基于相关积分的径向时频峰值滤波压制地震勘探随机噪声
发布时间：2019-04-13 16:22
【摘要】：地震勘探是目前寻找石油及天然气的主要方法,地震专家通过地震勘探资料可以识别地下岩层的地质结构,从而推断油气煤田的地下位置,为石油、天然气的开采提供有力的查找依据。地震勘探分成三个部分：地震资料采集、地震资料处理、地震资料解释。其中,地震资料处理是地震勘探的重要部分,地震资料处理为下一步地震资料解释提供清晰可辩的地震资料。在地震资料处理中如何获得高信噪比的地震资料显得尤为重要,地震勘探资料的去噪方法研究已成为地震勘探领域的一个研究热点。时频峰值滤波算法是基于时频分析的理论来消减随机噪声。金雷等将其引用到地震勘探记录中压制随机噪声,并取得了较好的效果,时频峰值滤波不依赖于信号的先验信息,可以恢复出湮没在较低信噪比噪声的地震信号,提高地震勘探资料的信噪比。然而传统的时频峰值滤波采用的是固定窗长,未能兼顾随机噪声压制与有效信号幅值的保持；时频峰值滤波只是沿时间方向上进行滤波,是一维的时频滤波算法,忽略了地震勘探资料空间中道与道之间信息,不能体现地震资料道与道之间的相关性,吴宁等引入径向时频峰值滤波算法,将地震资料通过径向道变换,在径向域进行时频峰值滤波,然后将滤波后的记录进行径向道反变换。径向道变换能将有效信号进行拉伸,在径向道域进行时频峰值滤波可选取较大的滤波窗长,使有效信号在窗内尽可能是线性函数,从而选取较大的窗长进行滤波,不影响信号幅值的保持,大窗长滤波更有利于随机噪声的压制,改善了传统时频峰值滤波的不足,但是径向时频峰值滤波采用的是固定角度的采样方式,在采样轨线与同相轴夹角过大的地方,径向道变换对信号的拉伸不明显,径向道变换会产生畸变,如果仍选用较大的窗长,地震记录幅值会出现很大的衰减,使滤波出现误差。本文提出基于相关积分的径向时频峰值滤波。在径向道域中求出每道序列的相关积分值,利用噪声序列与信号序列的相关积分值的不同,区分出含有有效信号段和噪声段,选取大窗长对噪声部分进行时频峰值滤波,选取小窗长对信号部分进行时频峰值滤波,将滤波后的记录进行径向道反变换。处理模拟地震记录和实际地震记录表明本文方法能有效提高了有效信号的幅值,噪声也得到了很好的压制,有效提高地震勘探资料的信噪比。
[Abstract]:Seismic exploration is the main method to search for oil and natural gas at present. Seismic experts can identify the geological structure of underground strata by seismic exploration data, and then infer the underground position of oil and gas coalfield, which is petroleum. The exploitation of natural gas provides a powerful basis for searching. Seismic exploration is divided into three parts: seismic data acquisition, seismic data processing, seismic data interpretation. Seismic data processing is an important part of seismic exploration, and seismic data processing provides clear and recognizable seismic data for the next step of seismic data interpretation. How to obtain the seismic data with high signal-to-noise ratio in seismic data processing is particularly important. The research on denoising methods of seismic exploration data has become a hot spot in the field of seismic exploration. Time-frequency peak filtering algorithm is based on the theory of time-frequency analysis to reduce random noise. Jin Lei et al used it to suppress random noise in seismic exploration records, and achieved good results. The time-frequency peak filtering does not depend on the prior information of the signal, and can recover the seismic signal which is annihilated in the lower SNR noise, and the time-frequency peak filtering does not depend on the prior information of the signal. Improve the signal-to-noise ratio of seismic exploration data. However, the traditional time-frequency peak filter uses a fixed window length, which fails to keep the amplitude of the effective signal and the random noise suppression. Time-frequency peak filtering is a one-dimensional time-frequency filtering algorithm, which only filters along the time direction. It ignores the information between trace and track in seismic exploration data space, and can not reflect the correlation between seismic data track and track. Wu Ning et al introduced the radial time-frequency peak filtering algorithm. The seismic data were filtered by radial channel transform and time-frequency peak filtering in radial domain. Then the filtered records were transformed by radial channel inverse transform. Radial channel transform can stretch the effective signal, and the filter window length can be selected by the time-frequency peak filtering in the radial channel domain, so that the effective signal can be a linear function in the window as far as possible, thus the larger window length can be selected to filter. The large window length filter is more favorable to suppress random noise and improve the deficiency of traditional time-frequency peak filtering, but the radial time-frequency peak filter adopts fixed-angle sampling mode, which does not affect the maintenance of signal amplitude, and large window length filter is more conducive to the suppression of random noise. Where the angle between the sampling trajectory and the in-phase axis is too large, the radial channel transformation does not stretch the signal obviously, and the radial channel transformation will cause distortion. If the window length is still large, the amplitude of the seismic record will be greatly attenuated and the filtering error will occur. In this paper, a radial time-frequency peak filter based on correlation integral is proposed. The correlation integral values of each channel sequence are obtained in the radial channel domain. By using the difference between the correlation integral values of the noise sequence and the signal sequence, the effective signal segment and the noise segment are distinguished, and the noise part is filtered by the time-frequency peak value of the noise part with large window length. Small window length is selected to filter the time-frequency peak value of the signal, and the filtered record is transformed by the radial channel inverse transform. The processing of simulated seismic records and actual seismic records shows that the proposed method can effectively increase the amplitude of effective signals, suppress the noise well, and effectively improve the signal-to-noise ratio of seismic exploration data.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.4
