Data Synthesis And Intelligent Recognition Of Seismic Anomalous Bodies

Seismic anomalous bodies is the channel of oil and gas migration and the area of oil and gas enrichment.The accurate identification of seismic anomalous bodies can provide an important basis for seismic data interpretation,reservoir prediction and well location deployment.In the intelligent identification of seismic anomalous bodies,abundant sample label data is the premise for accurate identification of anomalous bodies.Because the acquisition of 3D seismic signal is limited by exploration cost and strata complexity and variability,the sample data directly obtained from seismic signal is highly subjective,low precision and small amount of data,which can not meet the requirements of fine and intelligent recognition of seismic anomalous bodies.In addition,the identification of seismic anomalous bodies is affected by multiple factors,such as anomalous bodies wave field information,dominant frequency of seismic wavelet,signal-to-noise ratio,rotation angle of training data and so on,cause the final identification result to be seriously disturbed.The study of Thesis mainly focuses on two aspects: synthetic anomalous bodies data method and abnormal body intelligent identification.Combine knowledge of geophysics,signal processing,parallel programming,image segmentation algorithm and so on,a systematic study on the the geological pattern characteristics-synthetic data forward modeling-anomalous bodies intelligent identification-discussion on influence factors of anomalous bodies intelligent identification was carried out.The specific work and innovation are as follows:(1)A method for synthesizing sample labels of anomalous bodies in 3D seismic signals by combining the actual logging signals and the geological model characteristics of anomalous bodies in seismic signals is studied..Firstly,the joint distribution of lithological parameters is calculated,then Monte Carlo rejection sampling(MetropolisHastings,M-H)is used to obtain the lithological information of the target layer in the joint distribution,and finally the sampling information is integrated into the stratigraphic frame model.Considering the lithological background of the anomalous bodies,the depositional law,the characteristics of the geological pattern of the anomalous bodies and other factors,the characteristics of anomalous bodies are synthesized in the model integrated with lithologic information.Combined with the three-dimensional U-Net network algorithm,the reliability and practicability of the method are verified.(2)The intelligent recognition of karst cave abnormal body is carried out,and the factors affecting the recognition of anomalous body are comprehensively discussed from qualitative and quantitative aspects.Through the analysis of various influencing factors,the aim is to improve the accuracy and efficiency of intelligent identification of anomalous bodies.Firstly,in order to simulate the characteristics of seismic anomalous body wave field and speed up the efficiency of seismic anomalous body data synthesis,the Phase Shift Plus Interpolation(PSPI)forward and migration method is parallelized.Then,the parallelized PSPI algorithm is used to synthesize the cave sample labels for intelligent identification of cave anomalous body.Finally,combined with the U-Net network algorithm and the abnormal body synthesis method,the factors affecting the intelligent identification of faults and karst caves are discussed and studied from the qualitative and quantitative perspectives.Through the analysis of various influencing factors,the aim is to improve the accuracy and efficiency of intelligent identification of abnormal bodies.