基于结构导向TFPF算法的地震勘探随机噪声压制

发布时间：2018-10-30 13:51

【摘要】：石油、天然气在国民经济发展中起着至关重要的作用,但是却是不可再生资源,于是探索未知地下结构,寻找石油、天然气变得迫在眉睫。地震勘探作为探寻地下油气资源的有效手段,也随着寻找地下资源步伐的加快逐渐发展起来。在采集地震勘探记录的过程中,往往会混入很强的随机噪声,降低资料的信噪比(SNR),难以识别地震资料中的有效信号,严重影响地下结构精确成像和反演。因此,压制地震勘探记录中的随机噪声,提高地震资料信噪比变得尤为重要。时频峰值滤波(time-frequency peak filtering,TFPF)可以在较低信噪比下较好恢复有效信号,该算法无偏估计的条件为被处理信号是线性信号。为了满足线性条件,利用地震数据横向连续性,构建时空滤波轨线沿同相轴方向进行时空TFPF滤波。由于时空滤波轨线与同相轴方向一致,滤波轨线上的信号近似线性,因此时空TFPF在压制强地震勘探随机噪声的过程中,能很好的保持有效信号幅度。然而时空时频峰值滤波通常采用固定滤波轨线对地震勘探记录进行处理,而实际地震勘探记录中同相轴分布复杂,形态不规则,具有固定滤波轨线模型和参数的时空时频峰值滤波对不规则同相轴保持不理想,不能很好地保持同相轴边缘和端点等重要的结构特征。因此,需要依据同相轴结构建立与之一致的滤波轨线模型,并针对同相轴结构特征调整TFPF窗函数。为了解决时空TFPF滤波中轨线不能与同相轴完全匹配的问题,本文提出一种结构导向时频峰值滤波算法用于压制地震勘探随机噪声。该方法结合具有抗噪能力的最小吸收核同值区(SUSAN)算子度量地震勘探数据结构,并依据检测出的结构特征辨识地震勘探记录同相轴边缘和端点特征。在此基础上,基于信号结构特征构建时空滤波轨线,使得滤波轨线与同相轴形态一致,沿滤波轨线对地震同相轴重采样并进行时空时频峰值滤波。由于重采样信号线性程度更高,有利于保持信号特征。此外,提出基于结构导向的窗函数选取方法,在信号区采用传统对称窗函数,窗长较小,以保证信号恢复精度;在同相轴端点采用非对称窗函数,向信号方向滤波,避免端点处模糊;在噪声区域采用时间方向对称窗函数,窗长较大,即可以有效压制噪声,又能避免噪声空间一致性的影响。本文将结构导向时频峰值滤波应用于模拟地震勘探记录和实际共炮点地震勘探记录的处理。实验结果表明,本文方法在低信噪比条件下能够较好地辨识地震资料结构,与径向时频峰值滤波方法相比,结构导向时频峰值滤波算法更有效地压制地震勘探随机噪声,增强了同相轴的连续性,且在同相轴端点处无模糊问题,更好的恢复了整幅地震勘探记录。
[Abstract]:Oil and natural gas play an important role in the development of national economy, but they are non-renewable resources. As an effective means to explore underground oil and gas resources, seismic exploration has developed gradually with the acceleration of the search for underground oil and gas resources. In the process of collecting seismic exploration records, strong random noise is often mixed in, so it is difficult to identify the effective signals in seismic data by reducing the signal to noise ratio (SNR),) of the data, which seriously affects the accurate imaging and inversion of underground structures. Therefore, it is very important to suppress the random noise and improve the signal-to-noise ratio of seismic data. The time-frequency peak filter (time-frequency peak filtering,TFPF) can recover the effective signal better at lower SNR. The unbiased estimation condition of the algorithm is that the processed signal is a linear signal. In order to satisfy the linear condition, using the lateral continuity of seismic data, the space-time TFPF filtering along the same phase axis is constructed. Because the space-time filtering track is consistent with the in-phase axis and the signal on the filtered rail is approximately linear, the space-time TFPF can keep the effective signal amplitude well in the process of suppressing the random noise in strong seismic exploration. However, the time-frequency peak filtering usually uses fixed filtering track to process seismic exploration records, but the in-phase axis distribution is complex and the shape is irregular in the actual seismic exploration records. The time-frequency peak filtering with fixed filtering trajectory model and parameters is not ideal for the irregular in-phase axis and can not preserve the important structural characteristics such as the edge of the in-phase axis and the endpoints. Therefore, it is necessary to establish a filtering rail line model in accordance with the in-phase axis structure and adjust the TFPF window function according to the characteristics of the in-phase axis structure. In order to solve the problem that the track line can not exactly match with the cophase axis in spatio-temporal TFPF filtering, a structure-guided time-frequency peak filtering algorithm is proposed to suppress random noise in seismic exploration. The method measures the seismic exploration data structure with the minimum absorption kernel (SUSAN) operator which has the ability to resist noise, and identifies the edge and endpoint features of the in-phase axis of seismic exploration records according to the detected structural features. On this basis, the space-time filtering rail line is constructed based on the signal structure feature, which makes the filtering rail line consistent with the in-phase axis shape, and resamps the seismic cophase axis along the filtering rail line and carries out time-frequency peak filtering. Because of the higher linearity of resampling signal, it is advantageous to keep the signal feature. In addition, a structure-oriented window function selection method is proposed, in which the traditional symmetrical window function is used in the signal region, and the window length is small to ensure the signal recovery accuracy. The asymmetric window function is used at the endpoints of the same phase axis to filter in the direction of the signal to avoid blurring at the endpoints. In the noise region, the time-directional symmetric window function is adopted, and the window length is large, which can suppress the noise effectively and avoid the influence of noise spatial consistency. In this paper, the time-frequency peak filter guided by structure is applied to the processing of simulated seismic exploration records and actual common shot point seismic exploration records. The experimental results show that the proposed method can identify the structure of seismic data well under the condition of low signal-to-noise ratio. Compared with the radial time-frequency peak filtering method, the structure-guided time-frequency peak filtering algorithm is more effective in suppressing random noise in seismic exploration. The continuity of the same phase axis is enhanced, and there is no ambiguity at the end of the same phase axis, so the whole seismic exploration record is better restored.
【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】：P631.4

【相似文献】