Wavelets and non-linear principal components analysis for process monitoring

A non-linear principal component analysis (PCA) algorithm is proposed For p rocess performance monitoring based upon an input-training neural network. Prior to assessing the capabilities of the monitoring scheme on an industri al dryer, the data is first pre-processed to remove noise and spikes throug h wavelet de-noising. The wavelet coefficients obtained are used as the inp uts for the non-linear PCA algorithm. Performance monitoring charts with no n-parametric control limits are then applied to identify the occurrence of non-conforming operation prior to interrogating differential contribution p lots to help identify the potential source of the fault. Encouraging result s were achieved. (C) 1999 Elsevier Science Ltd. All rights reserved.