主动源实验中重复激发的弱信号处理方法研究

发布时间：2018-05-05 21:32

本文选题：信号处理 + 主动源实验　；参考：《中国科学技术大学》2017年硕士论文

【摘要】：从低信噪比的重复激发地震记录中提取微弱的有效信号是利用主动源实验数据研究地下结构的第一步,提取信号的准确性直接影响着后续数据处理的可靠性。本文探究并评估了叠加类(线性叠加、相似性加权叠加、时间域相位加权叠加、时频域相位加权叠加、改进的时频域相位加权叠加)和直接分离类(奇异值分解法、盲源分离法)方法提取弱信号的效果,并将以上方法用于解决主动源实验数据弱信号处理问题中。本文首先选用可控震源互相关子波和气枪震源子波构建出多道简单合成波形用于探究上述多种方法提取弱信号能力,通过定量分析各处理结果的走时、波形、频率成分等方面变化程度从而对各方法的好坏进行了较为全面的评估。然后,本文使用一维速度模型及实际噪声水平构建两种震源的人工合成地震剖面数据,仿照真实数据处理流程并结合合成测试的理论分析结果探究出适合实际资料处理的方法和流程。最后,通过主动源实验的实际数据来验证本文研究结果。合成测试和实际数据处理的结果表明:(1)当人工震源重复激发次数较少时,使用非线性叠加方法提取弱信号得到的结果优于线性叠加;当人工震源重复激发次数较多时,宜使用线性叠加方法提取弱信号,且可降低提取方法对有效信号波形等信息的影响;(2)零相位化处理可降低对最小相位子波信号提取过程中走时和波形的影响;(3)可控震源处理中,对原始数据处理完后需进行互相关处理来压缩子波避免不同震相间波形叠加后再提取有效信号;气枪震源数据预处理后直接叠加,可造成提取结果中震相到时信息破坏。本文提出对预处理后数据进行零相位化处理后再进行提取,有利于降低噪声、处理方法等对走时和波形的影响,将两种处理流程结果相结合亦可为震相到时拾取提供参考。
[Abstract]:It is the first step to study underground structure by using active source experimental data to extract weak effective signals from repeated excitation seismic records with low signal-to-noise ratio (SNR). The accuracy of extracting signals directly affects the reliability of subsequent data processing. In this paper, we investigate and evaluate the superposition classes (linear superposition, similarity weighted superposition, time-domain phase-weighted superposition, time-frequency phase-weighted superposition, improved time-frequency-domain phase-weighted superposition) and direct separation (singular value decomposition method). Blind source separation (BSS) method is used to solve the weak signal processing problem of active source experiment data. In this paper, the vibroseis cross-correlation wavelet and air-gun source wavelet are used to construct multi-channel simple synthetic waveforms to explore the ability of extracting weak signals from the above methods, and the waveforms of each processing result are quantitatively analyzed. Frequency components and other aspects of the degree of change, so as to evaluate the quality of each method is relatively comprehensive. Then, using the one-dimensional velocity model and the actual noise level, the synthetic seismic profile data of two kinds of seismic sources are constructed. According to the real data processing flow and the theoretical analysis results of synthetic test, the method and process suitable for practical data processing are explored. Finally, the results of this paper are verified by the actual data of the active source experiment. The result of synthetic test and actual data processing shows that when the number of repeated excitation of artificial source is less, the result of extracting weak signal by nonlinear superposition method is better than that of linear stack, and when the number of repeated excitation of artificial source is more than that of linear stack, It is advisable to use linear superposition method to extract weak signals, and to reduce the influence of extraction methods on the information such as effective signal waveforms. The zero phase processing can reduce the travel time and the influence of waveforms on the minimum phase wavelet signal extraction. After processing the original data, cross-correlation is needed to compress the wavelet to avoid the superposition of the waveforms between different seismic phases and then extract the effective signal. After the air gun source data is preprocessed and superimposed directly, the phase arrival information in the extraction result can be destroyed. In this paper, it is proposed that the pre-processing data can be extracted after zero phase processing, which is helpful to reduce the noise and the effect of the processing methods on the travel time and waveform. The combination of the two processing flow results can also provide a reference for the seismic phase arrival pickup.
【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P631.4

【相似文献】