摘要
BP神经网络的分类功能可以用于过程监测与故障诊断,但面对比较复杂的工业过程,数据量过大通常会导致网络的训练时间较长、收敛速度较慢。将主元分析与BP神经网络相结合,提出基于主元分析和BP神经网络理论的工业过程监测方法,根据TE过程数据变量确定神经网络的结构,然后使用TE过程数据训练集训练网络,并建立监测模型。通过对测试结果的分析,认为该模型训练时间短、故障误检率低,具有更好的过程监测效果。
As a classic classification function,BP neural network can be used for process monitoring and fault diagnosis,but in the face of complicated processes,a large amount of process data makes the training time of the network longer and the errors more difficult to converge.This paper determines the structure of the neural network combining principal component analysis with BP neural network,then uses the TE process data training set to train the network,and establishes a monitoring model,and it is applied to the TE process for testing finally.The results show that the model based on principal component analysis and BP neural network method has short training time and low false detection rate,and has better process monitoring effect.
作者
王国柱
路通
池晓航
周强
杜志勇
WANG Guozhu;LU Tong;CHI Xiaohang;ZHOU Qiang;DU Zhiyong(Weihua Group,Changyuan 453400,China;School of Electrical Engineering and Automation,Henan Institute of Technology,Xinxiang 453003,China;School of Electrical Engineering,Zhengzhou University,Zhengzhou 450001,China)
出处
《河南工学院学报》
CAS
2021年第1期17-21,共5页
Journal of Henan Institute of Technology
基金
国家自然科学基金项目(61802116)
河南省科技攻关项目(212102210139,202102210061)
河南工学院高层次人才科研启动基金项目(KQ1806)。
关键词
主元分析
BP神经网络
过程监测
故障诊断
principal component analysis
BP neural network
process monitoring
fault diagnosis
