郎宪明 李平 曹江涛 芦思宇 李俨 任泓

摘 要：针对管道运行状态数据的非平衡性会造成管道泄漏诊断准确率下降的问题，提出了一种基于非平衡数据的管道泄漏检测与定位方法.首先，将管道各工况非平衡数据采用基于K均值聚类的欠采样方法处理，使其达到数据平衡.然后，将Fischer-Burmeister函数引入到双支持向量机学习过程中，以避免目标函数求解时矩阵的求逆计算，并将平衡数据作为改进双支持向量机算法的输入，识别管道泄漏.采用相关分析法实现泄漏点定位.根据Flowmaster搭建的管道模型，运用该方法识别管道泄漏.仿真实验表明，与经典双支持向量机和拉格朗日双支持向量机相比，该方法能更快速识别管道泄漏孔径及定位.

关键词：非平衡数据；K均值聚类；双支持向量机；泄漏检测；泄漏点定位

中图分类号：TP277 文献标志码：A

Abstract：As the data imbalance of pipeline working conditions decreases the accuracy of the pipeline leakage diagnosis， a method of pipeline leak detection and location based on imbalance data was proposed. First， the imbalance data of different working conditions were processed by K-means clustering algorithm and under-sampling to achieve the balance data. Then， the Fischer-Burmeister function was introduced into the learning process of the twin support vector machine （TWSVM）， in order to avoid the matrix inversion calculation， and the balance data were input into the improved TWSVM to distinguish the pipeline leakage. Leak location was obtained by the cross-correlation function method. Moreover， a flow model of pipeline was put forward based on the Flowmaster software， and the proposed method was used to identify pipeline leakage. The experimental results show that the proposed method is more effective than the classical TWSVM and the Lagrange TWSVM to identify the pipeline leakage aperture and location.

Key words：imbalance data； K-means； twin support vector machine； leak detection； leakage location

管道運输以安全可靠，经济实用的特点在油气运输中被广泛应用[1-2].由于输送介质具有易燃易爆的特性，一旦管道发生泄漏，可能引发人员伤亡和环境污染等重大事故，所以需要对管道运行状态进行在线监控，及时发现泄漏并做相应处理.

管道运行过程的监控数据具有明显的分布不均衡的特征，即正常运行状态的样本数量远远多于泄漏样本及其他工况样本的数量.采用经典支持向量机方法建立管道泄漏检测模型时，多数方法是以平衡数据作为学习样本.文献[3]通过获取管道同沟敷设光纤振动信号的特征向量，建立管道泄漏、人工挖掘和人走动3种情况的平衡数据样本，用于支持向量机（SVM）的学习和测试，从而判断管道沿线是否有异常事件发生；……