一种引入吸收衰减的压缩感知薄储层识别方法

常规基于压缩感知稀疏反演方法是基于频域平稳地震褶积模型进行的,而实际地下介质是黏弹性的,这使得该方法的反演反射系数振幅缺失、位置偏移。针对该问题,通过引入时频衰减因子,构建改进的衰减感知矩阵,将稀疏反演从常规的频率域拓展至衰减频率域。反演结果在一定程度上可恢复缺失的振幅,增强弱信号的识别能力。由于噪声影响,以上处理结果中仍存在噪声干扰,故在衰减频率域稀疏反演的基础上,引入平滑的高斯函数,对反演目标函数进行优化以压制残存的噪声干扰;之后将反演结果与褶积宽频子波可生成高分辨率地震剖面,由此形成了一种引入吸收衰减的压缩感知薄储层识别方法。薄层理论模型及含有河道砂储层的实际地震资料的处理结果表明,本文方法较常规方法,有效地增强了薄层弱信号的振幅及横向连续性,可在保证信噪比的情况下,提高地震资料的分辨率及薄储层的识别能力。

基于多路特征的改进生成对抗网络水下图像增强算法

针对水下图像颜色偏差、细节模糊以及对比度低等问题,提出多路特征增强生成对抗网络进行水下图像增强算法。首先,为实现对水下图像的色彩平衡,通过构建蓝绿通道与红色通道之间的衰减矩阵,设计红色通道补偿算法。其次,分别搭建生成网络和判别网络。为获得不同层次特征信息,生成网络由不同深度的特征提取网络组成,并搭建特征增强模块,并行增强多路特征,提高特征的质量。同时,为提高网络细节增强能力,搭建多尺度判别网络,对生成图像进行判别。最后,构建多个损失函数优化网络参数,提高网络鲁棒性。实验结果表明,增强图像相较原始图像色彩更加丰富,对比度较高。与现有的算法相比,该方法能够有效恢复水下图像的真实色彩和纹理细节,显著提高水下图像质量。

卫星定位误差椭球的几何特征研究

在卫星导航定位的误差分析中,准确估计定位误差椭球的几何特征有利于更好的描述定位精度,为估算定位精度参数提供依据。本文研究卫星定位中误差椭球的轴方向、轴长和轴比。通过分析精度衰减(Dilutionof Precision,DOP)矩阵的特征向量和特征值,分离其中的三维定位部分和与钟差有关部分,构建了定位误差椭球的理论模型,并给出误差椭球轴方向、轴长和轴比的计算公式。结合GPS星座,在全球范围内选取8个有代表性的地点对该模型开展仿真分析,表明该理论模型可以准确的描述定位误差椭球的几何特征,误差椭球的轴方向、轴长和轴比与模型给出的理论值非常符合。同时,利用8个地点中2个地点附近的IGS测站秒级实测数据进行定位解算,对定位误差椭球进行了分析,结果表明测站误差椭球诸参数的理论值与实测结果基本一致。

Seismic signatures of reservoir permeability based on the patchy-saturation model

Modeling of seismic responses of variable permeability on the basis of the patchy-sa^ration model provides insights into the seismic characterization of fluid mobility. We linked rock-physics models in the frequency domain and seismic modeling on the basis of the propagator matrix method. For a layered patchy-saturated reservoir, the seismic responses represent a combination of factors, including impedance contrast, the effect of dispersion and attenuation within the reservoir, and the tuning and interference of reflections at the top and bottom of the reservoir. Numerical results suggest that increasing permeability significantly reduces the P-wave velocity and induces dispersion between the high- and low-frequency elastic limit. Velocity dispersion and the layered structure of a reservoir lead to complex reflection waveforms. Seismic reflections are sensitive to permeability if the impedance of the reservoir is close to that of the surroundings. For variable layer thickness, the stacked amplitudes increase with permeability for high-velocity surrounding shale, whereas the stacked amplitudes decrease with permeability for low-velocity surrounding shale.

UE运动仿真系统的设计和实现

5G的大规模部署和性能优化,离不开高效的5G测试系统。目前,路测是评估移动通信网络性能的常用方法,需要人工改变用户终端(User Equipment,UE)的位置,因而存在高成本和低效率的缺点。利用衰减量变化模拟UE运动,基于衰减矩阵开发出自动化的UE运动仿真系统,可以部分替代人工路测,从而达到节约测试成本、加速产品开发过程的目的。基于此,笔者重点探讨UE运动仿真系统的设计和实现。

Time-Dependent Density Matrix Renormalization Group Coupled with n-Mode Representation Potentials for the Excited State Radiationless Decay Rate:Formalism and Application to Azulene

We propose a method for calculating the nonradiative decay rates for polyatomic molecules including anharmonic effects of the potential energy surface(PES)in the Franck-Condon region.The method combines the n-mode repre-sentation method to construct the ab initio PES and the nearly exact time-dependent density matrix renormalization group method(TD-DMRG)to simulate quantum dynamics.In addition,in the framework of TD-DMRG,we further develop an algorithm to calculate the final-state-resolved rate coefficient which is very useful to analyze the contribution from each vibrational mode to the transition process.We use this method to study the internal conversion(IC)process of azulene after taking into account the anharmonicity of the ground state PES.The results show that even for this semi-rigid molecule,the intramode anharmonicity enhances the IC rate significantly,and after considering the two-mode coupling effect,the rate increases even further.The reason is that the anharmonicity enables the C-H vibrations to receive electronic energy while C-H vibrations do not contribute on the harmonic PES as the Huang-Rhys factor is close to 0.

一种改进的自适应指数加权衰减记忆滤波算法

对卡尔曼滤波在GPS定位应用中估值发散的抑制进行了研究,首先解决了由于各个测量噪声之间相关导致噪声协方差阵无法满足对角矩阵的问题;接着对基于多重衰减因子矩阵的指数加权的衰减自适应算法的一步预测均方误差进行了UD分解,将分解矩阵应用于卡尔曼滤波过程更新,并给出了多重衰减记忆因子矩阵的求解过程。实验结果表明,这种算法增强了数值稳定性,提高了定位精度,具有抑制滤波发散的作用。