基于小波分析和集成学习的短时交通流预测

针对短时交通流具有随机性和不确定性等特征,提出一种基于小波分析和集成学习的组合预测模型.首先,对原始交通流数据的平均行程时间序列应用Mallat算法进行多尺度小波分解,且对各尺度上分量进行单支重构;其次,对于各重构的单支序列分别使用极端梯度提升模型(extreme gradient boosting,XGBoost)进...     

飞行计划集中处理的流量瓶颈点的识别与优化

为建设全国级的空中交通流量管理系统提供智力支持,将飞行计划集中处理这一手段,运用于预战术飞行流量管理阶段的流量管理方法和技术的研究之中,提供了一条新的研究思路.首先,利用过点时间预测算法计算过点时间,并根据过点时间统计出关键航路点的流量预测值;接下来,通过航路点容量模型识别航路上飞行流量大于容量的航路点,即流量瓶颈点;...     

喜鹊在山地次生林中鸟类群落组织结构形成的作用

通过山地次生林鸟类群落组织结构的研究,表明喜鹊在鸟类群落组织形成中是关键类群之一.关键类群是通过巢地资源的被利用来实现的.5年中,利用喜鹊巢繁殖的鸟类达5种,利用率为25%~40.7%.     

AN EXPLICIT FINITE VOLUME SPATIAL MARCHING METHOD FOR REDUCED NAVIER–STOKES EQUATIONS

This paper develops a spatial marching method for high-speed flows based on a finite volume approach. The method employs the reduced Navier– Stokes equations and a pressure splitting in the streamwise direction based on the Vigneron strategy. For marching from an upstream station to one downstream the modified five-level Runge–Kutta integration scheme due to Jameson and Schmidt is used. In addition, for shock handling and for good convergence properties the method employs a matrix form of the artificial dissipation terms, which has been shown to improve the accuracy of predictions. To achieve a fast rate of convergence, a local time-stepping concept is used. The method retains the time derivative in the governing equations and the solution at every spatial station is obtained in an iterative manner. The developed method is validated against two test cases: (a) supersonic flow past a flat plate; and (b) hypersonic flow past a compression corner involving a strong viscous–inviscid interaction. The computed wall pressure and wall heat transfer coefficients exhibit good general agreement with previous computations by other investigators and with experiments.     

A NUMERICAL PROCEDURE FOR PREDICTING MULTIPLE SOLUTIONS OF A SPHERICAL TAYLOR–COUETTE FLOW

A new numerical procedure for predicting multiple solutions of Taylor vortices in a spherical gap is presented. The steady incompressible Navier–Stokes equations in primitive variables are solved by a finite- difference method using a matrix preconditioning technique. Routes leading to multiple flow states are designed heuristically by imposing symmetric properties. Both symmetric and asymmetric solutions can be predicted in a deterministic way. The current procedure gives very fast convergence rate to the desired flow modes. This procedure provides an alternative way of finding all possible stable steady axisymmetric flow modes.     

ON THE STRUCTURE OF STEADY FLOW THROUGH DUAL-INTAKE ENGINE PORTS

The influence of intake port design on the flow field in a dual-intake valve engine was investigated using computational fluid dynamics, in order to study the effect of inlet port design on the in-cylinder flow. A detailed 3D computational grid incorporating all the features of the Ford Zetec production engine inlet ports, valves and cylinder head was initially created and the flow structure modelled at 5 and 10 mm valve lifts under steady flow conditions. Comparisons of computational results with experimental data obtained by laser Doppler anemometry indicate that the flow characteristics have been predicted well in most regions. Flow generated by different intake port designs was also simulated by introducing air into the cylinder at different directions to the inlet valve axes and the effects of port deactivation, throttling and exhaust gas recirculation were examined. The implications of the results for intake port design are discussed.     

NUMERICAL SIMULATION OF THREE-DIMENSIONAL INCOMPRESSIBLE FLOW BY A NEW FORMULATION

A new mathematical formulation, called the pseudovorticity–velocity formulation, of the three-dimensional incompressible Navier–Stokes equations is presented as an alternative to the vorticity–velocity approach. For the model lid-driven cavity flow problem in two and three dimensions, combined with an explicit mixed spectral /finite different numerical scheme the proposed formulation is found to be efficient and very accurate as compared with the results available in the literature. In particular, the simulation results demonstrate an attractive feature of the present formulation compared with the vorticity–velocity approach, namely that the divergence-free condition of the velocity field can always be achieved on a non-staggered mesh.     

A PRESSURE-CORRECTION METHOD FOR THE SOLUTION OF INCOMPRESSIBLE VISCOUS FLOWS ON UNSTRUCTURED GRIDS

An unstructured grid, finite volume method is presented for the solution of two-dimensional viscous, incompressible flow. The method is based on the pressure-correction concept implemented on a semi-staggered grid. The computational procedure can handle cells of arbitrary shape, although solutions presented herein have been obtained only with meshes of triangular and quadrilateral cells. The discretization of the momentum equations is effected on dual cells surrounding the vertices of primary cells, while the pressure-correction equation applies to the primary-cell centroids and represents the conservation of mass across the primary cells. A special interpolation scheme s used to suppress pressure and velocity oscillations in cases where the semi-staggered arrangement does not ensure a sufficiently strong coupling between pressure and velocity to avoid such oscillations. Computational results presented for several viscous flows are shown to be in good agreement with analytical and experimental data reported in the open literature.     

A PARALLEL BLOCK-STRUCTURED MULTIGRID METHOD FOR THE PREDICTION OF INCOMPRESSIBLE FLOWS

In this paper a parallel multigrid finite volume solver for the prediction of steady and unsteady flows in complex geometries is presented. For the handling of the complexity of the geometry and for the parallelization a unified approach connected with the concept of block-structured grids is employed. The parallel implementation is based on grid partitioning with automatic load balancing and follows the message-passing concept, ensuring a high degree of portability. A high numerical efficiency is obtained by a non-linear multigrid method with a pressure correction scheme as smoother. By a number of numerical experiments on various parallel computers the method is investigated with respect to its numerical and parallel efficiency. The results illustrate that the high performance of the underlying sequential multigrid algorithm can largely be retained in the parallel implementation and that the proposed method is well suited for solving complex flow problems on parallel computers with high efficiency.     

COMPUTATION OF DILUTE PARTICULATE LAMINAR FLOW OVER A BACKWARD-FACING STEP

Particle-laden flows are calculated for a classical laminar backward- facing step problem. The particle tracks are calculated using a recently developed exponential Lagrangian tracking scheme. The behaviour of the particle-laden flow is considered for various inlet for Reynolds number, Stokes numbers and void fractions. Doping the flow with low-Stokes-number particles has the effect of increasing the inlet inertia of the flow and this increases the strength of the recirculation behind the step. High-Stokes- number particles are dominated by gravitational effects which affect the flow accordingly. Differences between the single-phase flow and the particle-laden flows are therefore dependent on the Stokes number and increase linearly with void fraction.