海洋多震源混合采集高保真数据分离研究

发布时间：2018-03-08 04:30

本文选题：多震源混合采集　切入点：加权多级中值滤波　出处：《吉林大学》2017年硕士论文　论文类型：学位论文

【摘要】：多震源混合采集技术作为一项高效率高密度的地震采集革新技术,对提高成像质量、节省勘探施工成本以及增强地下介质照明度起到了突出的推动作用,但受制于海上震源激发装备条件和勘探船开采成本等因素,海洋多震源混合采集的工业化进程仍处于初步阶段。为加快完成多震源混合采集从基础理论研究到实际采集的过渡,文章基于前人研究基础,针对海洋多震源混合采集技术开展了系统研究。本文主要阐释了海洋多震源混合采集的方法原理、混采数据分离方法以及缺失道集重构方法,针对目前常规勘探中存在的照明度不够、缺失道集等问题,从多震源混合采集系统设计、稀疏变换重构地震数据等不同角度相应地提出了解决方案。通过模拟算例验证了方法的可行性,为未来海洋混合震源采集技术的商业化发展奠定了理论基础。本文通过分析比较常规单震源采集和多震源混合采集,对多震源混合采集的基本理论和数学模型进行了详细解释。用图例表达出多震源混合采集SDR、STR两个关键参数和混合度的概念,通过引入混合震源算子概念,根据不同混合震源算子将多震源混合采集分为:“完全欠定型多震源混合采集”和“非完全欠定型多震源混合采集”,并对与之相应多震源混合采集数据进行了分离处理和研究。基于多震源混合采集观测系统中子震源间距不同,分别对两种海洋多震源混合采集观测系统:“海洋多震源混合采集多船观测系统”、“海洋多震源混合采集单船观测系统”的正演模拟进行了详细介绍,强调了不同观测系统的采集优点。通过延迟时间激发顺序的不同,介绍了四种不同类型的编码设计形式。由时间域多震源混合采集数据分离的角度出发,阐释了多震源混合采集中“伪分离”这一基础概念。在改进多级中值滤波方法的基础上,本文提出基于加权多级中值滤波迭代的多震源混合采集数据分离方法。将时间域混采数据伪分离记录转换至F-K域进行信噪分析,设计不同的选择窗口进行噪声阈值滤除,改善了F-K滤波无法完全压制混合采集数据中混叠噪声的缺点,提高了分离数据保真度。本文在设计F-K选择窗口的基础之上,引入一个权系数,将原滤波方法改进成加权多级中值滤波,并设计了迭代算法,在震源混合度较高的情况下进行了模型计算,与常规多震源地震混采数据分离技术相比具有更高的分离效率和更高质量的分离信噪比。由于在地震采集过程中存在“残道、坏道”的现象,为了使多震源混合采集分离记录具有较高的保真度,须在波场数据分离的同时对缺失道集数据进行重新构建。鉴于此,本文引入用于地震数据重构的Shearlet稀疏基,设计了一种在压缩感知框架下基于Shearlet变换(剪切波变换)的混合采集数据分离和地震数据重构方法。该方法首先对缺失的多震源混合采集数据进行时域变换,经预处理后通过Shearlet变换得到相应的稀疏表示,引入快速迭代收缩阈值算法实现Shearlet域中稀疏系数的恢复,最后对稀疏系数进行Shearlet反变换实现地震缺失道集数据的重构和混采数据的分离。利用Shearlet变换的最佳稀疏逼近性质在分离残留噪声的同时最大程度地保留有效信号,并设计迭代算法不断优化分离结果。通过模拟数据结果验证得出:与其他常规混采数据分离方法相比,该方法具有更佳的分离结果和计算效率。
[Abstract]:Multi source mixed acquisition technology is a high efficiency and high density seismic acquisition technology innovation, to improve the image quality, save construction cost and enhance the exploration of underground medium illumination has played a prominent role in promoting, but subject to the sea source equipment conditions and exploration ship mining costs and other factors, the marine multi source mixed collection of industrialization is still in the preliminary stage. Multi source mixed acquisition from basic theory research to the actual acquisition in order to speed up the transition, the article based on the basis of previous research, the hybrid acquisition technology of ocean source are researched. This paper mainly explains the principle and method of multi source marine mixed collection, mixing and deletion gathers data reconstruction method of separation method at present, the conventional exploration in lighting is not enough, the problem missing trace set, from the design of hybrid acquisition system of multiple sources, sparse Different angle transform to reconstruct seismic data corresponding solution is put forward. Simulation results verify the feasibility of the method, which laid a theoretical foundation for the future development of commercial marine hybrid source acquisition technology. By comparing the conventional single source acquisition and multi source mixed acquisition, a detailed explanation of the basic theory and mathematical model multi source mixed collection. Expression of multiple sources with mixed acquisition SDR legend, STR two key parameters and the mixing degree concept, through the introduction of hybrid source operator concept, according to the different mixed source operator multi source mixed collection is divided into: "no stereotypes of multiple sources of mixed collection" and "non completely under stereotypes of multiple sources mixed acquisition", and the corresponding multi source data collection and the mixture were separated and processing and research. Multi source mixed acquisition system based on the distance between the neutron source Different of the two kinds of ocean source hybrid acquisition system: "ocean source mixed acquisition observation system" ships "forward ocean source mixed acquisition observation system" single ship simulation is introduced in detail, emphasize different advantages. Through the observation system acquisition delay time excitation sequence, the four different types of encoding form design. By the time domain separation of multiple source mixed data acquisition perspective, explains the multiple sources of mixed collection in "pseudo separation" of the basic concept. Based on improved multilevel median filtering method, this paper proposes to separate multiple sources of data collection mixed weighted multistage median filtering based on iteration the time domain method. The mixed noise analysis of mining data records a pseudo separate conversion to F-K domain, the selection window design for different noise threshold filtering, F-K filtering to improve Completely mixed and pressed shortcomings of aliasing noise in acquisition data, improve the separation of data fidelity. Based on the selection window in the design of F-K, by introducing a weight coefficient, the improved filtering method as a weighted multistage median filtering, and the iterative algorithm for the design of a high degree of mixing in the source under the condition of model calculation. With the conventional multi source seismic data compared with the mixed mining separation technology has higher separation efficiency and higher quality separation signal-to-noise ratio. Due to the existence of residual in the seismic acquisition process, bad sectors "phenomenon, in order to make the multi source mixed collection and separation of records with high fidelity, must be in the separation of wave field data at the same time to gather data to reconstruct the missing. In view of this, this paper introduced Shearlet based sparse seismic data reconstruction, a Shearlet transform based on compressed sensing frame design (shear wave For the data acquisition and separation) hybrid seismic data reconstruction method. Firstly, the lack of multi source mixed data domain transform, after the pretreatment by Shearlet transform to obtain the corresponding sparse representation, the fast iterative shrinkage thresholding algorithm for sparse coefficient in Shearlet domain recovery, the sparse coefficient inverse transform Shearlet the lack of seismic gathers data reconstruction and mixed mining data separation. Using Shearlet transform optimal sparse approximation properties in the separation of residual noise and maximize the retention of the effective signal, and design iterative algorithm to optimize the separation results. The simulation results verify that the data and other data mining method compared to conventional coagulation separation, separation this method has better results and calculation efficiency.

【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P631.46

【相似文献】