全文下载:2012050546
刘喜梅 ,钟丽丽,步妮
(青岛科技大学 自动化与电子工程学院,山东 青岛 266042)
摘要:针对动态主元分析(DPCA)在故障识别方面的缺点,引入了BP神经网络,将DPCA与BP神经网络相结合,增强了对故障的辨识能力。该方法通过对各测量变量的自相关性分析,来降低动态主元分析中增广矩阵的维数,从而降低了分析过程的计算量。最后,将DPCA-BP应用于田纳西-伊斯曼过程的故障诊断中,验证了所提出方法的有效性。
关键词:DPCA;故障诊断;BP网络
中图分类号: TP 277文献标志码: A
Fault Diagnosis Method Based on DPCA-BP
LIU Xi-mei,ZHONG Li-li,BU Ni
(College of Automation and Electronic Engineering,Qingdao University of Science and Technology , Qingdao 266042, China)
Abstract:BP neural network is introduced to improve the dynamic principal component(DPCA)analysis in fault identification.By analyzing the autocorrelation of the measured variables, the dimension of the augmented matrix in dynamic principal component analysis is reduced, and the difficulty of calculation is decreased. Finally, the proposed DPCA-BP method is verified to be effective on the fault diagnosis by the application in the Tennessee-Eastman process.
Key words:DPCA ; fault diagnosis ; BP neural network
收稿日期:2012-08-03
基金项目:青岛市应用基础研究计划项目(2011).
作者简介:刘喜梅(1961—),女,教授.