分组趋势预测在温度控制故障识别中的应用  

Trend Forecasting in Group of Reactor Temperature Control Fault Identification

作　　者：邹向琪[1] 宋彤[1] 

机构地区：[1]大连理工大学电子信息与电气工程学部,辽宁大连116024

出　　处：《测控技术》2011年第6期29-32,共4页Measurement & Control Technology

摘　　要：以间歇式反应釜为例,对工业上广泛存在的大惯性时滞时变系统对象进行研究,给出了一种分组趋势预测的新方法。该方法采用数据分组方式将故障检测的范围进行分段,根据每组所在的时间域及过程特点给出相应的故障指标。根据采样数据的分析,结合过程设备特性实现故障识别和间歇反应釜工程应用,使用故障树的分析方法,逐级判断出故障原因,验证了分组趋势预测的有效性:在保证高精度的基础上,输出当前系统出现的故障和发生时间,为工业系统中时滞时变系统对象的故障判别提供了一种有效的解决方案。Time-varying delay system objects which are widely used in industry by the temperature of BSTR for example is studied,and a new method of grouping forecasting is presented.Using this method,fault detection range will be broken into pieces by data clustering,the appropriate fault indicators are given according to the time domain of each group and process feature.Fault identification is achieved by the analysis of sampling data combined characteristics of process equipment.It verifies the effectiveness of grouping trend forecasting combined with batch reactor application using fault tree analysis.It is on the basis of high accuracy and output the breakdown of the current system and the time.It provides an effective solution for the fault identification of the time-varying delay system objects in industrial systems.

关 键 词：故障诊断 时滞时变 间歇反应釜 趋势预测 自动识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]