We present a framework for monitoring, diagnosis, and control of indus
trial processes. This framework utilizes the multiresolution analysis
capability of wavelet theory. Wavelet coefficient patterns at differen
t scales, under a variety of process conditions, are noted to form pro
cess fingerprints, these fingerprints yield process fault diagnosis. T
his knowledge facilitates efficient control. (C) 1998 Elsevier Science
Ltd. All rights reserved.