Laguerre滤波器在抗噪语音识别特征提取中的应用

为克服FIR滤波器存在的通阻带特性差、滤波器阶次高等缺点给语音识别系统带来的不利影响,采用Laguerre滤波器组代替过零峰值幅度特征提取中使用的FIR滤波器组进行前端处理。在仔细研究FIR滤波器参数确定方法的基础上,叙述了Laguerre滤波器原理及参数计算方法,并给出了计算结果。孤立词、非特定人语音识别实验结果表明,使用Laguerre滤波器不仅使识别系统抗噪性能优于使用FIR滤波器,而且滤波器阶数也大为下降。     

在VC下读取地震勘探CST格式数据

本文分析了地震勘探数据CST的存放格式,并详细介绍了在VC环境下对该数据进行读取、格式转换以及存储为本机文件格式的一些处理。     

HPF高性能语言在地震资料并行处理中的应用

HPF（High Performance Fortran)是HPF论坛1993年推出的一种高性能数据并行语言。文章利用合作单位开发的p-HPF并行编译系统,在地震资料处理中得到了大规模的工业应用,基准测试程序和实际资料处理结果表明,基于HPF的地震资料并行处理具有可编程性强,可移植性好和并行效率高等特点,具有很好的应用前景。     

遥感技术在油气非地震勘探中的应用

油气遥感技术是油气非地震勘探中的一种有效方法。其理论依据是油气藏烃类微渗漏理论。本文通过实例表明：①在复杂的地表条件开展油气非地震勘探工作中，遥感微地貌解译可以为研究勘探区地球化学场和地球物理场的分布规律提供科学依据。②通过遥感影像异常内外地物波谱数据与化探多指标数据的因子分析可以进一步探讨影像异常与化探异常的相互关系。③对遥感图像进行特征主成分选择（ＦＰＣＳ）和ＨＳＩ逆变换等计算机处理，可提取与烃蚀变有关的异常，并得到了化探异常的验证。     

煤矿采区三维地震勘探技术的应用与效果分析

近年来,随着我国煤炭开采整合工作的持续推进,煤矿企业所面临的局面正在不断改观,无论是矿产资源开发规模还是集约化水平都得到了不同程度的提升。但是,在一些矿区和小煤窑生产区,留下了大量因无序开采而造成的采空区,对现代煤矿企业的生产活动带来了消极影响。文章以此为视角,对煤矿采区三维地震勘探技术的应用与效果进行了分析,以期为相关工作提供借鉴信息。     

卷积神经网络在地震断层自动识别中的应用

断层解释是地震资料解释的重要组成部分。长期以来,采用的地震断层解释方法对人工经验依赖性较高,例如相干体分析、人工蚂蚁算法、边缘检测等方式就非常依赖地质解释人员的经验。这样的计算方法容易因为人工错误而产生误差。因此,探究地震断层自动识别非常重要。     

深度学习在地震初至拾取中的应用综述

随着计算机硬件技术的不断提高,人工智能技术在各个行业的广泛应用,地震学界开始探索用深度学习算法处理地震初至波数据,为初至波的拾取研究提供了崭新的方向。为了解决传统地震波初至拾取方法对低信噪比资料拾取精度较低、算法鲁棒性较差等缺点,提高地震初至拾取速度和效率,减少耗费的人力以及人为拾取所产生的误差,该文系统地介绍了地震事件初至波拾取的常用传统方法,详细地阐述了深度学习领域内各种经典神经网络模型在地震初至波拾取中应用现状。通过实例对比分析卷积神经网络、深度信念网络、生成对抗神经网络和深度递归神经网络等模型在地震初至波拾取中的应用效果,讨论总结了深度学习在地震初至波拾取领域内应用存在问题和应用前景,为今后地震初至拾取研究提供新的思路。     

Korteweg-de Vries方程的时空谱配置方法

针对全直线上的KdV方程构造了时空全离散Legendre-Hermite谱配置格式,也就是在时间方向上用Legendre-Gauss-Lobatto节点为配置点,在空间方向上用Hermite-Gauss节点为配置点,构造得到一个非线性矩阵方程,将其化为非线性方程组,利用通常的不动点迭代求解,数值实验表明这种方法的有效性.     

三维Navier-Stokes方程的差分-谱方法混合法在GPU上的实现与优化

差分-谱方法通常在槽道湍流的直接数值模拟中使用,本文主要研究差分-谱方法在单GPU卡上的实现。由于GPU的硬件发展十分迅速,不同的GPU硬件对双精度计算的支持有所不同,本文首先验证GPU上数值计算的精度,用差分-谱混合法求解标量扩散方程,并将GPU和CPU上获取的数值结果与解析解进行对比,以确定GPU上数值算法实现的精确度。标量扩散方程在Nvidia S2050单GPU卡上求解,获得接近20倍的加速比,三维不可压缩Navier-Stokes方程达到了25倍的加速比。     

Generalized Laguerre spectral method for Fisher's equation on a semi-infinite interval

In this paper, we propose a generalized Laguerre spectral method for Fisher's-type equation with inhomogeneous boundary conditions on a semi-infinite interval. By reformulating the equation with suitable functional transform, it is shown that the generalized Laguerre approximations are convergent on a semi-infinite interval with spectral accuracy. An efficient and accurate algorithm based on the generalized Laguerre approximations to the transformed equation is developed and implemented. Numerical results show the efficiency of this approach and coincide well with theoretical analysis.     

