卡钻风险预警与识别方法研究
发布时间：2018-11-15 19:11
【摘要】：在石油勘探的过程中,钻井事故与复杂问题总是客观存在的。卡钻事故占整个钻井事故的40%～50%,由卡钻引起的资金耗费占非生产耗费的50%以上。在国家科技重大专项(合同号：2011ZX05021-006)“钻井实时监测与技术决策系统”中,设置了如何预警、识别各种井下风险事故的研究任务,而卡钻风险事故是最为复杂的钻井风险事故之一。本文就基于实时数据对卡钻事故的预警和识别展开了研究工作。 人工神经网络算法是一种非线性、强自适应学习能力的数据信息处理方法,其极强的非线性逼近能力能真实表示出输入变量与输出变量之间的非线性关系。针对卡钻事故中参数变化规律,将事故与数据变化之间的非线性关系通过人工神经网络对特征参数基于样本进行训练建立联系,在实时数据传输的基础上,能够将指导卡钻事故的各个参数变化非线性映射成对卡钻事故的识别。 本文根据已经发生的卡钻事故记录数据,深入分析和总结了卡钻事故发生的类别和特点。对卡钻事故发生过程进行研究,并用对应参数变化表示整个过程的发展,确定了各个卡钻事故在整个过程中的特征参数。 卡钻风险是卡钻事故发生早期的异常反应,对卡钻风险进行识别能够在卡钻事故发生之前进行预警。将井下钻具活动方式分类为正常钻进状态和钻具的上下活动状态,配以钻具静止时间等实时计算参数,针对卡钻风险的发生能够较好的区别各类卡钻对应发生状态。根据各状态下卡钻风险征兆规律,运用神经网络算法对井下发生异常进行识别,并结合实时计算参数建立卡钻风险预警模型。 根据卡钻事故发生之后对应的表征规律,卡钻事故发生之后对各钻井状态下呈现相同规律,运用神经网络算法实现对各状态下发生了的卡钻事故进行识别,结合卡钻事故发生之前对卡钻风险预警的分类情况,确定卡钻事故的类别,建立了卡钻事故分类识别模型。 在卡钻风险预警及卡钻事故识别模型的基础上设计了卡钻风险预警与识别软件。对有一定规律的卡钻风险,可进行预警和识别,其结果部分与实际相符。通过实例进一步完善之后,可以对现场提供卡钻事故预警及识别参考。
[Abstract]:In the process of petroleum exploration, drilling accidents and complex problems always exist objectively. Drilling jam accounts for 40% of the total drilling accident, and the capital cost caused by drilling jam accounts for more than 50% of the non-production cost. In the National Science and Technology Major Project (contract number: 2011ZX05021-006), "drilling Real-time Monitoring and Technical decision system", the research task of how to early warning and identify various downhole risks and accidents has been set up. The drilling risk accident is one of the most complex drilling risk accidents. In this paper, the early warning and recognition of drilling jam accidents based on real-time data are studied. Artificial neural network (Ann) algorithm is a kind of data information processing method with nonlinear and strong adaptive learning ability. Its strong nonlinear approximation ability can truly express the nonlinear relationship between input variable and output variable. In view of the rule of parameter change in drill jam accident, the nonlinear relation between accident and data change is linked by artificial neural network to train characteristic parameters based on sample, and on the basis of real time data transmission, the relationship between the nonlinear relation between the accident and the change of data is established by means of artificial neural network. The nonlinear mapping of the parameters used to guide the drill jam accident can be used to identify the drill jam accident. Based on the recorded data of drilling accidents, this paper analyzes and summarizes the types and characteristics of drilling accidents. The development of the whole process is represented by the change of corresponding parameters, and the characteristic parameters of each drill jam accident in the whole process are determined. Drilling risk is the early abnormal reaction of drilling accident. The downhole drilling tools are classified as the normal drilling state and the upper and lower moving state of the drill tool and the real-time calculation parameters such as the drilling tool static time can be used to distinguish the corresponding occurrence states of various kinds of drill jammed well in view of the occurrence of drilling jam risk. According to the regularity of the risk symptom of drilling jam under different conditions, the neural network algorithm is used to identify the abnormal underground and the early warning model of the risk of drilling jam is established by combining the real-time calculation parameters. According to the corresponding representation law after the drill jam accident, the same rule is presented for each drilling state after the drill jam accident occurs, and the neural network algorithm is used to realize the recognition of the drill jam accident in each state. According to the classification of early warning of drilling risk before the occurrence of drill jam accident, the classification and recognition model of drill jam accident is established. Based on the model of early warning and identification of jam risk, the software of early warning and recognition of drilling risk is designed. Early warning and identification can be carried out for the drilling risk with certain regularity, and the results are in accordance with the actual situation. After further improvement through examples, it can provide early warning and identification reference for drilling jam accident in the field.
