基于匹配滤波的时间域自适应全波形反演

发布时间：2018-02-11 18:22

本文关键词： 全波形反演匹配滤波小波变换多尺度目标函数　出处：《吉林大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着勘探对象逐渐向复杂区域转移,需要一种高精度、高分辨率的速度建模方法支撑地震数据的处理与解释。全波形反演方法通过实际数据与理论模拟数据的拟合,充分利用了叠前地震数据的全波场信息重建速度模型,是一种高精度速度建模方法,具有反映模型细节的能力。由于全波形反演理论的固有问题,该方法具有较强的非线性性,导致全波形反演方法在实际应用受到了限制。本文对时间域全波形反演方法进行了详细的研究,针对该方法存在的由强非线性造成的“跳周”现象,提出了基于匹配滤波的解决方法,从数据预处理的角度,对模拟波场数据进行改造,降低目标函数中局部极小点出现的可能性,达到提高反演稳定性的目的。本文提出了一种基于最小平方滤波的时间域全波形反演方法,有效地降低了反演对初始模型和低频信息的依赖性,提高了反演的稳定性。对时间域匹配滤波原理进行了研究,根据最小平方原理计算滤波算子,利用匹配滤波对模拟波场的相位进行调整,减小模拟波场与实际波场间的相位差。同时,利用滤波后的模拟数据构建一种新的目标函数,由于滤波对相位差的调整作用,使得新的目标函数具有较少的局部极小点,能够稳定地下降,降低了反演结果陷入局部极小点的可能性。由于匹配滤波具有统一地震记录特征能力,该方法不仅能够改善记录间的相位差异,而且可以在一定程度上消除子波差异对地震记录造成的影响,使得基于最小平方滤波的全波形反演方法在存在子波误差的情况下,依然具有较好的稳定性。本文对小波域的最小平方滤波方法进行了研究,在小波变换多尺度全波形反演的基础上,提出了基于小波域最小平方滤波的多尺度自适应全波形反演方法,提高了反演的稳定性。该方法在小波域实现匹配滤波,完成对模拟地震记录的修正,并利用小波变换的多尺度特性,从低频带开始反演,不断扩大反演频带范围,实现了多尺度反演。为验证方法的有效性,分别利用上述方法在Marmousi模型上进行了试算,对比试验结果表明,本文提出的算法具有较好的稳定性,一定程度上能够克服“跳周”现象的影响。
[Abstract]:As exploration objects gradually move to complex areas, a high precision and high resolution velocity modeling method is needed to support the processing and interpretation of seismic data. The velocity model reconstructed by fully utilizing the full wave field information of prestack seismic data is a high precision velocity modeling method with the ability to reflect the details of the model. Due to the inherent problems of the theory of full waveform inversion, the method has strong nonlinearity. As a result, the full waveform inversion method is limited in practical application. In this paper, the time-domain full-waveform inversion method is studied in detail, and the phenomenon of "cycle jump" caused by strong nonlinearity in this method is discussed. In this paper, a method based on matched filtering is proposed to modify the simulated wave field data from the point of view of data preprocessing so as to reduce the possibility of local minima in the objective function. In order to improve the inversion stability, a time-domain full-waveform inversion method based on least square filtering is proposed, which effectively reduces the dependence of the initial model and the low-frequency information. The stability of inversion is improved. The principle of matched filtering in time domain is studied, the filter operator is calculated according to the least square principle, and the phase of simulated wave field is adjusted by matching filter. The phase difference between the simulated wave field and the actual wave field is reduced. At the same time, a new objective function is constructed by using the filtered simulation data. Because of the adjustment of the phase difference by the filter, the new objective function has less local minima. This method not only improves the phase difference between the records, but also reduces the possibility of the inversion result falling into the local minima. Because the matched filtering has the ability of unified seismic record feature, this method can not only improve the phase difference among the records, but also reduce the probability of falling into the local minimum. In addition, the influence of wavelet difference on seismic records can be eliminated to some extent, so that the full waveform inversion method based on least square filter can be used in the case of wavelet error. In this paper, the least square filtering method in wavelet domain is studied, on the basis of multi-scale full waveform inversion of wavelet transform, A multi-scale adaptive full-waveform inversion method based on least square filtering in wavelet domain is proposed, which improves the stability of the inversion. Using the multi-scale characteristic of wavelet transform, the inversion frequency band range is enlarged and the multi-scale inversion is realized from the low frequency band. To verify the validity of the method, the above methods are used in the Marmousi model respectively. The experimental results show that the proposed algorithm has good stability and can overcome the influence of "jump cycle" to some extent.
【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】