油田地震数据处理软件的模块迭代规律挖掘-《计算机技术与发展》

文章信息/Info

Title:: Mining Module Iteration Rules of Oilfield Seismic DataProcessing Software

Author(s):: LI Chun-sheng; ZHOU Zhi-peng; ZHANG Ke-jia; FU Yu; LIU Tao; School of Computer and Information Technology,Northeast Petroleum University,Daqing 163319,China

Keywords:: seismic data; data monitoring; data mining; log mining; iterative rule

摘要:: 在油田的勘探开发过程中,需要对地震数据进行处理和解释得出地震剖面图、地震波速度、频率等,再根据这些数据结合地震波传播理论和石油地质学原理来分析地下地质构造的特点,为探测含油量和提出钻探井位提供依据。目前一些油田公司利用 GEOEAST、WGC、PARADIGM 等大型软件对地震数据进行处理,这些软件执行过程中需要多台节点( 主机)协同工作。而目前油田公司的应对方案是在确保任务可以完成的前提下,人工地将节点进行固定划分。但这种方案不可避免地会造成部分节点闲置或者负载等问题。针对这一问题,提出了通过监控油田地震数据处理软件的许可日志以及各节点的机器 KPI 数据来挖掘出一定作业量下软件执行过程中各功能模块的迭代规律,预测软件执行过程中所占资源情况,为节点资源的动态调度提供依据。并通过实验验证了这一方案的可行性。

Abstract:: In the process of oilfield exploration and development,it is necessary to process and interpret the seismic data to obtain the seismic profile,seismic wave velocity and frequency,and then analyze the characteristics of underground geological structure according to these data combined with the seismic wave propagation theory and the principle of petroleum geology, so as to provide a basis for detecting oil content and proposing drilling well location. At present,some oilfield companies use GEOEAST,WGC,PARADIGM and other large-scale software to process seismic data. During the implementation of these software,multiple nodes ( hosts) need to work together. At present,the response plan of the oilfield company is to manually divide the nodes under the premise of ensuring that the tasks can be completed. However,such scheme will inevitably cause some nodes to idle or load. To solve this problem,we propose to mine the iteration rules of each functional module in the software operation process under a certain workload by monitoring the license log of oilfield seismic data processing and interpretation software and the machine KPI data of each node,so as to predict the resources occupied in the software execution process and provide a basis for the dynamic scheduling of node resources. In addition, we also verify the feasibility of such scheme through experiments.