Research On Well-Seismic Analysis Method For Reservoir Prediction

With the continuous improvement of oil and gas exploration technology,the number of oil and gas exploration in the middle and shallow large anticlines has been gradually reduced,the output of crude oil is gradually reduced,the new recoverable reserves in the old area are reduced year by year,it's difficult to find the hidden oil reservoir.It is difficult to solve the problem of the reservoir prediction in the complex hidden oil and gas reservoir under the current situation,and the information of the well data or seismic data can not meet the requirements of the actual production.The conventional reservoir prediction is limited to the existing theories and methods,to describe the relationship between multidimensional data,multi solution problem of reservoir cannot improve the different data brought by prediction,so how to reduce the uncertainty in prediction is an urgent problem.We cannot interpret this information separately from well logging data and seismic data.We must adopt an interdisciplinary and multidisciplinary multidisciplinary integration and complementary development path to solve the problem that single seismic data cannot meet the needs of complex hidden reservoir predictions,and the diversity and complexity of the data related to reservoir prediction cause the multi-solution of conventional reservoir prediction to be different from the depth when the well data and seismic data are located and the scale is not Matching and other issues.To solve the above problems,this paper fully absorb the domestic and abroad combined with seismic reservoir prediction research results achieved in the field,to understand the related research status combined with seismic reservoir prediction,this paper puts forward a high precision seismic data,seismic data analysis method,storage method based on MapReduce and imbalanced data reservoir prediction optimization algorithm.Firstly,this paper analyzes the characteristics of the seismic data,establishes the method of extracting and collating seismic data,proposes the data fusion specification of wells,uses the distributed column storage technology to build large data sets that simulate well information,and optimizes the machine learning under unbalanced data,the experimental analysis shows that the well-seismic dataset proposed in this paper improves the efficiency of data extraction.The seismic attributes selected by the FSLDL algorithm include commonly used seismic attributes.The predicted longitudinal seismic profiles and transverse seismic profiles are in good agreement with actual reservoirs.