The application of artificial neural networks (ANNs) can be found in numerous fields, including image and speech recognition, natural language processing, autonomous vehicles. As well, intrusion detection, the subject this paper, relies heavily on it. Different detection models have been constructed using ANNs. While ANNs are relatively mature to construct models, some challenges remain. Among most notorious these bloated caused by large number parameters, non-interpretability models. Our...

Abstract The coseismic surface ruptures associated with large earthquakes contribute to severe damage near-fault buildings through fault deformation. However, previous studies simplified geological conditions and were based mainly on numerical physical simulations. In other words, the scarcity of earthquakes, especially for active thrust faults, limits understanding mechanisms building near faults. Herein, study selected 2008 Mw7.9 Wenchuan earthquake as an example. Based compass...

Seismic facies analysis is to study the sedimentary environment of stratigraphic sequence and provides an important basis for reservoir prediction. Most existing methods have low efficiency heavily rely on manual experience, therefore, it difficult interpret increasingly complex seismic data. Deep learning techniques can help solve these problems achieve automatic classification. We regard classification as a target segmentation problem propose new method training strategies. Our workflow...

Seismic facies interpretation provides a reference for analyzing geological conditions and predicting oil gas reservoirs. The application of deep learning in seismic reduces lot manual work interpreters' subjective effects existing conventional methods. Convolutional neural network (CNN) is widely used technique learning. However, CNN not the best model interpreting massive dataset due to its low efficiency classification accuracy. Given this issue, we propose an effective scheme, which...

Summary Least-squares reverse-time migration (LSRTM) is able to produce images with fewer artefacts, higher resolution and more accurate amplitudes over conventional method. Because of these benefits, LSRTM attracts greater interest in recent years. However, the computational cost much expensive than traditional In order reduce calculation time, we propose a method combine CPU GPU together accelerate multi-source using MPI CUDA. This can parallel both coarse-grain fine-grain make full use...

Previous No AccessSEG 2017 Workshop: Carbonate Reservoir E&P Workshop, Chengdu, China, 22-24 October 2017VSP reverse time migration based on complex wavefield decompositionAuthors: Hao Xue*Yang LiuHao Xue*State Key Laboratory of Petroleum Resources and Prospecting, China University Petroleum, Beijing, ChinaCNPC Geophysical Yang LiuState Chinahttps://doi.org/10.1190/carbonate2017-42 SectionsAboutPDF/ePub ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions...

Video saliency prediction is an important task in the field of computer vision. Most existing video methods only focus on image information, and audio information often ignored. This leads to incomplete perception mode, which makes it difficult achieve optimal performance. SENet excellent attention mechanism-based network. It significantly enhances performance 2D convolutional networks. However, whether 3D network can be applied this mechanism remains studied. In order solve above problems,...

Summary The previous pure acoustic anisotropic wave equations (PAAWEs) mainly derived from the decoupled elastic wavefield and had complicated spatial pseudo differential operator (SPDO). It can directly apply some spectral methods to these or employ series expansions approximate SPDO then solve P-wave equations. Considering need fairly large computational load low-order have accuracy loss in cases, here, we present corresponding solutions issues. square root term Alkhalifah’s accurate...

Summary The explicit rotated staggered-grid finite-difference method (ERSG-FDM) has been successfully applied to simulate seismic wave propagation in viscoelastic and anisotropic media. Nevertheless, the implicit stencils can effectively increase accuracy stability of (FD) numerical modeling. In this paper, order improve modeling elastic tilted transversely isotropic (TTI) media without increasing computational resources, we optimal coefficients ERSG-FDM extend globally (ISG-FDM) based on...

Quality factor Q is commonly used to measure the attenuation effect of seismic waves, which great importance inverse filtering, fluid identification and reservoir prediction. There are many methods estimate quality Q, mainly including logarithmic spectral ratio method centroid frequency shifting method. For VSP records, reliable value can be obtained using these two methods. However, for surface barely calculated because it difficult remove spectrum reflection coefficient from that record...

Summary In this study, we implement least-square reverse time migration (LSRTM) using one-step extrapolated exponential matrix method and wavefield decomposition imaging condition. One-step employs the Chebyshev polynomial expansion to approximate of it propagates waves free numerical dispersion noise. Using an analytical generated by method, can separate into upgoing downgoing both in source receiver explicit way, which improve efficiency iterative LSRTM algorithm. The condition is used...

Reverse-time migration (RTM) has been a popular seismic imaging method due to its high accuracy, no dip-angle limitation, and good adaptability for complex models. For the conventional RTM, cross-correlation condition is widely used may generate severe low-frequency noise in imaging. The wavefield decomposition approach can reduce such some extent. We apply recently developed eight-direction into RTM of 2D vertical profile (VSP) data. To further improve optimal finite-difference improved...