Prediction Of Reservoir Using Multi-component Seismic Data Under Multilayer Network Structure Of Machine Learning

Multi-component seismic data carry rich oil and gas reservoir informations,how to effectively use these oil and gas characteristics information to direct oil and gas reservoir prediction,so as to shorten the exploration cycle and reduce production costs.In recent years,machine learning algorithm has been widely used in oil and gas feature extraction,classification,recognition and prediction because of its strong model robustness and high generalization ability.Based on this,this study,based on the distribution of oil and gas reservoirs,based on the multi-component data,organically and effectively uses the difference between the sensitivity of the P-wave and the transversal waves to the oil and gas reservoirs,designed a multi-layer network based on machine learning to predict the distribution of seismic oil and gas reservoirs.First,in accordance with the Wright criterion.Second,using the verified clustering and unsupervised algorithms,the hidden layer is constructed to enhance network sharing and extract hydrocarbon features.Finally,A well-point sample obtained from the compromised algorithm of expected risk assessment under the idea of increasing the penalty value of the network is used as the input training sample for the support vector machine algorithm of supervised learning,and the hydrocarbon characteristics at the remaining positions are used as the learning sample set.Then conduct predictions from known to unknown seismic reservoirs.This scheme is applied to the prediction of seismic oil and gas reservoir in HG area.Combined with drilling and other geological data,it is shown that the seismic oil and gas reservoir boundary predicted by this scheme is clearer,the prediction results are basically consistent with the actual situation,and the prediction of the favorable exploration area in the study area is made.It provides a new prediction method for seismic reservoir prediction.This study demonstrated a technique flow for the identification of oil and gas reservoir by machine learning,which can be used to solve the problem in practical production.