Intelligent Oil And Gas Detection Based On Seismic Time-Frequency Information

With the continuous development of human oil and gas exploitation activities,there is a higher accuracy requirement for oil and gas detection methods.Time frequency analysis is an effective method to analyze non-linear and non-stationary signals by projecting one-dimensional time signal into two-dimensional time-frequency space to describe the time and frequency characteristics of the signal in the local range.The resolution of traditional time-frequency method is not high,so it is difficult to meet the requirements of high resolution in oil and gas detection.In this paper,MSST resolution time-frequency method is introduced into the field of geophysics for high-resolution time-frequency analysis of seismic signals.The traditional method of oil and gas detection based on the theory of two-phase medium is interfered by human factors,and can not make full use of the time-frequency information in seismic signals.In this paper,based on the MSST transformation and BPNN model,the MSST-BPNN model is proposed,and the intelligent oil and gas identification is carried out in jin72 study area.The comparison results show that the MSST-BPNN has high recognition accuracy in oil and gas detection,and can be used for the initial exploration The plane distribution of oil and gas can be predicted in this paper.On this basis,a comprehensive intelligent oil and gas detection system is developed,which integrates high-resolution time-frequency analysis,neural network model building and training,and intelligent oil and gas detection.The main achievements of this paper are as follows:(1)Based on the theory of two-phase medium,oil and gas detection was carried out in jin72 research area,and the coincidence rate with drilling and gas testing results was 76%.A set of signal analysis and oil and gas detection system for seismic data is developed.It has the function of spectrum analysis of seismic data,and can detect oil and gas and project well in a given area based on the theory of two-phase medium.(2)Based on the comparison of STFT transform,S transform and SST transform,the MSST transform method is introduced into the geophysical field.The comparison experiment between simulation signal and actual seismic signal shows that the MSST transform has higher time-frequency resolution and energy focusing.(3)Based on MSST transformation and BP neural network model,a MSST-BPNN model for oil and gas detection is established,and the intelligent oil and gas detection in jin72 study area is realized based on MSST-BPNN model.The oil and gas detection results of MSST-BPNN model are consistent with those of 25 wells in jin72 study area,and the recognition rate reaches 96%.(4)A set of intelligent oil and gas detection system is developed,which supports the use of a variety of time-frequency analysis conversion methods for time-frequency analysis of seismic trace data.The neural network model can be built and trained according to the given data set,and the application and verification of the trained model can be carried out to complete the intelligent oil and gas detection in the target area.