Machine Learning And Prediction Of Multiple-components Seismic Reservoirs Based On Color Fusion And Convolutio Feature Analysis

Multi-component seismic data carries abundant information of oil and gas reservoirs.How to effectively predict oil and gas reservoirs using the abundant information,so as to shorten the exploration cycle,reduce production costs and improve accuracy of prediction,has always been the goal of the industry.The sensitivity of P-wave and PS-wave to oil and gas reservoirs is often different,which can be used to improve the accuracy of oil and gas reservoir prediction.This study designed a machine learning and prediction method for multi-component seismic reservoir based on color feature fusion and convolution feature analysis,combining with the distribution law of oil and gas reservoirs and the response of seismic characteristics,relying on artificial intelligence technology.Unsupervised and supervised learning is used to extract the characteristics of oil and gas to realize accurate identification and prediction of seismic oil and gas reservoirs.Using the sensitivity difference of oil and gas between P-wave and PS-wave,this study proposes a multi-component seismic reservoir unsupervised learning method based on color feature fusion.First,a large number of compressional and shear wave seismic attributes were extracted using cluster analysis to conduct unsupervised learning to optimise the attributes.Then,using the different responses of compressional and shear waves to oil and gas and an understanding of rock physics,three types of composite attribute were constructed to highlight oil and gas anomalies by multi-component seismic attributes.Finally,the three composite attributes were transformed to the colour space by a first-order linear transformation and RGB colour blending,obtained unsupervised learning prediction results of multi-component seismic reservoirs based on color feature fusion.Meanwhile,this study exploratorily develops a machine learning and prediction method based on convolution characteristics analysis,using the distribution law of oil and gas reservoirs and its seismic response of characteristics.That is,the convolution kernel is obtained using the limited known information of oil and gas well,and the convolution calculation is carried out by using the P-wave and PS-wave seismic data as input to identify the seismic oil and gas characteristics.The scheme is applied to the model data and actual data,which verifies the effectiveness and feasibility of the method.These two schemes have been applied to the prediction of seismic reservoirs in FG area,and the expected prediction results were obtained.Applying this scheme to reservoir prediction shows that unsupervised learning and colour blending techniques could help the human eye perceive geological anomalies,highlight common hydrocarbon characteristics,reduce differences and decrease interpretation ambiguity.Supervised learning method based on convolution feature analysis can extract,classify and recognize oil and gas characteristics intelligently.A good prediction result can be obtained by using only a small section of oil and gas information.Compared with unsupervised learning prediction method based on color feature fusion,it has higher prediction efficiency and helps to reduce production costs.The prediction results of the two schemes are in good agreement with the actual situation,which indicates that the scheme is feasible and effective,It provides a new way for accurate identification and prediction of oil and gas reservoirs.