人工神经网络在地质学判别分析中的应用

将现代人工神经网络方法应用于地质学的判别分析，通过实例计算表明，无论是对于两组判别分析还是多组判别分析，人工神经网络方法都是切实可行的。     

Large eddy simulation of hydrocyclone—prediction of air-core diameter and shape

Hydrocyclones are widely used in the mining and chemical industries. An attempt has been made in this study, to develop a CFD (computational fluid dynamics) model, which is capable of predicting the flow patterns inside the hydrocyclone, including accurate prediction of flow split as well as the size of the air-core. The flow velocities and air-core diameters are predicted by DRSM (differential Reynolds stress model) and LES (large eddy simulations) models were compared to experimental results. The predicted water splits and air-core diameter with LES and RSM turbulence models along with VOF (volume of fluid) model for the air phase, through the outlets for various inlet pressures were also analyzed. The LES turbulence model led to an improved turbulence field prediction and thereby to more accurate prediction of pressure and velocity fields. This improvement was distinctive for the axial profile of pressure, indicating that air-core development is principally a transport effect rather than a pressure effect.     

复式河槽正常水深计算方法比较

明渠恒定均匀流正常水深在水力计算中十分重要。而对于复式河槽,当洪水漫滩后,直接运用曼宁方程推求正常水深,将带来很大误差。系统地总结了计算正常水深的各种方法,用二分法求解,并运用英国科学工程研究协会洪水水槽设施(SERC-FCF)的大量系列水槽实验成果加以比较。比较发现,精度由高到低依次为COHM方法、刘沛清法、断面倾斜分割法、动量传递法、谢汉祥法、断面叠加法。通过对各种方法详细分析,建议在计算天然复式河槽正常水深时,COHM方法是值得推广的一种好方法。     

Numerical evaluation of the damping‐solvent extraction method in the frequency domain

The damping‐solvent extraction method for the analysis of unbounded visco‐elastic media is evaluated numerically in the frequency domain in order to investigate the influence of the computational parameters—domain size, amount of artificial damping, and mesh density—on the accuracy of results. An analytical estimate of this influence is presented, and specific questions regarding the influence of the parameters on the results are answered using the analytical estimate and numerical results for two classical problems: the rigid strip and rigid disc footings on a visco‐elastic half‐space with constant hysteretic material damping. As the domain size is increased, the results become more accurate only at lower frequencies, but are essentially unaffected at higher frequencies. Choosing the domain size to ensure that the static stiffness is computed accurately leads to an unnecessarily large domain for analysis at higher frequencies. The results improve by increasing artificial damping but at a slower rate as the total (material plus artificial) damping ratio ζ_t gets closer to 0.866. However, the results do not deteriorate significantly for the larger amounts of artificial damping, suggesting that ζ_t≈0.6 is appropriate; a larger value is not likely to influence the accuracy of results. Presented results do not support the earlier suggestion that similar accuracy can be achieved by a large bounded domain with small damping or by a small domain with larger damping. Copyright © 2002 John Wiley & Sons, Ltd.     

Numerical modeling of toxic nonaqueous phase liquid removal from contaminated groundwater systems: mesh effect and discretization error estimation

Numerical modeling has now become an indispensable tool for investigating the fundamental mechanisms of toxic nonaqueous phase liquid (NAPL) removal from contaminated groundwater systems. Because the domain of a contaminated groundwater system may involve irregular shapes in geometry, it is necessary to use general quadrilateral elements, in which two neighbor sides are no longer perpendicular to each other. This can cause numerical errors on the computational simulation results due to mesh discretization effect. After the dimensionless governing equations of NAPL dissolution problems are briefly described, the propagation theory of the mesh discretization error associated with a NAPL dissolution system is first presented for a rectangular domain and then extended to a trapezoidal domain. This leads to the establishment of the finger‐amplitude growing theory that is associated with both the corner effect that takes place just at the entrance of the flow in a trapezoidal domain and the mesh discretization effect that occurs in the whole NAPL dissolution system of the trapezoidal domain. This theory can be used to make the approximate error estimation of the corresponding computational simulation results. The related theoretical analysis and numerical results have demonstrated the following: (1) both the corner effect and the mesh discretization effect can be quantitatively viewed as a kind of small perturbation, which can grow in unstable NAPL dissolution systems, so that they can have some considerable effects on the computational results of such systems; (2) the proposed finger‐amplitude growing theory associated with the corner effect at the entrance of a trapezoidal domain is useful for correctly explaining why the finger at either the top or bottom boundary grows much faster than that within the interior of the trapezoidal domain; (3) the proposed finger‐amplitude growing theory associated with the mesh discretization error in the NAPL dissolution system of a trapezoidal domain can be used for quantitatively assessing the correctness of computational simulations of NAPL dissolution front instability problems in trapezoidal domains, so that we can ensure that the computational simulation results are controlled by the physics of the NAPL dissolution system, rather than by the numerical artifacts. Copyright © 2014 John Wiley & Sons, Ltd.     

