Noise Suppression Of Seismic Signals Based On Wavelet Transform

As one of the commonly used and effective methods for earth resources and geological exploration,seismic exploration is widely used in the investigation of petroleum and natural gas.During the seismic data collection process,human and environmental factors often cause the seismic signal data acquired to contain a large amount of interference information,which seriously affects the interpretation of follow-up data and makes it difficult to analyze the target stratigraphic structure and lithology information.Therefore,if the random noise contained in the seismic signal is effectively suppressed and the interference information is eliminated,high-resolution seismic data can be obtained.it will be helpful for the observation and geological analysis of subsequent seismic data and lay a solid foundation for the exploration work.That is why this article focuses on the suppression of random noise in seismic signals.The main tasks are as follows:(1)Study on an improved wavelet threshold algorithm for Chaos fruit fly optimization.The commonly used wavelet threshold denoising algorithm need to determine the threshold value,usually based on the prior information of the signal to estimate the noise variance,thus making the wavelet threshold with strong guess and uncertainty.Based on this issue,In this paper,firstly,the signal is decomposed into high-frequency and low-frequency wavelet coefficients of multiple scales by wavelet decomposition,and then the generalized cross-validation(GCV)threshold selection function is used as the basis function for selected threshold.And using the improved Chaos fruit fly optimization algorithm to iteratively optimize the basis function to get the optimal wavelet threshold,then combined with the soft threshold function to thresholding of the high frequency wavelet coefficients.Finally,Reconstructed the signal after the noise.This paper algorithm of determining the thresholddoes not require prior information,which greatly reduces the guessing of threshold selection.(2)In order to solve the problem that the Fruit Fly Optimization Algorithm(FOA)is easy to fall into the local optimal solution.Therefore,in this paper the improvement strategy of Chaos Drosophila optimization algorithm is studied.For each optimal fruit fly that iterates from FOA algorithm,a constrained chaotic search is performed,and The constraints are adjusted by the chaotic factors so that the FOA algorithm avoids falling into a local extremum and obtains an optimal solution.(3)For the traditional non-local means filter(NLM)algorithm which set up the filter parameters do not consider the noise intensity in the noisy signaland ignore the structural defects of the signal.This paper studies the adaptive NLM algorithm based on wavelet entropy.The wavelet entropy is used to estimate the average noise variance in the search window of the NLM algorithm.Then the noise variance is used to help determine the measurement of the filter parameter in the search window.The objective is to achieve denoising based on local differences by signal structure characteristics and noise intensity.For improve the denoising effect.The algorithm and the selected contrast algorithm are applied to thenoisysim-ulated seismic records and actual seismic records.By observing the processed seismic record profile,it is verified that the proposed algorithm is superior to other comparison algorithms in suppressing noise and maintaining signal amplitude.The average signal-to-noise ratio(SNR)and average square root error(MSE)of the simulated seismic records after processing were compared from the quantitative perspective.In the improved NLM algorithm experiment,the average SNR after denoising by the original noisy signal,the traditional NLM algorithm and the denoised algorithm of this paper are 5.14 dB,16.23 dB,18.80 dBand the average MSE is 0.0074,0.00056,0.000307.It can be seen that the overall SNR and MSE of the seismic signal after the algorithm processing is superior to the traditional NLM algorithm,which all prove the effectiveness of the improve algorithm.