arXiv subjects

Song Fu

Publications and source records attributed to Song Fu.

29 records · Page 2Linked to original sources

arXiv2020

A two-stage fourth-order gas-kinetic CPR method for the Navier-Stokes equations on triangular meshes

A highly efficient gas-kinetic scheme with fourth-order accuracy in both space and time is developed for the Navier-Stokes equations on triangular meshes. The scheme combines an efficient correction procedure via reconstruction (CPR) framework with a robust gas-kinetic flux formula, which computes both the flux and its time-derivative. The availability of the flux time-derivative makes it straightforward to adopt an efficient two-stage temporal discretization to achieve fourth-order time accuracy. In addition, through the gas-kinetic evolution model, the inviscid and viscous fluxes are coupled and computed uniformly without any separate treatment for the viscous fluxes. As a result, the current scheme is more efficient than traditional explicit CPR methods with a separate treatment for viscous fluxes, and a fourth order Runge-Kutta approach. Furthermore, a robust and accurate subcell finite volume (SCFV) limiting procedure is extended to the CPR framework for troubled cells, resulting in subcell resolution of flow discontinuities. Numerical tests demonstrate the high accuracy, efficiency and robustness of the current scheme in a wide range of inviscid and viscous flow problems from subsonic to supersonic speeds.

math.NA↗

arXiv2020

Feature selection and processing of turbulence modeling based on an artificial neural network

Data-driven turbulence modeling has been considered an effective method for improving the prediction accuracy of Reynolds-averaged Navier-Stokes equations. Related studies aimed to solve the discrepancy of traditional turbulence modeling by acquiring specific patterns from high-fidelity data through machine learning methods, such as artificial neural networks. The present study focuses on the unsmoothness and prediction error problems from the aspect of feature selection and processing. The selection criteria for the input features are summarized, and an effective input set is constructed. The effect of the computation grid on the smoothness is studied. A modified feature decomposition method for the spatial orientation feature of the Reynolds stress is proposed. The improved machine learning framework is then applied to the periodic hill database with notably varying geometries. The results of the modified method show significant enhancement in the prediction accuracy and smoothness, including the shape and size of separation areas and the friction and pressure distributions on the wall, which confirms the validity of the approach.

physics.flu-dyn↗

arXiv2020

CoFF: Cooperative Spatial Feature Fusion for 3D Object Detection on Autonomous Vehicles

To reduce the amount of transmitted data, feature map based fusion is recently proposed as a practical solution to cooperative 3D object detection by autonomous vehicles. The precision of object detection, however, may require significant improvement, especially for objects that are far away or occluded. To address this critical issue for the safety of autonomous vehicles and human beings, we propose a cooperative spatial feature fusion (CoFF) method for autonomous vehicles to effectively fuse feature maps for achieving a higher 3D object detection performance. Specially, CoFF differentiates weights among feature maps for a more guided fusion, based on how much new semantic information is provided by the received feature maps. It also enhances the inconspicuous features corresponding to far/occluded objects to improve their detection precision. Experimental results show that CoFF achieves a significant improvement in terms of both detection precision and effective detection range for autonomous vehicles, compared to previous feature fusion solutions.

cs.CV↗

arXiv2020

DCANet: Learning Connected Attentions for Convolutional Neural Networks

While self-attention mechanism has shown promising results for many vision tasks, it only considers the current features at a time. We show that such a manner cannot take full advantage of the attention mechanism. In this paper, we present Deep Connected Attention Network (DCANet), a novel design that boosts attention modules in a CNN model without any modification of the internal structure. To achieve this, we interconnect adjacent attention blocks, making information flow among attention blocks possible. With DCANet, all attention blocks in a CNN model are trained jointly, which improves the ability of attention learning. Our DCANet is generic. It is not limited to a specific attention module or base network architecture. Experimental results on ImageNet and MS COCO benchmarks show that DCANet consistently outperforms the state-of-the-art attention modules with a minimal additional computational overhead in all test cases. All code and models are made publicly available.

