Application Of Seismic Multi-attribute Analysis In Fracture Identification Of Compact Reservoir

Different kinds of seismic attributes can be obtained through different mathematical and physical transform of seismic data.The seismic attributes contain information of stratigraphic lithology,reservoir physical properties,fluid type and fracture,and the seismic attribute analysis is one of the effective ways of prediction of the reservoir.These seismic attributes have their own emphasis when describing the geological features.The dip and azimuth attributes,coherence and curvature properties of the formation have good effect in fracture identification.Stratigraphic dip and azimuthal properties can be used to describe the basic changes in stratigraphic structure.Coherence has better effect in fault and channel identification.Curvature properties are sensitive to subtle changes in geological features and can be used for fracture identification.Since reservoir prediction using single seismic attribute may cause multi-results,it is usually preferred to attributes that can reflect the characteristics of the reservoir area,thereby reducing the multi-results,improve the accuracy of prediction.However,when multi-attribute is used,there is a problem of information redundancy.You can extract the redundant information of important information in the attribute through the algorithm of attribute optimization.Multi-attribute analysis in the application of reservoir prediction is mainly attribute conversion and attribute clustering two categories of methods.The multi-attribute analysis for the reservoir parameters prediction needs to establish the corresponding relationship between the seismic attribute and the reservoir parameters.When the relationship to be trained is more complicated or the sample points are less,the conventional method has some limitations and can't meet the accuracy requirements.With trained deep network,deep learning helps better understand features of the data interested and improve the prediction accuracy.In order to better excavate the relationship between seismic attributes and reservoir parameters,the use of deep belief network for seismic multi-attribute analysis can achieve better results.In this paper,the deep belief network is used for the multi-attribute analysis to predict the reservoir thickness and the development of the fractures.