化工管路泄漏的二级故障诊断系统  

A two-step fault diagnosis framework for chemical pipeline leakage

作　　者：田文德[1] 孙素莉[2] 

机构地区：[1]青岛科技大学化工学院,山东青岛266042 [2]青岛科技大学高分子科学与工程学院,山东青岛266042

出　　处：《华中科技大学学报（自然科学版）》2009年第S1期172-176,共5页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：教育部留学回国人员科研启动基金资助项目(2005-29);山东省优秀中青年科学家科研奖励基金资助项目(2006BS05005)

摘　　要：提出了一种基于二级混合结构的化工管路泄漏故障诊断方法,利用多元统计分析来监测管路流量,当流量异常时通过分析变量贡献来确定故障部位,最终采用管路流动机理模型进行故障诊断.为获取有效的稳态数据来建立多元统计分析模型,提出了一种自动确定采样滞后量的方法.实例应用结果表明,该系统诊断速度快,诊断精度与模型法相当,从而说明该二级诊断方法可行、有效.A two-level framework for pipeline leakage detection and diagnosis is proposed.In the first level,multivariate statistical analysis is used to monitor fluid flow process by the quantity of T2 and Q and their thresholds,and then leakage locations are identified by variable contributions to these quantities.The hyperbolic tangent function is utilized to automatically calculate steady-state sampling period for multivariate statistical modeling.Subsequently the second level is activated with dynamic pipeline flow mechanism model to perform detailed leakage diagnosis at the leakage locations.Case study on a complex pipeline shows that this method detects and identifies chemical pipeline leakage fault quickly and effectively,with a similar precision to modelbased method.

关 键 词：故障诊断 管路泄漏 多元统计 动态模型 二级系统 

分 类 号：TQ056[化学工程]