Research on micr oseismic denoising method based on CBDNet

Naise sppeaion s an important part of micrseimic momiloring techomology.Sigmul and naise can be separated by denoisig and fihering to improve the subesequent amlys.In this paper,we popoase a new denoising method besed on comvolutional blind denoising netwonk(CBDNet).The method is pnily modied for image denoising netwarck CBDNet to make it suitble for ome dimernsional data denoising At present,moast aof the existing ftering methods are proposed for the Gausian white nmoise denoising h comtrast the propesed method also leams the wind moise mnstruction noix trafc noie and mixed noise through the sategy of reidual leamig.The full anvohution subnetwark.is used to esimate the noise level,which significandy improves the sigmal.to nise mio and ibs perfommance of removing the comelated noises The model is trmined with dffent types of real naise and randoam noises The denoising esult is evaluated by comespanding indexes and compured with ofher denoing methods.The reuls show that the poposed method has better demoising performance than raditiomal methods,and it has a superior noise supession level for ail well construction noise and mixed noise.The proposed method can supress the inerference of time frequeney owedapped end to end and still he nmoise suppesion and event detection capability even if the sigmul is superimpased on other types of noie.