基于改进shapelet转换的油井异常工况识别  

Oil well abnormal condition identification based on improved shapelet transformation

作　　者：王立达 李克文[1] 牛小楠 田继林 WANG Li-da;LI Ke-wen;NIU Xiao-nan;TIAN Ji-lin(College of Computer Science and Technology,China University of Petroleum(East China),Qingdao 266580,China)

机构地区：[1]中国石油大学(华东)计算机科学与技术学院,山东青岛266580

出　　处：《计算机工程与设计》2025年第4期1079-1086,共8页Computer Engineering and Design

基　　金：国家自然科学基金重大基金项目(51991365);山东省自然科学基金项目(ZR2021MF082)。

摘　　要：针对现有的油井异常工况识别通常是根据示功图数据直接给出判断结果,不具备可解释性,不利于生产人员判断处理等问题,提出一种基于shapelet转换的油井异常工况识别方法。根据示功图时间序列特点,通过计算示功图曲线的差值序列得到额外特征,限制shapelet作用范围避免错误匹配,引入间隔shapelet捕捉长时特征三点针对性改进,提高shapelet转换算法在油井异常工况数据集的应用效果。实验结果表明,该方法在保证了较高分类准确率的同时,具备较好可解释性,可为异常工况识别与处置提供可靠建议。The existing oil well abnormal condition identification method give the judgment results directly according to the indicator diagram,which is not interpretable and is not conducive to the judgment and processing of production personnel.To solve these problems,a method of recognizing abnormal oil well conditions based on shapelet transformation was proposed.Three targeted improvements were proposed according to the time series characteristics of indicator diagram,including calculating the difference sequence of the indicator diagram to get additional features,limiting the matching range of shapelet to avoid mismatching and introducing the interval shapelet to capture long-term features.Experimental results show that this method not only guarantees high classification accuracy,but has good interpretability,and provides reliable suggestions for the identification and disposal of abnormal conditions.

关 键 词：油井 示功图 异常检测 工况识别 时间序列分类 最大区分子序列 可解释性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]