Research On Seismic Waveform Classification Method Based On Support Vector Machine

Waveform classification technology has been widely used in petroleum exploration.This technology can conveniently and accurately classify and predict underground rock formations as well as oil and gas reservoirs,providing an effective way for exploration personnel to identify oil and gas reservoirs.However,it has been found that seismic data classification theory and practice still have problems such as incomplete seismic data,high noise,many artificial errors in the process of horizon tracking,complex feature extraction,and unsatisfactory classification methods.This paper focuses on studying these issues:Firstly,it analyzes seismic waveform classification process and different application methods,the Chebyshev polynomial fitting interpolation method and the structure-oriented filtering method are applied to the data preprocessing,and an automatic tracking method based on waveform features and correlation is implemented in horizon tracking.The method of Principal component analysis is used to select the features of the horizon data,and classification algorithm based on combination of decision tree and support vector machine implements seismic waveform classification.Secondly,a seismic waveform classification software application platform based on support vector machines is developed.Finally,the development platform is used to process the actual work area data,and the effects of support vector machine with and without decision trees are compared and analyzed,the results show that the classification algorithm based on the combination of decision tree and support vector machine has superiority in seismic waveform classification.After applying the relevant methods proposed in this thesis,better tracking effect and actual classification effect are achieved,and the classification result is improved.The research in this thesis has certain significance and value in seismic exploration theory and practice...