定性趋势分析在过程故障诊断中的应用研究  

Applied Research of Qualitative Trend Analysis in Process Fault Diagnosis

作　　者：张贝克[1] 张海洋[1] 马昕[1] 

机构地区：[1]北京化工大学信息科学与技术学院,北京100029

出　　处：《系统仿真学报》2008年第10期2750-2753,共4页Journal of System Simulation

基　　金：国家863高技术研究发展计划资助项目(2003AA412310);黑龙江省科技攻关项目(GB06A106)

摘　　要：为了用计算机对传感器数据的瞬态演化进行建模,提出了定性趋势分析(Qualitative Trend Analysis,QTA),趋势是基于三种基元:{不变,上升,下降},用这三个基元组合的七种形状来描述过程的趋势,这种定性的过程趋势描述是一种很有用的方法。系统介绍用于过程故障诊断的定性趋势分析(Qualitative Trend Analysis,QTA)两个重要算法:趋势提取算法和趋势识别算法。为了使得趋势提取算法和趋势识别算法有更大的适应性,采用了分级阈值的办法来动态地获得趋势提取和识别中所需要的阈值。获得过程的趋势后,用一种趋势匹配的策略构架来诊断故障。对于过程工业的故障诊断,定性趋势分析的优点是能够简洁、快速地诊断故障。Qualitative process trend representation is a useful approach to model temporal evolution of sensor data by computer, and the trend bases on the three primitives： {Steady, Increasing, Decreasing}, using these seven shapes combined by three primitives to describe the process trend. Two important algorithm of Qualitative Trend Analysis （QTA） for process fault diagnosis was systematically introduced--trend extraction algorithm and trend identification algorithm. To make the algorithm have larger adaptability, the method of grading threshold was introduced to achieve the threshold dynamically required in trend extraction and identification. A framework of trend match methodology is used to diagnosis fault after capturing the trends. The advantage of Qualitative Trend Analysis （QTA） diagnosis faults compactly and quickly for fault diagnosis of process industry.

关 键 词：定性趋势分析(QTA) 趋势提取 趋势识别 故障检测与诊断 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]