Seismic activity prediction using computational intelligence techniques in northern Pakistan

Earthquake prediction study is carried out for the region of northern Pakistan. The prediction methodology includes interdisciplinary interaction of seismology and computational intelligence. Eight seismic parameters are computed based upon the past earthquakes. Predictive ability of these eight seismic parameters is evaluated in terms of information gain, which leads to the selection of six parameters to be used in prediction. Multiple computationally intelligent models have been developed for earthquake prediction using selected seismic parameters. These models include feed-forward neural network, recurrent neural network, random forest, multi layer perceptron, radial basis neural network, and support vector machine. The performance of every prediction model is evaluated and McNemar’s statistical test is applied to observe the statistical significance of computational methodologies. Feed-forward neural network shows statistically significant predictions along with accuracy of 75% and positive predictive value of 78% in context of northern Pakistan.     

地震油气储层的小样本卷积神经网络学习与预测

地震储层预测是油气勘探的重要组成部分,但完成该项工作往往需要经历多个环节,而多工序或长周期的研究分析降低了勘探效率.基于油气藏分布规律及其在地震响应上所具有的特点,本文引入卷积神经网络深度学习方法,用于智能提取、分类并识别地震油气特征.卷积神经网络所具有的强适用性、强泛化能力,使之可以在小样本条件下,对未解释地震数据体进行全局优化提取特征并加以分类,即利用有限的已知含油气井段信息构建卷积核,以地震数据为驱动,借助卷积神经网络提取、识别蕴藏其中的地震油气特征.将本方案应用于模型数据及实际数据的验算,取得了预期效果.通过与实际钻井信息及基于多波地震数据机器学习所预测结果对比,本方案利用实际数据所演算结果与实际情况有较高的吻合度.表明本方案具有一定的可行性,为缩短相关环节的周期提供了一种新的途径.     

AUTOMATED FIRST ARRIVAL PICKING: A NEURAL NETWORK APPROACH1

A back-propagation neural network is successfully applied to pick first arrivals (first breaks) in a background of noise. Network output is a decision whether each half-cycle on the trace is a first or not. 3D plots of the input attributes allow evaluation of the attributes for use in a neural network. Clustering and separation of first break from non-break data on the plots indicate that a neural network solution is possible, and therefore the attributes are suitable as network input. Application of the trained network to actual seismic data (Vibroseis and Poulter sources) demonstrates successful automated first-break selection for the following four attributes used as neural network input: (1) peak amplitude of a half-cycle; (2) amplitude difference between the peak value of the half-cycle and the previous (or following) half-cycle; (3) rms amplitude ratio for a data window (0.3 s) before and after the half-cycle; (4) rms amplitude ratio for a data window (0.06 s) on adjacent traces. The contribution of the attributes based on adjacent traces (4) was considered significant and future work will emphasize this aspect.     

基于机器学习算法的银川台阵资料自动化震相分析和定位

近年来快速发展的机器学习算法显著提高了震相拾取的精度和效率.采用卷积神经网络和递归神经网络的震相识别方法对银川台阵2019年6～7月的连续波形数据进行事件检测和P、S震相拾取,并通过快速震相关联和事件定位得到了银川地区较全的地震目录.结果表明,当震相数小于10时,虽然可以检测出较多事件,但分布呈弥散状,与区域地震活动特...     

基于卷积神经网络的断层预测方法

针对传统相干体属性在预测断层时存在断层假象以及易受噪声影响等缺点,本文提出一种利用卷积神经网络进行断层预测的方法。首先构建适合实际工区断层特征的卷积神经网络模型,然后利用部分分频地震数据和人工解释出的断层标签进行网络模型训练,最后把训练好的模型应用到整个三维地震数据中进行断层预测。实际地震数据预测结果表明基于卷积神经网络断层预测结果与地震数据吻合较好,并且在断层细节刻画上要优于传统地震相干体属性方法。      

基于深度卷积神经网络的地震数据随机噪声压制

本文以提高地震数据的成像质量为目标,提出一种智能的卷积神经网络降噪框架,从带有噪声的地震数据中自适应地学习地震信号。为了加速网络训练和避免训练时出现梯度消失现象,我们在网络中加入残差学习和批标准化的方法,并采用了ReLU激活函数和Adam优化算法优化网络。此外,Marmousi和F3数据集被用来对网络进行训练和测试,经过充分训练的网络不仅能在学习中保留地震数据特征,而且能去除随机噪声。首先充分地训练网络,从中提取出随机噪声,并保留学习到的地震数据特征,之后通过重建地震数据估算测试集中的波形特征。合成记录和实际数据的处理结果显示了深度卷积神经网络在随机噪声压制任务中的潜力,并通过实验验证表明了深度卷积神经网络框架有很好的去噪效果。      

