改进动态主元分析的工业过程故障预警研究

传统动态主元分析(DPCA)进行工业过程故障预警时,对所有变量选择相同时间间隔。为克服DPCA中没有考虑到变量延迟、动态变化速度不同的问题,采用变量延迟对齐、时间间隔可变等方法,对DPCA中扩展矩阵的组成方法进行改进。数值仿真结果表明,改进DPCA可以有效减少故障漏报。将该方法应用于原油初馏过程故障预警,在准确预警故障的基础上减少了漏报。

Fault Early Warning of Industrial Process Based on Improved Dynamic Principal Component Analysis

The traditional dynamic principal component analysis(DPCA)selects the same time interval for all variables in the process of industrial process fault early warning.In order to overcome the problems of variable delay and different speed of dynamic change that are not considered in DPCA,the method of variable delay alignment and variable time interval is adopted to improve the composition of extended matrix in DPCA in this paper.The numerical simulation results show that the improved DPCA can effectively reduce the missing alarms.This method is applied to the early warning of crude oil initial distillation process,and the results show that the missing alarms are reduced on the basis of accurate early warning.