基于压缩感知理论的地震波场数据重建
发布时间：2018-08-13 14:54
【摘要】：随着勘探过程中采集条件的限制以及后续工作处理中对废道废炮的剔除,不可避免的使地震波场数据存在不完整性,对地震资料的后续处理产生一定的影响,因此有必要对地震波场数据进行重建。但是,受奈奎斯特(Nyqusit)采样定理的限制,传统的地震信号重建方法对采样率有较高的要求。压缩感知(Compressive/Compressed sensing,CS)理论主要包括信号的稀疏化处理、测量矩阵的构造以及重建算法三个部分。重建算法是CS理论的核心内容,即通过低维的稀疏信号精确的恢复出原始信号。本文基于CS理论对其重建算法进行了深入研究。选取正交匹配追踪算法(Orthogonal Matching Pursuit,OMP)作为重建算法,并对OMP的实现原理、算法结构等进行具体的研究同时做出信号仿真。针对OMP算法的重构速度比较慢和迭代次数需要给定的问题,发展了一种改进方案,将最优匹配追踪算法(Optimized OMP,OOMP)的约束最优匹配原子选择策略和自适应变步长梯度投影(Adaptive Variance Step Gradient Projection method,AVSGP)算法的后向投影思想与其进行结合。仿真实验证明,在稀疏条件同等的情况下,改进的OMP算法在重构时间及效果方面明显优于传统OMP算法。针对重建地震波场数据中混有随机噪声的情况,本文选取曲波阈值法进行去噪,对于处理大规模数据的情况引入分块思想及二次曲波变换对曲波阈值进行改进。同时与CS理论相结合,可以重构出去噪后的地震信号。通过仿真实验证明,在稀疏条件同等的条件下,改进的曲波阈值法与原始曲波阈值法相比在去噪方面有明显提高。最后,对本文的研究工作进行总结,通过对仿真模型的建立验证出方法的有效及实用性。
[Abstract]:With the limitation of acquisition conditions in the process of exploration and the elimination of abandoned road guns in the subsequent work, the seismic wave field data will inevitably be incomplete, which will have a certain impact on the subsequent processing of seismic data. Therefore, it is necessary to reconstruct seismic wave field data. However, due to the limitation of Nyquist (Nyqusit) sampling theorem, the traditional seismic signal reconstruction method requires high sampling rate. The theory of Compressive/Compressed sensing CS mainly includes three parts: signal sparse processing, measurement matrix construction and reconstruction algorithm. The reconstruction algorithm is the core of CS theory, that is to say, the original signal can be recovered accurately by low dimensional sparse signal. In this paper, the reconstruction algorithm based on CS theory is studied. The orthogonal matching tracking algorithm (Orthogonal Matching pursuit OMP) is selected as the reconstruction algorithm, and the realization principle and algorithm structure of OMP are studied in detail and the signal simulation is made at the same time. Aiming at the problem that the reconstruction speed of OMP algorithm is slow and the number of iterations need to be given, an improved scheme is developed. The constrained optimal matching atomic selection strategy of the optimal matching tracking algorithm (Optimized OMP) and the backward projection idea of the adaptive variable step size gradient projection (Adaptive Variance Step Gradient Projection method AVSGP algorithm are combined. Simulation results show that the improved OMP algorithm is superior to the traditional OMP algorithm in reconstruction time and effect under the same sparse condition. In order to solve the problem of mixed random noise in reconstructed seismic wave field data, this paper selects Qu Bo threshold method for denoising, and introduces the block idea and the quadratic Qu Bo transform to improve the Qu Bo threshold when dealing with large scale data. At the same time, combined with CS theory, the seismic signal after noise can be reconstructed. The simulation results show that the improved Qu Bo threshold method is better than the original Qu Bo threshold method in denoising under the same sparse condition. Finally, the research work of this paper is summarized, and the effectiveness and practicability of the method are verified by the establishment of simulation model.
【学位授予单位】：东北石油大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.4
本文编号：2181318
[Abstract]:With the limitation of acquisition conditions in the process of exploration and the elimination of abandoned road guns in the subsequent work, the seismic wave field data will inevitably be incomplete, which will have a certain impact on the subsequent processing of seismic data. Therefore, it is necessary to reconstruct seismic wave field data. However, due to the limitation of Nyquist (Nyqusit) sampling theorem, the traditional seismic signal reconstruction method requires high sampling rate. The theory of Compressive/Compressed sensing CS mainly includes three parts: signal sparse processing, measurement matrix construction and reconstruction algorithm. The reconstruction algorithm is the core of CS theory, that is to say, the original signal can be recovered accurately by low dimensional sparse signal. In this paper, the reconstruction algorithm based on CS theory is studied. The orthogonal matching tracking algorithm (Orthogonal Matching pursuit OMP) is selected as the reconstruction algorithm, and the realization principle and algorithm structure of OMP are studied in detail and the signal simulation is made at the same time. Aiming at the problem that the reconstruction speed of OMP algorithm is slow and the number of iterations need to be given, an improved scheme is developed. The constrained optimal matching atomic selection strategy of the optimal matching tracking algorithm (Optimized OMP) and the backward projection idea of the adaptive variable step size gradient projection (Adaptive Variance Step Gradient Projection method AVSGP algorithm are combined. Simulation results show that the improved OMP algorithm is superior to the traditional OMP algorithm in reconstruction time and effect under the same sparse condition. In order to solve the problem of mixed random noise in reconstructed seismic wave field data, this paper selects Qu Bo threshold method for denoising, and introduces the block idea and the quadratic Qu Bo transform to improve the Qu Bo threshold when dealing with large scale data. At the same time, combined with CS theory, the seismic signal after noise can be reconstructed. The simulation results show that the improved Qu Bo threshold method is better than the original Qu Bo threshold method in denoising under the same sparse condition. Finally, the research work of this paper is summarized, and the effectiveness and practicability of the method are verified by the establishment of simulation model.
【学位授予单位】：东北石油大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.4
【参考文献】
相关硕士学位论文 前1条
1 王培茂;地震勘探数据压缩相关技术研究及应用[D];吉林大学;2004年
，本文编号：2181318
本文链接：https://www.wllwen.com/kejilunwen/diqiudizhi/2181318.html
