were implemented. All the parallel methods and optimization techniques implemented were comprehensively analysed and evaluated by the three-dimensional grid of the M6 wing and CHN-T1 aeroplane. The results show that using the Cuthill–McKee grid renumbering and loop fusion optimization techniques can improve memory access performance by 10%. The proposed reduction strategy, combined with multi-level memory access optimization, has a significant acceleration effect, speeding up the hot spot subroutine with data races three times. Compared with the serial CPU version, the overall speed-up of the GPU codes can reach 127. Compared with the parallel CPU version, the overall speed-up of the GPU codes can achieve more than thirty times the result in the same Message Passing Interface (MPI) ranks.

Vortex detection plays a fundamental role in turbulence research and engineering problems. However, due to the lack of a mathematically rigorous vortex definition, as well as the absence of any vortex-oriented database, both traditional and machine learning detection methods achieve only limited performance. In this paper, we develop a deep learning model for vortex detection using a weak supervision approach. In order to avoid the need for a vast amount of manual labeling work, we employ an automatic clustering approach to encode vortex-like behavior as the basis for programmatically generating large-scale, highly reliable training labels. Moreover, to speed up the clustering method, a multi-view U-Net (MVU-Net) model is proposed to approximate the clustering results using the knowledge distillation technique. A multi-view learning strategy is further applied to integrate the information across multiple variables. In addition, we propose a physics-informed loss function, which enables our model to explicitly consider the characteristics of flow fields. The results on eight real-world scientific simulation applications show that the proposed MVU-Net model significantly outperforms other state-of-the-art methods on both efficiency and accuracy.