【作者】 张金鹏；

【导师】 王孝良；

【作者基本信息】 大连理工大学 ， 控制理论与控制工程， 2011， 硕士

【摘要】 本文在深入探讨当前国内石油钻井工程事故诊断预报领域现状的基础上,分析了我国目前钻井事故诊断中存在的问题,并归纳的两点主要原因：第一、钻井传感器信号异常的捕获不够准确及时,导致井下事故时体现在传感器信号变化被忽略,无法做出相应的事故预报。第二、钻井事故诊断的方法盲目借鉴国外钻井事故诊断的方法,忽略了我国油气存储的地质的特殊性以及我国目前的钻井装备技术水平,当井下事故发生时,不能够做出及时准确的预报。针对我国钻井事故诊断中存在的这两个难点,本文提出了基于RDT (Related Dynamic Threshold,自相关动态门限)&FDF (Fuzzy Data Fusion,模糊数据融合)方法的钻井事故诊断系统的研究。本文的主要工作如下：首先,介绍了各个钻井相关传感器在钻井作业中的作用,以及他们所体现的钻井工程相关物理量的意义。之后介绍了钻井事故的成因、危害以及钻井事故发生时体现在传感器上的特征表现。其次,提出了一种新的基于自相关动态门限的钻井传感器信号异常检测的方法,提高了对钻井传感器信号异常的捕获能力。同时,对比了D-S证据理论、神经网络和灰色关联理论方法在钻井事故诊断中的应用,针对上述方法的不足提出了基于模糊数据融合的钻井故障诊断方法,更加贴近钻井作业模糊性和不确定性强的特点。最后,介绍了整体系统的实现,钻井事故诊断系统作为整套油井服务系统三大子系统之一,由数据处理模块,和诊断模块两大部分组成,介绍了自定义数据字典的格式以及钻井重要参数的计算推导和事故诊断中间参数的配置方法。给出了系统的运行效果。经测试系统捕获传感器信号异常的能力比较强,且可以实现随钻井作业的进度和当前钻进岩性的不同动态调整传感器门限。钻井事故的诊断结合人工确认的方法,能够确保诊断依据的正确性,较好地将人工经验与智能诊断相结合。更多还原

【Abstract】 Base on the deep research of present situation on domestic oil drilling engineering accident diagnosis this paper has analyzed the problems in this field. There are two main reasons lead to the not so good drilling accident diagnosis effect:Firstly, the catch of sensor signal abnormalities is not accurate or in time so that lead to the system would neglect the change of the sensor signal abnormalities. Secondly, many researchers use foreign advanced method in the drilling engineering accident diagnosis and forecast in spite of whether or not fit to domestic geological particularity technology level. When some accident happened in the well, these methods can not make an accurate forecast.To deal with the two problems above, a new drilling engineering accident diagnosis and forecast method based on based on RDT&FDF (Related Dynamic Threshold and Fuzzy Data Fusion) is proposed in this paper. Main tasks in this paper are as follows:Firstly, this paper introduces the all kinds of sensors in the drilling process and their effect. And then the reason and danger of drilling accident are expounded in the next paragraph. In the end of the first part this paper states the sensor reflections when the accident happened in the well.Secondly, this paper rise up the accuracy on the catch of sensor signal abnormalities while using the related dynamic threshold method. At the same time, contrast with application of D-S Evidence theory, neural net and gray theory in the drilling accident diagnosis, the method based on fuzzy data fusion is much more close to the characteristic fuzziness and uncertainty of the drilling process.At last, the petroleum drilling accident forecast system contains two main parts:data processing module and diagnosis module. In this part, database design, drilling parameters calculation process and middle parameters of the accident diagnosis is introduced and the result is given in the last part of this paper.After testing the new method has a good capacity to catch of sensor signal abnormalities and can change the threshold of the sensor in the different drilling stage or in the different rock stratum. Meanwhile, this method combines the artificial experience with the software accident diagnosis to ensure the accuracy of the accident forecast result.更多还原