Spherical harmonic-generalized Laguerre pseudospectral method for three-dimensional exterior problems

In this paper, we develop the mixed pseudospectral method for three-dimensional exterior problems. Some basic results on the mixed spherical harmonic-generalized Laguerre interpolation are established, which play important roles in the related pseudospectral methods. As examples, we provide the mixed pseudospectral schemes for two exterior problems with convergence analysis. Numerical results demonstrate the efficiency of this approach.     

Spherical Harmonic–Generalized Laguerre Spectral Method for Exterior Problems

In this paper, we propose the mixed spherical harmonic-generalized Laguerre spectral method for three-dimensional exterior problems. Some approximation results are established. As an example, a model problem is considered. The convergence of proposed scheme is proved. Numerical results demonstrate the efficiency of this approach.     

Semiconductor nanodevice simulation by multidomain spectral method with Chebyshev, prolate spheroidal and Laguerre basis functions

A new approach based on spectral method with efficient basis functions is proposed in the paper to simulate semiconductor nanodevice by solving the Schrödinger equation. The computational domain is partitioned at heterojunctions into a number of subdomains. The envelope functions in subdomains are expanded by various efficient basis functions and then patched by the BenDaniel-Duke boundary conditions to preserve exponential order of accuracy. Importantly, the consideration to choose the basis functions depends on the oscillatory characteristics of envelope functions. Three kinds of basis functions including prolate spheroidal wave functions, Chebyshev polynomials, and Laguerre-Gaussian functions are used according to the mathematical features in this work. In addition, the determinations of optimum values of scaling factor in Laguerre-Gaussian functions and bandwidth parameter in prolate spheroidal wave functions are also discussed in detail. Several quantum well examples are simulated to validate the effectiveness of the present scheme. The relative errors of energy levels achieve the order of 10⁻¹² requiring merely a few grid points.     

Chebyshev rational spectral method for long-short wave equations

We consider the initial boundary value problem of the long-short wave equations on the whole line. A fully discrete spectral approximation scheme is developed based on Chebyshev rational functions in space and central difference in time. A priori estimates are derived which are crucial to study numerical stability and convergence of the fully discrete scheme. Then, unconditional numerical stability is proved. Convergence of the fully discrete scheme is shown by the method of error estimates. Finally, numerical experiments are presented to demonstrate the efficiency and accuracy of the convergence results.     

谱元法在求解刚架结构动力学问题中的应用

把谱元法应用于刚架结构的动力学响应计算和分析中.建立了杆和梁的谱单元动力学刚度阵,针对刚架结构组装了整体动力学刚度阵,建立了整体结构的运动方程,计算了结构的固有频率和时域响应,并与采用有限元方法得到的结果进行了对比.从结果中可以看出谱元法在数值模拟中的独特优势.     

混合基函数勒让德谱元方法

针对勒让德谱元方法,构造了一类混合局部基函数,并证明了其线性无关特点.在此基础上,给出了一种勒让德谱元计算格式:在元素端点采用局部拉格朗日基函数,在元素内部采用调整后的局部勒让德多项式.进行了正确性和精度测试,数值实验结果表明该方法能够有效地实现高计算精度,且其计算矩阵比经典谱元方法更为简单,并具有良好的数据重用性和稀疏特点.     

用于海洋地震勘探的数字包设计与实现

提出了一种高精度、高性能的用于海洋地震勘探的拖缆数字包,并给出其设计与实现.每个数字包含有16个数据采集通道,每个通道都可以直接连接水听器,而水听器产生的模拟信号被由FPGA控制的24-bit高精度Δ-∑ADC实时采集.同时,基于串行并发总线结构,设计和实现了系统的实时控制和远距离数据传输与同步.测试结果表明:该拖缆数字包满足海洋地震勘探的要求.     

Dispersion analysis and computational efficiency of elastic lattice methods for seismic wave propagation

Discrete particle methods or elastic lattice methods represent a 3D elastic solid by a series of interconnected springs arranged on a regular lattice. Generally, these methods only consider nearest neighbour interactions, i.e. they are first-order in space. These interconnected springs interacted through a force term (Hooke's Law for an elastic body), which when viewed on a macroscopic scale provide a numerical solution for the elastodynamic wave equations. Along with solving the elastodynamic wave equations these schemes are capable of simulating elastic static deformation. However, as these methods rely on nearest neighbour interactions they suffer from more pronounced numerical dispersion than traditional continuum methods. By including a new force term, the numerical dispersion can be reduced while keeping the flexibility of the nearest neighbour interaction rule. We present results of simulations where the additional force term reduces the numerical dispersion and increases the accuracy of the elastic lattice method solution. The computational efficiency and parallel scaling of this method on multiple processors is compared with a finite-difference solution to assess the computational cost of using this approach for simulating seismic wave propagation. We also show the applicability of this method to modelling seismic propagation in a complex Earth model.