Research On The Method Of Regular Interference Suppression Based On Non-stationary Filtering

2D filtering is the main method of regular interference suppression.However,the biggest defect of the filtering method based on 2D Fourier transform is that the filter cannot change with time and space.This loss of time and space resolution results in the inability to focus on the local characteristics of regular interference to suppress it.For example,the traditional F-K filter cannot adapt to the time and space characteristics when suppressing regular interference.The current time-varying and space-varying 2D filtering methods are difficult to realize the point-by-point change of the filter,and the filter factor usually generates distortion during the filtering process,which greatly reduces their application in seismic signal processing,such as methods based on feedback filtering and window shifting.Therefore,this paper presents a non-stationary filtering model based on 2D Fourier transform,which allows filters to change point by point with time and space,and does not limit the types and variations of 2D filters.After that,a fast 2D non-stationary filtering algorithm in the time-space domain is developed,which greatly reduce the amount of calculation while ensuring the accuracy of the filtering result by scaling and extracting the reference filtering factor.Finally,as an application,an adaptive linear interference suppression method and a non-stationary filter adaptive inversion method are proposed respectively.The former determines the filter parameters from the prior information of effective signal,which avoids the prediction of linear interference.The latter uses the distribution of seismic signals in the Radon domain as a priori information,and can adaptively invert the filter parameters required for multiple suppression from the given seismic data.