本文编号：2457741
[Abstract]:Seismic exploration is the main method to search for oil and natural gas at present. Seismic experts can identify the geological structure of underground strata by seismic exploration data, and then infer the underground position of oil and gas coalfield, which is petroleum. The exploitation of natural gas provides a powerful basis for searching. Seismic exploration is divided into three parts: seismic data acquisition, seismic data processing, seismic data interpretation. Seismic data processing is an important part of seismic exploration, and seismic data processing provides clear and recognizable seismic data for the next step of seismic data interpretation. How to obtain the seismic data with high signal-to-noise ratio in seismic data processing is particularly important. The research on denoising methods of seismic exploration data has become a hot spot in the field of seismic exploration. Time-frequency peak filtering algorithm is based on the theory of time-frequency analysis to reduce random noise. Jin Lei et al used it to suppress random noise in seismic exploration records, and achieved good results. The time-frequency peak filtering does not depend on the prior information of the signal, and can recover the seismic signal which is annihilated in the lower SNR noise, and the time-frequency peak filtering does not depend on the prior information of the signal. Improve the signal-to-noise ratio of seismic exploration data. However, the traditional time-frequency peak filter uses a fixed window length, which fails to keep the amplitude of the effective signal and the random noise suppression. Time-frequency peak filtering is a one-dimensional time-frequency filtering algorithm, which only filters along the time direction. It ignores the information between trace and track in seismic exploration data space, and can not reflect the correlation between seismic data track and track. Wu Ning et al introduced the radial time-frequency peak filtering algorithm. The seismic data were filtered by radial channel transform and time-frequency peak filtering in radial domain. Then the filtered records were transformed by radial channel inverse transform. Radial channel transform can stretch the effective signal, and the filter window length can be selected by the time-frequency peak filtering in the radial channel domain, so that the effective signal can be a linear function in the window as far as possible, thus the larger window length can be selected to filter. The large window length filter is more favorable to suppress random noise and improve the deficiency of traditional time-frequency peak filtering, but the radial time-frequency peak filter adopts fixed-angle sampling mode, which does not affect the maintenance of signal amplitude, and large window length filter is more conducive to the suppression of random noise. Where the angle between the sampling trajectory and the in-phase axis is too large, the radial channel transformation does not stretch the signal obviously, and the radial channel transformation will cause distortion. If the window length is still large, the amplitude of the seismic record will be greatly attenuated and the filtering error will occur. In this paper, a radial time-frequency peak filter based on correlation integral is proposed. The correlation integral values of each channel sequence are obtained in the radial channel domain. By using the difference between the correlation integral values of the noise sequence and the signal sequence, the effective signal segment and the noise segment are distinguished, and the noise part is filtered by the time-frequency peak value of the noise part with large window length. Small window length is selected to filter the time-frequency peak value of the signal, and the filtered record is transformed by the radial channel inverse transform. The processing of simulated seismic records and actual seismic records shows that the proposed method can effectively increase the amplitude of effective signals, suppress the noise well, and effectively improve the signal-to-noise ratio of seismic exploration data.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.4
【参考文献】
相关期刊论文 前2条
1 金雷,李月,杨宝俊;用时频峰值滤波方法消减地震勘探资料中随机噪声的初步研究[J];地球物理学进展;2005年03期
2 刘喜武;张宁;勾永峰;李华;;地震勘探信号时频分析方法对比与应用分析[J];地球物理学进展;2008年03期
，本文编号：2457741
本文链接：https://www.wllwen.com/kejilunwen/diqiudizhi/2457741.html