cs.CV↗

arXiv2020

Hypersonic attachment-line instabilities with large sweep Mach numbers

This study aims to shed light on hypersonic attachment-line instabilities with large sweep Mach numbers. Highly swept flows over a cold cylinder that give rise to large sweep Mach numbers are studied. High fidelity base flows are obtained by solving full Navier-Stokes equations with a high-order shock-fitting method. Using local and global stability theories, an attachment-line mode is found to be dominant for the laminar-turbulent transition along the leading edge that agrees well with experimental observations Gaillard1999. The behavior of this mode explains the reason why the transition occurs earlier as the sweep Mach number is above 5. Also, this attachment-line mode is absent if the base flow is calculated with boundary-layer assumptions, indicating that the influence of inviscid flow outside the boundary layer can not be ignored as is normally done. It is clearly demonstrated that the global modes display the features of both attachment-line modes, as in sweep Hiemenz flow, and cross-flow-like modes further downstream along the surface. In contrast to incompressible flows, the mode is shown to be of inviscid nature. Moreover, the leading-edge curvature has a destabilizing effect on the attachment-line mode for large spanwise wave numbers but a stabilizing effect for small spanwise wave numbers.

physics.flu-dyn↗

arXiv2020

Image-based flow decomposition using empirical wavelet transform

We propose an image-based flow decomposition developed from the two-dimensional (2D) tensor empirical wavelet transform (EWT) (Gilles 2013). The idea is to decompose the instantaneous flow data, or its visualisation, adaptively according to the averaged Fourier supports for the identification of spatially localised structures. The resulting EWT modes stand for the decomposed flows, and each accounts for part of the spectrum, illustrating fluid physics with different scales superimposed in the original flow. With the proposed method, decomposition of an instantaneous 3D flow becomes feasible without resorting to its time series. Examples first focus on the interaction between a jet plume and 2D wake, where only experimental visualisations are available. The proposed method is capable of separating the jet/wake flows and their instabilities. Then the decomposition is applied to an early stage boundary layer transition, where direct numerical simulations provided a full data-set. The tested inputs are the 3D flow data and its visualisation using streamwise velocity & λ2 vortex identification criterion. With both types of inputs, EWT modes robustly extract the streamwise-elongated streaks, multiple secondary instabilities and helical vortex filaments. Results from bi-global stability analysis justify the EWT modes that represent the streak instabilities. In contrast to Proper Orthogonal Decomposition or Dynamic Modal Decomposition that extract spatial modes according to energy or frequency, EWT provides a new strategy as to decompose an instantaneous flow from its spatial scales.

physics.flu-dyn↗

arXiv2019

Low-Latency High-Level Data Sharing for Connected and Autonomous Vehicular Networks

Autonomous vehicles can combine their own data with that of other vehicles to enhance their perceptive ability, and thus improve detection accuracy and driving safety. Data sharing among autonomous vehicles, however, is a challenging problem due to the sheer volume of data generated by various types of sensors on the vehicles. In this paper, we propose a low-latency, high-level (L3) data sharing protocol for connected and autonomous vehicular networks. Based on the L3 protocol, sensing results generated by individual vehicles will be broadcasted simultaneously within a limited sensing zone. The L3 protocol reduces the networking latency by taking advantage of the capture effect of the wireless transmissions occurred among vehicles. With the proposed design principles, we implement and test the L3 protocol in a simulated environment. Simulation results demonstrate that the L3 protocol is able to achieve reliable and fast data sharing among autonomous vehicles.

cs.NI↗

arXiv2019

Cooper: Cooperative Perception for Connected Autonomous Vehicles based on 3D Point Clouds

