Adaptive mesh refinement computation of acoustic radiation from an engine intake

A block-structured adaptive mesh refinement (AMR) method was applied to the computational problem of acoustic radiation from an aeroengine intake. The aim is to improve the computational and storage efficiency in aeroengine noise prediction through reduction of computational cells. A parallel implementation of the adaptive mesh refinement algorithm was achieved using message passing interface. It combined a range of 2nd- and 4th-order spatial stencils, a 4th-order low-dissipation and low-dispersion Runge–Kutta scheme for time integration and several different interpolation methods. Both the parallel AMR algorithms and numerical issues were introduced briefly in this work. To solve the problem of acoustic radiation from an aeroengine intake, the code was extended to support body-fitted grid structures. The problem of acoustic radiation was solved with linearised Euler equations. The AMR results were compared with the previous results computed on a uniformly fine mesh to demonstrate the accuracy and the efficiency of the current AMR strategy. As the computational load of the whole adaptively refined mesh has to be balanced between nodes on-line, the parallel performance of the existing code deteriorates along with the increase of processors due to the expensive inter-nodes memory communication costs. The potential solution was suggested in the end.     

基于混合网格的三维Navier-Stokes方程并行算法

提出了一种基于混合网格的三维Navier-Stokes方程的并行计算方法。Navier-Stokes的求解采用了基于面的有限体积方法,该方法适用于任何网格类型。采用一方程Spalart-Allmaras模型来计算紊流黏性。并行计算采用区域分裂的方法,利用METIS网格分区系统实现了各节点的加载平衡。节点间的数据交换通过调用MPI库函数来实现,采用非阻断通讯的方式来减少数据交换时间。充分利用FORTRAN90的动态存储特性来减少对内存的需求。最后,通过对绕DLR-F6外形(翼身组合体+挂架+发动机短舱)黏性流动的数值模拟,验证了该并行程序的准确性,高性能并行计算以及处理复杂几何外形的能力。     

Design, implementation and evaluation of parallel pipelined STAP onparallel computers

Performance results are presented for the design and implementation of parallel pipelined space-time adaptive processing (STAP) algorithms on parallel computers. In particular, the issues involved in parallelization, our approach to parallelization, and performance results on an Intel Paragon are described. The process of developing software for such an application on parallel computers when latency and throughput are both considered together is discussed and tradeoffs considered with respect to inter and intratask communication and data redistribution are presented. The results show that not only scalable performance was achieved for individual component tasks of STAP but linear speedups were obtained for the integrated task performance, both for latency as well as throughput. Results are presented for up to 236 compute nodes (limited by the machine size available to us). Another interesting observation made from the implementation results is that performance improvement due to the assignment of additional processors to one task can improve the performance of other tasks without any increase in the number of processors assigned to them. Normally, this cannot be predicted by theoretical analysis     

A Parallel Adaptive 3D MHD Scheme for Modeling Coronal and Solar Wind Plasma Flows

A parallel adaptive mesh refinement (AMR) scheme is described for solving the governing equations of ideal magnetohydrodynamics (MHD) in three space dimensions. This solution algorithm makes use of modern finite-volume numerical methodology to provide a combination of high solution accuracy and computational robustness. Efficient and scalable implementations of the method have been developed for massively parallel computer architectures and high performance achieved. Numerical results are discussed for a simplified model of the initiation and evolution of coronal mass ejections (CMEs) in the inner heliosphere. The results demonstrate the potential of this numerical tool for enhancing our understanding of coronal and solar wind plasma processes. This revised version was published online in June 2006 with corrections to the Cover Date.     

High-order implicit discontinuous Galerkin schemes for unsteady compressible Navier–Stokes equations

Efficient solution techniques for high-order temporal and spatial discontinuous Galerkin(DG) discretizations of the unsteady Navier–Stokes equations are developed. A fourth-order implicit Runge–Kutta(IRK) scheme is applied for the time integration and a multigrid preconditioned GMRES solver is extended to solve the nonlinear system arising from each IRK stage. Several modifications to the implicit solver have been considered to achieve the efficiency enhancement and meantime to reduce the memory requirement. A variety of time-accurate viscous flow simulations are performed to assess the resulting high-order implicit DG methods. The designed order of accuracy for temporal discretization scheme is validate and the present implicit solver shows the superior performance by allowing quite large time step to be used in solving time-implicit systems. Numerical results are in good agreement with the published data and demonstrate the potential advantages of the high-order scheme in gaining both the high accuracy and the high efficiency.     