地震应急信息自动分类方法研究

地震应急信息的高效处理为地震应急救援工作提供了重要支撑。本文根据地震应急信息分类的需求,构建了一种高效便捷的地震信息分类处理方法。以震前、震时、震后为时间主线,将地震应急信息分为震前基础背景信息、地震震情灾情信息及震后应急救援信息,并采用“关键词分类”的方法,在计算机语言的支持下,将多渠道汇集的应急信息进行自动分类,在一定程度上缩短了应急信息加工处理与服务的时间,能快速高效地为应急指挥提供信息服务。     

Shear wave velocity prediction using seismic attributes and well log data

Formation’s properties can be estimated indirectly using joint analysis of compressional and shear wave velocities. Shear wave data is not usually acquired during well logging, which is most likely for cost saving purposes. Even if shear data is available, the logging programs provide only sparsely sampled one-dimensional measurements: this information is inadequate to estimate reservoir rock properties. Thus, if the shear wave data can be obtained using seismic methods, the results can be used across the field to estimate reservoir properties. The aim of this paper is to use seismic attributes for prediction of shear wave velocity in a field located in southern part of Iran. Independent component analysis (ICA) was used to select the most relevant attributes to shear velocity data. Considering the nonlinear relationship between seismic attributes and shear wave velocity, multi-layer feed forward neural network was used for prediction of shear wave velocity and promising results were presented.     

A comparative study of three neural network forecast combination methods for simulated river flows of different rainfall—runoff models

Abstract

The performances of three artificial neural network (NN) methods for combining simulated river flows, based on three different neural network structures, are compared. These network structures are: the simple neural network (SNN), the radial basis function neural network (RBFNN) and the multi-layer perceptron neural network (MLPNN). Daily data of eight catchments, located in different parts of the world, and having different hydrological and climatic conditions, are used to enable comparisons of the performances of these three methods to be made. In the case of each catchment, each neural network combination method synchronously uses the simulated river flows of four rainfall—runoff models operating in design non-updating mode to produce the combined river flows. Two of these four models are black-box, the other two being conceptual models. The results of the study show that the performances of all three combination methods are, on average, better than that of the best individual rainfall—runoff model utilized in the combination, i.e. that the combination concept works. In terms of the Nash-Sutcliffe model efficiency index, the MLPNN combination method generally performs better than the other two combination methods tested. For most of the catchments, the differences in the efficiency index values of the SNN and the RBFNN combination methods are not significant but, on average, the SNN form performs marginally better than the more complex RBFNN alternative. Based on the results obtained for the three NN combination methods, the use of the multi-layer perceptron neural network (MLPNN) is recommended as the appropriate NN form for use in the context of combining simulated river flows.     

基于模糊粗糙集的机器学习储层参数预测

因为地震数据的三维空间分布优势,地震属性已经被广泛应用于含油气性预测、储层厚度预测、孔隙度预测等。但也存在地震属性之间信息冗余、属性与储层物性参数关系模糊的问题。针对这两个问题,将模糊粗糙理论和机器学习引入到储层参数预测中来。通过模糊粗糙集理论对地震属性进行约简,去除冗余信息,得到最优化的地震属性组合;将约简后的属性作为机器学习的输入,实现从地震属性到储层物性参数的非线性映射。该方法既保留了地震属性中有效信息,又避免了因输入变量过多而导致的网络模型训练困难。实际数据应用表明,属性约简的机器学习预测结果分辨率更高,并与数据吻合更好。      

全卷积神经网络在建筑物震害遥感提取中的应用研究

为解决建筑物震害信息提取自动化程度不高的问题,本文将全卷积神经网络应用于建筑物震害遥感信息提取。以玉树地震后获取的玉树县城区0.2m分辨率航空影像作为建筑物震害信息提取试验数据源,将试验区地物划分为倒塌建筑物、未倒塌建筑物和背景3类。对427个500×500像素的子影像进行人工分类与标注,选取393个组成训练样本集,34个用于验证。利用训练样本集对全卷积神经网络进行训练,采用训练后的网络对验证样本进行建筑物震害信息提取及精度评价。研究结果表明:建筑物震害遥感信息提取总体分类精度为82.3%,全卷积神经网络方法能提高信息提取自动化程度,具有较好的建筑物震害信息提取能力。     