Autonomous vehicles may make wrong decisions due to inaccurate detection and recognition. Therefore, an intelligent vehicle can combine its own data with that of other vehicles to enhance perceptive ability, and thus improve detection accuracy and driving safety. However, multi-vehicle cooperative perception requires the integration of real world scenes and the traffic of raw sensor data exchange far exceeds the bandwidth of existing vehicular networks. To the best our knowledge, we are the first to conduct a study on raw-data level cooperative perception for enhancing the detection ability of self-driving systems. In this work, relying on LiDAR 3D point clouds, we fuse the sensor data collected from different positions and angles of connected vehicles. A point cloud based 3D object detection method is proposed to work on a diversity of aligned point clouds. Experimental results on KITTI and our collected dataset show that the proposed system outperforms perception by extending sensing area, improving detection accuracy and promoting augmented results. Most importantly, we demonstrate it is possible to transmit point clouds data for cooperative perception via existing vehicular network technologies.

cs.CV↗

arXiv2018

Linear instability of Poiseuille flows with highly non-ideal fluids

The objective of this work is to investigate linear modal and algebraic instability in Poiseuille flows with fluids close to their vapour-liquid critical point. Close to this critical point, the ideal gas assumption does not hold and large non-ideal fluid behaviours occur. As a representative non-ideal fluid, we consider supercritical carbon dioxide (CO$_2$) at pressure of 80 bar, which is above its critical pressure of 73.9 bar. The Poiseuille flow is characterized by the Reynolds number ($Re=ρ_{w}^{*}u_{r}^{*}h^{*}/μ_{w}^{*}$), the product of Prandtl ($Pr=μ_{w}^{*}C_{pw}^{*}/κ_{w}^{*}$) and Eckert number ($Ec=u_{r}^{*2}/C_{pw}^{*}T_{w}^{*}$), and the wall temperature that in addition to pressure determines the thermodynamic reference condition. For low Eckert numbers, the flow is essentially isothermal and no difference with the well-known stability behaviour of incompressible flows is observed. However, if the Eckert number increases, the viscous heating causes gradients of thermodynamic and transport properties, and non-ideal gas effects become significant. Three regimes of the laminar base flow can be considered, subcritical (temperature in the channel is entirely below its pseudo-critical value), transcritical, and supercritical temperature regime. If compared to the linear stability of an ideal gas Poiseuille flow, we show that the base flow is more unstable in the subcritical regime, inviscid unstable in the transcritical regime, while significantly more stable in the supercritical regime. Following the corresponding states principle, we expect that qualitatively similar results will be obtained for other fluids at equivalent thermodynamic states.

physics.flu-dyn↗

arXiv2018

The new mode of instability in viscous high-speed boundary layer flows

The new mode of instability found by Tunney et al. is studied with viscous stability theory in this article. When the high-speed boundary layer is subject to certain values of favorable pressure gradient and wall heating, a new mode becomes unstable due to the appearance of the streamwise velocity overshoot ($U(y)>U_\infty$) in the base flow. The present study shows that under practical Reynolds numbers, the new mode can hardly co-exist with conventional first mode and Mack's second mode. Due to the requirement for additional wall heating, the new mode may only lead to laminar-turbulent transition under experimental (artificial) conditions.

physics.flu-dyn↗

arXiv2014

Characterizing Workload of Web Applications on Virtualized Servers

With the ever increasing demands of cloud computing services, planning and management of cloud resources has become a more and more important issue which directed affects the resource utilization and SLA and customer satisfaction. But before any management strategy is made, a good understanding of applications' workload in virtualized environment is the basic fact and principle to the resource management methods. Unfortunately, little work has been focused on this area. Lack of raw data could be one reason; another reason is that people still use the traditional models or methods shared under non-virtualized environment. The study of applications' workload in virtualized environment should take on some of its peculiar features comparing to the non-virtualized environment. In this paper, we are open to analyze the workload demands that reflect applications' behavior and the impact of virtualization. The results are obtained from an experimental cloud testbed running web applications, specifically the RUBiS benchmark application. We profile the workload dynamics on both virtualized and non-virtualized environments and compare the findings. The experimental results are valuable for us to estimate the performance of applications on computer architectures, to predict SLA compliance or violation based on the projected application workload and to guide the decision making to support applications with the right hardware.

cs.PF↗