基于改进PCA的复杂多阶段过程故障检测

Fault Detection of Complex Multi-stage Process Based on Improved PCA

为解决复杂多阶段过程难以进行有效监控的问题,提出了时空近邻标准化和主成分分析的故障检测方法。TSNS使用从时间到空间上嵌套近邻集的均值和标准差标准化各阶段样本,能够将多阶段数据高斯化为单一阶段的数据,分离故障样本,同时去除前后时刻样本间的时序相关性。TSNS能帮助PCA有效克服动态性、非线性和多阶段特征的影响,显著提高PCA的故障检测率。通过使用青霉素发酵过程设计故障检测实验,将TSNSPCA与一些经典方法进行比较,发现其具有更高的故障检测率。

To solve the problem that it is difficult to monitor complex multi-stage processes effectively,the time-space nearest neighborhood standardization and principal component analysis(TSNS-PCA)method is proposed for fault detection.TSNS uses the means and standard deviations of the nested neighbor set from time to space to standardize the samples of each stage.TSNS can Gaussianize the multi-stage data into single-stage data,separate the fault samples,and remove the temporal correlation between samples at previous and subsequent moments in the process data.TSNS helps PCA to overcome the effects of dynamics,non-linearities,and multi-phase features effectively,significantly improving the fault detection rate of PCA.Through the fault detection experiment of the penicillin fermentation process,TSNS-PCA is compared with some classical methods,which proves that it has a higher fault detection rate in complex multi-stage processes.