Efficient DEM‐based overland flow routing using integrated recursive algorithms

Two primary concerns in performing watershed overland flow routing are the numerical instability and computational efficiency. The stability of executing an explicit scheme has to be maintained by observing the Courant–Friedrich–Lewy criterion, which is adopted to confirm that the numerical marching speed is larger than the wave celerity. Moreover, there is another criterion of time step devised in previous studies to avoid back‐and‐forth refluxing between adjacent grids. The situation of refluxing usually occurs on flat regions. In light of this, the selection of a small time increment to honor both restrictions simultaneously is believed to decrease the computational efficiency in performing overland flow routing. This study aims at creating a robust algorithm to relax both restrictions. The proposed algorithm was first implemented on a one‐dimensional overland plane to evaluate the accuracy of the numerical result by comparing it with an analytical solution. Then, the algorithm was further applied to a watershed for 2D runoff simulations. The results show that the proposed integrated algorithm can provide an accurate runoff simulation and achieve satisfactory performance in terms of computational speed.     

Computer-aided design of bus route maps

The bus route map is a diagram that aims to convey necessary information for map readers to find an appropriate way of moving from an origin to a destination. Design of bus route map is a complicated and time-consuming task that requires careful consideration of readability and aesthetics. This paper proposes a new computational method for designing bus route maps. The method helps us to reduce six types of undesirable elements in bus route maps, i.e., gap, shift, crossing, overlap, misalignment, and acute bend. The method consists of two phases: line layout phase determines the relative order of bus routes on each road segment and map layout phase calculates the actual position of bus routes drawn on a map. This paper applies the method to the design of bus route maps of Chiba City, Japan. The result supports the effectiveness of the method as well as reveals open topics for future research.     

河元背地区绳索取心钻进泵压计算模型的建立与应用

泵压是反映金刚石绳索取心钻进时井内作业是否正常的重要参数。为了进一步研究泵压波动变化与实际钻进作业之间的关系,指导钻进工作的快速、安全开展,结合江西相山河元背地区CUSD2井实际泵压波动数据及相关钻进资料,建立了适用于本井的循环系统压力损失计算模型。对比分析模型计算的理论泵压与实际泵压的波动变化趋势,将模型应用于实际钻进,在此基础上预测后续地层的泵压波动区域范围,发现7 MPa的泵压安全值不再适用于后续地层钻进,调整设置泵压安全值为10 MPa,并利用邻井泵压波动数据进一步验证其可靠性。适当调整修正理论模型,使其可应用于河元背以及相山地区绳索取心钻进中,指导该地区钻探工作。     

平面点集凸壳的一种快速算法

提出一种计算平面点集凸壳的快速算法———八方向极值快速凸壳算法。该算法首先对平面点集进行一次扫描,从而快速查找到东、南、西、北、东南、西南、东北、西北8个方向上的极值点,构造出一个更接近凸壳的初始凸壳,从而在后续的点集扫描中可以排除更多的内点,使该算法计算效率更高。该算法的空间复杂度为O(N);其时间复杂度虽然无法突破最坏情况下O(NlogN)的理论下限,但其期望时间复杂度已达到线性水平,并且可以容易地扩展到三维和高维空间。     

基于GPU加速的Boussinesq类波浪传播变形数值模型

Boussinesq波浪模型是一类相位解析模型,在时域内求解需要较高的空间和时间分辨率以保证计算精度。为提高计算效率,有必要针对该类模型开展并行算法的研究。与传统的中央处理器(CPU)相比,图形处理器(GPU)有大量的运算器,可显著提高计算效率。基于统一计算设备架构CUDA C语言和图形处理器,实现了Boussinesq模型的并行运算。将本模型的计算结果同CPU数值模拟结果和解析解相比较,发现得到的结果基本一致。同时也比较了CPU端与GPU端的计算效率,结果表明,GPU数值模型的计算效率有明显提升,并且伴随数值网格的增多,提升效果更为明显。