【学位授予单位】：西南石油大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TE28
本文编号：2334191
[Abstract]:In the process of petroleum exploration, drilling accidents and complex problems always exist objectively. Drilling jam accounts for 40% of the total drilling accident, and the capital cost caused by drilling jam accounts for more than 50% of the non-production cost. In the National Science and Technology Major Project (contract number: 2011ZX05021-006), "drilling Real-time Monitoring and Technical decision system", the research task of how to early warning and identify various downhole risks and accidents has been set up. The drilling risk accident is one of the most complex drilling risk accidents. In this paper, the early warning and recognition of drilling jam accidents based on real-time data are studied. Artificial neural network (Ann) algorithm is a kind of data information processing method with nonlinear and strong adaptive learning ability. Its strong nonlinear approximation ability can truly express the nonlinear relationship between input variable and output variable. In view of the rule of parameter change in drill jam accident, the nonlinear relation between accident and data change is linked by artificial neural network to train characteristic parameters based on sample, and on the basis of real time data transmission, the relationship between the nonlinear relation between the accident and the change of data is established by means of artificial neural network. The nonlinear mapping of the parameters used to guide the drill jam accident can be used to identify the drill jam accident. Based on the recorded data of drilling accidents, this paper analyzes and summarizes the types and characteristics of drilling accidents. The development of the whole process is represented by the change of corresponding parameters, and the characteristic parameters of each drill jam accident in the whole process are determined. Drilling risk is the early abnormal reaction of drilling accident. The downhole drilling tools are classified as the normal drilling state and the upper and lower moving state of the drill tool and the real-time calculation parameters such as the drilling tool static time can be used to distinguish the corresponding occurrence states of various kinds of drill jammed well in view of the occurrence of drilling jam risk. According to the regularity of the risk symptom of drilling jam under different conditions, the neural network algorithm is used to identify the abnormal underground and the early warning model of the risk of drilling jam is established by combining the real-time calculation parameters. According to the corresponding representation law after the drill jam accident, the same rule is presented for each drilling state after the drill jam accident occurs, and the neural network algorithm is used to realize the recognition of the drill jam accident in each state. According to the classification of early warning of drilling risk before the occurrence of drill jam accident, the classification and recognition model of drill jam accident is established. Based on the model of early warning and identification of jam risk, the software of early warning and recognition of drilling risk is designed. Early warning and identification can be carried out for the drilling risk with certain regularity, and the results are in accordance with the actual situation. After further improvement through examples, it can provide early warning and identification reference for drilling jam accident in the field.
【学位授予单位】：西南石油大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TE28
【参考文献】
相关期刊论文 前6条
1 刘金华;工程参数录井在钻井工程中的应用[J];断块油气田;2003年05期
2 李凤霞;崔茂荣;王丽华;马勇;;综合录井技术在实时监测钻井事故中的应用[J];断块油气田;2007年03期
3 史玉升,梁书云;钻进过程实时状态监控与事故诊断专家系统——人工智能在钻井工程中的应用之一[J];地质与勘探;1999年02期
4 郝进义;;数据库设计规范及设计技巧研究[J];计算机光盘软件与应用;2012年12期
5 陈锐;李黔;尹虎;袁本福;;钻井风险实时监测与诊断系统设计及应用[J];断块油气田;2013年01期
6 陈锐;李黔;尹虎;;案例推理技术在钻井风险预测中的应用[J];断块油气田;2013年02期
，本文编号：2334191
本文链接：https://www.wllwen.com/kejilunwen/shiyounenyuanlunwen/2334191.html
