| With the rapid development of petroleum exploration,oilfield data is growing explosively.Oil companies have accumulated a large amount of reservoir data.As a result,it is improtant to manage the oilfield via modern information and intelligence technology.Generally,reservoir data feature large scale,diversity of data,and complex data structure.In this situation,it is difficult for traditional data analysis methods to satisfy the practical needs of oil companies.In recent years,big data technologies,e.g.,Hadoop and Spark,can handle and analyze big data quickly and effectively.Thus,they are suitable for analyzing big reservoir data.Based on data technologies such as Hadoop and Spark,we study the technique and system for reservoir data processing and analysis.The main work is shown as follows.(1)Design a general framework of reservoir data analysis based on big data technology.This framework consists of five functional modules,i.e.,data acquisition,data access,data storage,data analysis,and data application.According to the characteristics of reservoir data and practical requirements,we carefully devise each module in the proposed framework.(2)Desingn a distributed storage scheme of reservoir data using HBase.Reservoir data are heterogeneous,diverse,and multi-source.Considering these features,we employ HBase to devise a distributed storage framework,where reservoir data table and HBase table are built to solve the distributed storage problem.At the same time,we implement HBase secondary index with Coprocessor.(3)Use C4.5 decision tree to present a new algorithm for stratigraphic classification.In order to mine useful knowledge from reservoir data,we make use of C4.5 decision tree to classify the oil layer and perform the classification algorthm based on Spark.(4)Design and develop a reservoir data analysis system.Based on Hadoop,we employ big data technology to develop a reservoir data analysis system,which includes the following main functions:distributed storage of reservoir data,reservoir data analysis,oil stratigraphic classification,and graphical display. |
