基于DPCA方法的流程工业系统故障检测  

Fault Detection of Process Industry System Based on DPCA Method

作　　者：亚森江·加入拉 苏佩建 许晨星 Yasenjiang Jiarula;Su Peijian;Xu Chenxing(College of Mechanical Engineering,Xinjiang University,Urumqi 830017,China)

机构地区：[1]新疆大学机械工程学院,乌鲁木齐830017

出　　处：《煤矿机械》2022年第3期166-168,共3页Coal Mine Machinery

基　　金：国家自然科学基金项目(52065065);新疆大学博士科研启动基金项目(BS190216)。

摘　　要：故障检测在流程工业系统故障诊断中有着至关重要的作用,是设备正常运行的重要保障,及时准确发现并排除故障对于安全生产意义重大。运用动态主成分分析(DPCA)方法对田纳西·伊斯曼(TE)过程中的几种典型故障进行检测。通过对原始数据矩阵添加滞后变量,利用正常样本计算出的平方预测误差(SPE)统计量的阈值,对所有故障数据进行检测。仿真实验表明,使用DPCA方法的故障检测比传统的主成分分析(PCA)方法有更高的准确率。Fault detection plays an important role in the fault diagnosis of process industry system.It is an important guarantee for the normal operation of equipment.It is of great significance for safety production to find and eliminate faults in time and accurately.The dynamic principal component analysis(DPCA)method was used to detect several typical faults in Tennessee Eastman(TE)process.By adding lag variables to the original data matrix,calculated the threshold of squared prediction error(SPE)statistics from normal samples,all fault data were detected.The simulation experiment shows that the fault detection using DPCA method has higher accuracy than the traditional principal component analysis(PCA)method.

关 键 词：流程工业系统 故障检测 DPCA 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]