三维Euler方程组隐式LU分解的高性能并行计算

许多非定常无粘流体力学问题的数值模拟都需要利用Euler方程组来进行计算,而由于在隐格式下,所选取的时间步长可以比在显格式下时大得多,所以隐格式越来越受到重视,其中隐式LU分解是最常用的方法之一。对三维Euler方程组,采用隐式LU分解进行计算时,网格点所在的各个对角阵面之间存在数据依赖关系,本文分析了采用区域分解且边界上用显格式代替隐格式进行计算的高效性,在长方体建筑物内的爆炸模拟表明,在有112个CPU的某MPP巨型机上,并行计算效率超过60%。本文还分析了计算结果与串行计算时的差异,以及利用区域重叠减小这种差异的方法,同时考虑了对处理器进行合理的逻辑组织,将计算网格映射到处理器网格,以最大限度减少通信开销的方法。文中最后以一个爆炸毁伤的例子实际说明了所述方法的可行性与高效性。     

Pseudo-linear inertial navigation algorithm

A new method is illustrated for processing the output of a set of triad orthogonal rate gyros and accelerometers to reconstruct vehicle navigation parameters (attitude, velocity, and position). The paper introduces two vectors with dimensions 4 × 1 as velocity and position quaternions. The navigation equations for strapdown systems are nonlinear but after using these parameters, the navigation equations are converted into a pseudo-linear system. The new set of navigation equations has an analytical solution and the state transition matrix is used to solve the linear time-varying differential equations through time series. The navigation parameters are updated using the new formulation for strapdown navigation equations. Finally, the quaternions of velocity and position are converted into the original position and velocity vectors. The combination of the coning motion and a translational oscillatory trajectory is used to evaluate the accuracy of the proposed algorithm. The simulations show significant improvement in the accuracy of the inertial navigation system, which is achieved through the mentioned algorithm.     

面向机载应用的领域专用加速器研究

机载环境下数据处理规模的剧增以及人机混合智能的应用使得传统的以CPU 为核心计算单元的架构 已不能满足计算需求。在满足延时、精度等指标的情况下,选用高能效的处理器或处理器组合来快速准确地处 理这些数据成为机载计算领域面临的重要问题。按照常规处理器、领域专用加速器两大类型对各自主要代表性 处理器的架构特点进行了分析和总结,得出了各类处理器在机载情况下的主要适用场景和应用情况。根据领域 专用的设计思想开发了面向数据关联应用的专用加速器,对数据关联算法中的统计距离计算和分簇处理这两个 计算瓶颈进行了定制化的加速设计,并在基于FPGA 的平台上进行了测试验证,结果表明,加速器对于统计距 离计算的加速效果约为FT2000/4 单核性能的10 倍,对于分簇处理的加速效果约为FT2000/4 单核性能的3 倍, 整体运算速度相比FT2000/4 处理器的单核提升了5 倍。     

受约束多刚体系统动力学的计算机辅助分析

 以Newton-Euler法为基础建立了受约束多刚体系统动力学的一种新的递推算法,采用奇异值分解(SVD)对约束系统动力学方程缩聚求解。编制了计算机辅助分析通用程序CADACMS。     

悬臂式挠性薄片气体动压径向轴承分析研究

本文建立了悬臂式挠性薄片气体动压径向轴承的特性分析计算方法,用差分方法耦合求解径向轴承的雷诺方程和薄片弹性变形方程。给出了径向轴承刚度、阻尼的计算方法和稳定性计算方法。计算分析了设计的悬臂式薄片气体动压径向轴承的静特性和动特性。计算结果表明,对于所有的偏心率范围和可能的工作运行范围内是恒定稳定的。      

单列叶栅紊流流场有限分析数值研究

本文将有限分析方法用于曲线座标系上紊流 N-S方程数值计算,研究了高雷诺数时叶栅粘性紊流流场。有限分析方法在网格单元上对非线性偏微分程进行线化处理,在解析边界条件下求出线化偏微分方程的解析解,以此解构造离散方程。有限分析方法能够根据对流的方向和大小自动改变格式系数,并具有数值扩散小、精度高和稳定性好等优点。本文以 k-ε紊流模型模化紊流,以壁面函数方法处理近壁区流动参数。      

基于ADE-ELM的涡轴发动机建模方法

提出了基于自适应微分进化-极端学习机(ADE-ELM)求解平衡方程的高精度涡轴发动机实时部件级模型建立方法.基于牛顿-拉夫逊(N-R)迭代模型,以迭代计算前模型平衡方程残差为输入,迭代收敛后平衡方程猜值修正量为输出,训练极端学习机,并采用自适应微分进化(ADE)算法优化极端学习机(ELM)参数,提高猜值修正量映射精度.ADE算法中采用sigmoid型自适应缩放因子,提高了微分进化算法的寻优能力.在涡轴发动机不同飞行状态下的测试结果表明,以N-R迭代算法模型为基准,基于ADE-ELM的发动机模型,最大建模误差约为一次通过算法的1/3,运算耗时约为一次通过算法的1/3,验证了算法的有效性.      

