OFF-LINE DIAGNOSIS OF DETERMINISTIC FAULTS IN CONTINUOUS DYNAMIC MULTIVARIABLE PROCESSES USING SPEECH RECOGNITION METHODS

Faults or special events which occur occasionally in continuous proces ses generate dynamic patterns in a large number of process variables. However, the patterns arising from the same fault can exhibit differen t time durations (depending on the operating conditions), magnitudes a nd directions. Any robust fault diagnosis method must be able to corre ctly classify these faults under these different conditions. This pape r presents an off-line fault diagnosis method based on pattern recogni tion principles for multivariate dynamic data. The method consist of a filtering and scaling step, where the magnitude dependent information is removed, and a similarity assessment step via dynamic time warping (DTW). DTW is a flexible pattern matching method used in the area of speech recognition. The method presented in this paper is designed to classify faults independently of their magnitude, duration, direction and plant production level. As a further feature extraction step, prin cipal component analysis is used to reduce the dimension of the multiv ariate problem and enhance the distance-based classification. Case stu dies from the Tennessee-Eastman plant are used to test the method and to illustrate its advantages and limitations. (C) 1998 Published by El sevier Science Ltd. All rights reserved.