Seismic modelling for reservoir studies: a comparison between convolutional and full-waveform methods for a deep-water turbidite sandstone reservoir

Two seismic modelling approaches, that is, two-dimensional pre-stack elastic finite-difference and one-dimensional convolution methods, are compared in a modelling exercise over the fluid-flow simulation model of a producing deep-water turbidite sandstone reservoir in the West of Shetland Basin. If the appropriate parameterization for one-dimensional convolution is used, the differences in three-dimensional and four-dimensional seismic responses from the two methods are negligible. The key parameters to ensure an accurate seismic response are a representative wavelet, the distribution of common-depth points and their associated angles of incidence. Conventional seismic images generated by the one-dimensional convolutional model suffer from lack of continuity because it only accounts for vertical resolution. After application of a lateral resolution function, the convolutional and finite-difference seismic images are very similar. Although transmission effects, internal multiples and P-to-S conversions are not included in our convolutional modelling, the subtle differences between images from the two methods indicates that such effects are of secondary nature in our study. A quantitative comparison of the (normalized root-mean-square) amplitude attributes and waveform kinematics indicates that the finite-difference approach does not offer any tangible benefit in our target-oriented seismic modelling case study, and the potential errors from one-dimensional convolution modelling are comparatively much smaller than the production-induced time-lapse changes.     

Acquisition design for direct reflectivity and velocity estimation from blended and irregularly sampled data

Blended acquisition along with efficient spatial sampling is capable of providing high-quality seismic data in a cost-effective and productive manner. While deblending and data reconstruction conventionally accompany this way of data acquisition, the recorded data can be processed directly to estimate subsurface properties. We establish a workflow to design survey parameters that account for the source blending as well as the spatial sampling of sources and detectors. The proposed method involves an iterative scheme to derive the survey design leading to optimum reflectivity and velocity estimation via joint migration inversion. In the workflow, we extend the standard implementation of joint migration inversion to cope with the data acquired in a blended fashion along with irregular detector and source geometries. This makes a direct estimation of reflectivity and velocity models feasible without the need of deblending or data reconstruction. During the iterations, the errors in reflectivity and velocity estimates are used to update the survey parameters by integrating a genetic algorithm and a convolutional neural network. Bio-inspired operators enable the simultaneous update of the blending and sampling operators. To relate the choice of survey parameters to the performance of joint migration inversion, we utilize a convolutional neural network. The applied network architecture discards suboptimal solutions among newly generated ones. Conversely, it carries optimal ones to the subsequent step, which improves the efficiency of the proposed approach. The resultant acquisition scenario yields a notable enhancement in both reflectivity and velocity estimation attributable to the choice of survey parameters.     

多波地震深度学习的油气储层分布预测案例

有机并有效利用纵波与转换横波在油气储层敏感度上存在的差异,有助于突出地震油气储层特征,有助于提高地震油气储层分布边界刻画的精度.基于此,本文设计了一种卷积神经网络与支持向量机方法相结合的多波地震油气储层分布预测的深度学习法（Deep Learning Method）.首先,利用莱特准则剔除所生成的多波地震属性中可能存在的异常值降低网络变体数量.然后,通过能突出多波地震油气储层特征的聚类算法和无监督学习算法构建隐藏层,用于增加网络共享,提取油气特征.最后,将增加网络罚值后的井点样本作为支持向量机预测的输入样本,以降采样后的C3卷积层属性作为学习集,进行从已知到未知的地震油气储层的预测.本方案应用于HG地区晚三叠统HGR组的碳酸盐岩油气储层预测,所预测的地震油气储层边界更加清晰,预测结果与实际情况基本吻合.应用结果表明：本论文方案不仅具有可行性,且具有有效性.     

Estimation of groundwater storage from seismic data using deep learning

Convolutional neural networks can provide a potential framework to characterize groundwater storage from seismic data. Estimation of key components, such as the amount of groundwater stored in an aquifer and delineate water table level, from active-source seismic data are performed in this study. The data to train, validate and test the neural networks are obtained by solving wave propagation in a coupled poroviscoelastic–elastic media. A discontinuous Galerkin method is applied to model wave propagation, whereas a deep convolutional neural network is used for the parameter estimation problem. In the numerical experiment, the primary unknowns estimated are the amount of stored groundwater and water table level, while the remaining parameters, assumed to be of less of interest, are marginalized in the convolutional neural network-based solution. Results, obtained through synthetic data, illustrate the potential of deep learning methods to extract additional aquifer information from seismic data, which otherwise would be impossible based on a set of reflection seismic sections or velocity tomograms.     

