摘要
针对线性PCA方法难以提取非线性统计特征信息,本文在输入训练神经网络基础上提出了一种非线性PCA故障检测方法。同时探讨了非线性PCA过程性能监视系统的设计方案及其在间歇生产过程中的应用,仿真实验结果证明算法的有效性。
In this paper, a nonlinear PCA (principal component analysis) method for fault detection based on input-train neural network (IT-net) is proposed for the reason that linear PCA method is difficult to capture nonlinear statistical characteristic information. Design scheme of process perform monitoring system based on the algorithm and its application to batch processes are discussed. The algorithm validity is demonstrated by simulation results.
出处
《系统仿真学报》
CAS
CSCD
2001年第z1期149-151,共3页
Journal of System Simulation
关键词
神经网络
PCA
性能监视
故障检测
neural network
PCA
performance monitoring
fault detection