Study On Multi-attribute Fusion Mode Of Seismic Fault Interpretation

The strike-slip fault in a certain area in the west is mainly a negative flower monotype structure,in which a series of faults of different scales and scales are developed,and the seismic reflection characteristics are chaotic and abnormal.The recognition effect of traditional interpretation methods has limitations and cannot meet the identification requirements of different sequence fractures.Therefore,this paper proposes an intelligent identification method for strike-slip faults,aiming to explore the effectiveness and feasibility of applying machine learning methods guided by geological knowledge in the field of fault identification.In this paper,multiple single-frequency seismic data are obtained by spectral decomposition technology,and the multi-scale and multi-frequency seismic sensitive attributes in the longitudinal and horizontal directions are analyzed,and the multi-attribute fusion is realized by constructing a spatial topology by combining the structure,scale and seismic response characteristics of the strike-slip fault,which by jointly analyzing the fault distribution in various seismic sensitive attributes,determining the type and location of the break point and section,and constructing a third-order spatial topology by connecting it by the point and line method.The development location and appearance of the strike-slip fault in the target area are clearly displayed.In order to quickly reproduce the interpretation effect of this method,this paper makes sample labels according to the topological structure,analyzes the fracture recognition effect of different networks in numerical models,selects a deep aggregation neural network to train the strike-slip fault,and completes the fracture prediction of the target work area.The results of three-dimensional actual data test show that compared with the conventional single seismic attribute and principal component analysis method,the fault prediction effect of the multi-attribute fusion method proposed in this paper is significantly improved,the fault structure is clear,the structural details are increased,and the resolution of sequential faults is clear.It is proved that the multi-attribute fusion fault interpretation method based on topology structure can be guided by geological knowledge and geological experience,constrain the learning mode of neural network,greatly improve the recognition effect of strike-slip fault,and the method has been quickly and efficiently popularized and applied in neighboring work areas.