复杂工况下热油管道漏失识别与定位方法研究  

Leak Detection and Positioning Method for Heat Pipelines in Complicated Conditions

作　　者：梁伟[1] 张来斌[1] 郭存杰[1] 陈志刚[1] 

机构地区：[1]中国石油大学(北京)机电工程学院,北京102249

出　　处：《石油矿场机械》2009年第6期12-16,共5页Oil Field Equipment

基　　金：国家高技术研究发展计划(863计划)项目(2008AA06Z209);中国石油天然气集团公司创新基金(2006-A类);教育部新世纪优秀人才支持计划资助项目(NCET-05-0110);北京市教育委员会共建项目专项资助

摘　　要：针对复杂工况下热原油管道泄漏难以准确识别的难题,提出采用基于多元支持向量机的管道泄漏诊断方法,并建立了识别模型,可以在小样本情形下完成模型的训练工作,实现多种工况下对压力波动信号的分类识别,从而提高评判泄漏的有效性和准确性。为了解决热输原油管道负压波波速受油品及温度等因素影响较大导致的定位误差,分析了管道沿程轴向温降及修正负压波波速的方法,并采用牛顿-柯特斯积分方法对传统泄漏定位公式进行了改进。现场试验表明,基于多元支持向量机的检测方法能有效地识别管道运行异常状态,改进的漏点定位算法使得定位精度从原来的2.5%提高到1.0%。For solving the difficulty of leakage detection in complicated conditions for heat pipelines, a method based on MSVM （Multi-Support Vector Machine） was proposed and the diagnosis model was established. Model train could be completed in few samples to distinguish differentconditions of pipelines. Moreover, in heat pipelines pressure velocity was affected by oil and pipeline axial temperature drop. Positioning usually had obvious error. For solving this problem, axial temperature drop was analyzed and pressure velocity was revised. By means of Newton-Cotes integration method, positioning formula was improved. The field experiments showed that abnormal conditions including leakage of pipelines could be distinguished effectively by using the method based on MSVM and the improved positioned formula made positioning accuracy increased from 2.5% to 1.0%.

关 键 词：热输油管道 泄漏检测 多元支持向量机 负压波 轴向温降 

分 类 号：TE973[石油与天然气工程—石油机械设备]