钻井风险实时诊断技术概念设计  

RESEARCH ON REAL-TIME DIAGNOSING DRILLING RISK TECHNOLOGY

作　　者：乐守群[1] 梁海波[1] 赵庆 李建民[1] 靳文博[1] 

机构地区：[1]西南石油大学石油工程学院 [2]中国石油集团工程技术研究院工艺研究所

出　　处：《钻采工艺》2011年第1期7-10,112,共4页Drilling & Production Technology

基　　金："十一五"国家油气重大科技攻关项目(编号:2008ZX05000-021-006)的部分研究成果

摘　　要：由于石油钻井工程的复杂性、不确定性及井下工作状态的隐蔽性和一些测量参数的滞后性,再加之决策者在压力下做出判断的不可靠性,经常导致不能及时发现钻井风险,甚至产生重大悲剧事件。而BP神经网络可以通过自学习、自适应获取信息之间的内在联系机制,有效解决了信息复杂性、不确定性和隐蔽性的问题,实现信息之间非线性映射,是一种很好的多参数智能融合方法。然而以前利用BP神经网络的钻井风险诊断方法的研究并不多见,且只考虑了风险发生时的一些地表采集的风险征兆参数,没有考虑引起钻井风险发生的原因。文章将风险发生的征兆参数和引起风险发生的工程力学参数结合起来运用到BP神经网络上,建立了融合钻井风险力学因素和钻井风险征兆参数的钻井风险诊断模型来实时诊断钻井风险。这样的钻井风险智能诊断方法更具有科学性,并且更便于钻井风险预测。As a result of the drilling project complexity, the uncertainty and some logging parameters of the lag, coupled with policy-makers to make judgments under pressure is unreliability, it causes always the drilling risk can’t be discovered timely, even it will cause the significant tragedy event. The BP neural network has self-learning system, and self-adapting access to intrinsic link mechanisms between information. It effectively solved the complexity, uncertainty and hidden issues of information, realized nonlinear mapping between the information, is a good multi-parameter intelligent fusion. However former downhole risk diagnosis method by using BP neural network had only considered some surface gathering risk indication parameter while the risk has happened, had not considered the causes of the downhole risk. The risk indication parameter and the engineering mechanics parameter were used in the BP neural network to diagnose downhole risk by real-time. So, the downhole risk diagnosis method is more scientific, and is advantageous for the downhole risk forecast.

关 键 词：钻井风险 钻井事故 风险诊断 神经网络 井下风险 

分 类 号：TE22[石油与天然气工程—油气井工程]