粗集理论在地震储层预测中的应用

在地震储层预测中,可采用的地震属性种类繁多,但太多地震属性常常会起到干扰作用,影响储层的预测精度,因此,为提高地震储层预测精度,把粗糙集理论融入到地震属性的优化中,利用粗糙集理论所具有的提取有用属性、简化信息处理的能力,优选出地震属性中的敏感属性是本文的研究目的,本文采用了一种基于属性方差的自组织神经网络量化方法,并运用基于区别矩阵的属性频率约简算法对地震属性进行优选,实例分析表明:该方法可行有效,可以最大限度地删除冗余地震属性,用优选出的敏感属性组合对多种储层参数进行预测均已取得了较好的效果.     

Detection of seismic reflections from seismic attributes through fractal analysis

The concept of fractals is used here for the identification of seismic reflectors with special emphasis on thin‐bed delineation, which is generally overlooked during standard data processing. A new fractal analysis scheme is applied to both synthetic and real field seismic data. The fractal dimensions of the three seismic attributes – amplitude, phase, and instantaneous frequency – have been analysed and evaluated. A change in fractal dimension is found to occur whenever there is a reflection. However, the resolution in the delineation of reflectors varies, depending on the attribute under consideration and the method of fractal dimension estimation used. Fractal analysis is performed on both noise‐free and noisy synthetic data to establish the noise tolerance limit for both the ‘divider method’ and the ‘Hurst method’. It is then tested with different peak frequencies of the source wavelet to establish the criteria for using the divider method and the Hurst method. The divider method is found to be suitable for high peak frequency source wavelets (> 25 Hz), while the Hurst method is best suited for low peak frequency source wavelets (< 25 Hz). Finally, when applied to the digitally processed and migrated field seismic data, it could even delineate reflectors which otherwise went undetected on the migrated time section.     

Short-term prediction of influent flow in wastewater treatment plant

Predicting influent flow is important in the management of a wastewater treatment plant (WWTP). Because influent flow includes municipal sewage and rainfall runoff, it exhibits nonlinear spatial and temporal behavior and therefore makes it difficult to model. In this paper, a neural network approach is used to predict influent flow in the WWTP. The model inputs include historical influent data collected at a local WWTP, rainfall data and radar reflectivity data collected by the local weather station. A static multi-layer perceptron neural network performs well for the current time prediction but a time lag occurs and increases with the time horizon. A dynamic neural network with an online corrector is proposed to solve the time lag problem and increase the prediction accuracy for longer time horizons. The computational results show that the proposed neural network accurately predicts the influent flow for time horizons up to 300 min.     

基于卷积神经网络的地震波形智能识别研究

随着世界上多个国家和地区的地震预警系统投入运行,误报和漏报等问题逐渐突显,特别是将标定以及强干扰波形误识别为大震事件,快速、精确地区分地震与其他波形是一个难题。针对于此,该研究提出了基于卷积神经网络地震波形智能识别方法。首先收集并处理了2012—2017年中国境内福建以及周边邻省共683个地震和478个爆破事件,并对这些样本筛选、截取和基线校正等预处理,共得到了27 500条三通道波形。在此基础上,构建了3 s波形输入的卷积神经网络模型(SW-CNN)。结果表明：模型对地震、噪声、爆破和异常波形的识别率分别为97.9、99、99.2和99.3%。相比于人工手动分类识别,该模型更省时和更稳定,为地震预警目前所面临的问题提供了一个新的解决方法。     

基于深度学习的地震震级分类

为了探索地震加速度时程记录的震级信息,训练卷积神经网络基于地震震级大小对地震记录进行分类,将K-NET和KiK-net中将近12万个地震记录作为样本,对其进行信息筛选和归一化,之后将地震加速度时程记录用作输入,训练卷积神经网络模型以M5.5为分类界限来区分大震和小震。结果显示,在训练集中基于该模型的分类准确率为93.6％,在测试集中的准确率为92.3％,具有良好的分类效果,这表明大震记录与小震记录之间存在一些根本的区别,即可通过地震动加速度时程记录获取一定的震级信息。