基于隐马尔科夫模型的石油钻井事故预警方法-《郑州大学学报(工学版)》

文章信息/Info

Title:: An Accident Prewarning Method of Oil Drilling Based on Hidden Markov Model

Author(s):: Zhu Xiaodong1; Yuan Kunjie1; Wang Yanling2; 1. School of Electrical Engineering, Zhengzhou University, Zhengzhou, Henan, 450001; 2. School of Information Engineering, Zhengzhou University, Zhengzhou, Henan, 450001

Keywords:: Hidden Markov Model; Trend Feature Extraction; Drilling Accident; Early Warning System

摘要:: 为了实时准确地对石油钻井过程中发生的事故进行预警，构造一种基于连续隐马尔科夫模型（CHMM）的石油钻井事故预警方法，利用参数的长短期均值构造出反映原始参数趋势变化的特征量，用特征向量建立CHMM模型，通过历史数据训练CHMM模型，经前向算法得出匹配概率，并利用统计方法来确定预警系统的阈值，对于每一种钻井故障，给出了综合几个参数的预警结果判断策略.，实验证明CHMM预警模型能够及时有效的对钻井异常情况做出预警.。

Abstract:: In order to predict the accidents of oil drilling process in real time and accurately, an early warning method based on Continuous Hidden Markov Model (CHMM) was constructed. The patameters of average vaule in long-time and short-time were utilized to construct the characteristic which could reflect the changing trend of the original parameters. The CHMM model was established based on the characteristic variables and was trained by the history data. Then the matching probability was calculated by the forwardalgorithm and the threshold of the warning system was determined by the statistical method. For each kind of drilling faults, the strategy for judging the warning results based on multiple parameters was designed. The results of experiment proved that the CHMM waring model could predict the drilling accidents more promptly and effectively.