火箭发动机两相喷雾燃烧的并行虚拟机仿真

采用计算流体动力学方法对液体火箭发动机内部多维喷雾两相燃烧过程进行了数值模拟。气相控制方程组用欧拉坐标系下的Ｎａｖｉｅｒ－Ｓｔｏｋｅｓ方程组描述,液相控制方程组在Ｌａｇｒａｎｇｉａｎ坐标系下进行描述。气、液两相作用通过方程组的源项互相耦合,编制了串行和并行程序,并在并行虚拟机环境下进行了测试。计算结果显示并行计算的效率较高。     

Application of a Highly Parallel Processor to Radar Data Processing

The application of a highly parallel computer known as PEPE (Parallel Element Processing Ensemble) to radar data processing is described. The PEPE computer consists of a large number of identical processing elements which operate in parallel and are controlled by a common control unit. Each of the processing elements is assigned to a particular radar target such that many targets can be tracked in parallel. PEPE is designed to augment a conventional computer rather than to stand alone. The total data processing load associated with a radar tracking system is distributed between PEPE and the sequential machine in a manner than maximizes the overall system efficiency and desensitizes the system performance to fluctuations in traffic levels. The use of PEPE provides very high data processing throughput potential to a radar data processing system.     

Neural networks in nonlinear aircraft flight control

This paper describes an approach for incorporating a neural network with real-time learning capability in a flight control architecture. The architecture is also applicable, in general, for the control of processes described by nonlinear differential equations of motion in which there exists a control for each degree of freedom. The main features are that the defining equations of motion for the process to be controlled are poorly known with respect to their functional forms, and that the functional forms, themselves, may undergo sudden and unexpected variation. It is well known that such systems are difficult to control, particularly when the effect of the control action enters nonlinearly. Numerical results based on 6DOF simulations of a high performance aircraft are presented to illustrate the potential benefits of incorporating neural networks as a part of a flight control system architecture     

Systematic Geometric Error Modeling for Workspace Volumetric Calibration of a 5-axis Turbine Blade Grinding Machine

A systematic geometric model has been presented for calibration of a newly designed 5-axis turbine blade grinding machine. This machine is designed to serve a specific purpose to attain high accuracy and high efficiency grinding of turbine blades by eliminating the hand grinding process. Although its topology is RPPPR (P: prismatic; R: rotary), its design is quite distinct from the competitive machine tools. As error quantification is the only way to investigate, maintain and improve its accuracy, calibration is recommended for its performance assessment and acceptance testing. Systematic geometric error modeling technique is implemented and 52 position dependent and position independent errors are identified while considering the machine as five rigid bodies by eliminating the set-up errors of workpiece and cutting tool. 39 of them are found to have influential errors and are accommodated for finding the resultant effect between the cutting tool and the workpiece in workspace volume. Rigid body kinematics techniques and homogenous transformation matrices are used for error synthesis.     

Analysis of Polyphase Commutator Generators for Wind-Power Applications

This paper describes the mathematical modeling of the ac polyphase hase commutator generator by means of Park's equations. For clarity, a two-phase, balanced-operation machine is analyzed. Equations ions of performance are developed in terms of familiar parameters. The machine is shown to have attractive characteristics for variablespeed peed constant-frequency power generation, with possible application to wind-power systems.     

Performance limits for processor networks with divisible jobs

Ultimate performance limits to the aggregate processing speed of networks of processors that are processing a divisible job are described. These take the form of either closed-form expressions or numerical procedures to calculate the equivalent processing speed of an infinite number of processors. These processors are interconnected in either a linear daisy chain with load origination from the network interior or a tree topology. The tree topology is particularly general as a natural way to perform load distribution in a professor network topology with cycles (e.g., hypercube, toroidal network) is to use an embedded spanning tree. Such limits on performance are important as they provide an ideal baseline against which to compare the performance of finite configurations of processors.     

N-S方程有限分析数值研究

 本文将有限分析方法用于曲线座标系上紊流N-S方程的数值计算。分别计算了单列和串列叶栅内部流场,计算中采用了k-ε紊流模型和壁面函数。计算结果与试验结果相比较,吻合程度令人满意。有限分析方法在网格单元上对N-S方程进行线化处理,以解析边界条件作为约束,得出解析解,在解析解基础上构造离散代数方程。有限分析方法的最大特点是可以适应对流速度大小和方向,自动调整格式系数,因而具有数值扩散小和稳定性高等优点。     

基于Heun法的模糊控制系统

针对有些常微分方程求解难且不易用图形直观观察的特点,根据求解常微分方程的Heun法构造了一种模糊控制系统以辨识由常微分方程所描述的系统。这种模糊控制系统在预测状态的同时也能够逼近系统微分方程中的函数。