Research On Extracting And Cluster Analysis Method Of Pre-stack Seismic Signal Based On Deep Learning

Seismic facies analysis is an important technique in the process of petroleum exploration and can be used as a basis for inferring geological environment.The traditional method is to use the relationship between seismic data parameters as the discrimination means,including amplitude,polarity,reflection abundance,layer velocity,etc.During the data processing,the accuracy will also be reduced due to various processing methods themselves.Prestack seismic signals contain more information.The paper uses prestack data combined with neural networks and machine learning methods to study seismic facies analysis.Prestack seismic signals have a large amount of information and high data dimensions.They are directly used for cluster analysis and the calculations are too large and the results are not accurate enough.This paper uses neural networks to perform dimensionality reduction on seismic data,and finally uses the reduced seismic features to perform cluster analysis.The specific test process is as follows:(1)Use the physical model to train the network,combine the signal data of the model to build a suitable model,and adjust the model parameters.(2)Use the trained network to extract signal features and perform cluster analysis.(3)Validate the model using actual data.