2011
IMAACA 2011 Proceeding
Multi-scale extension of discriminant PLS for fault detection and diagnosis
Authors: Mohammad Sadegh Emami Roodbali, Mehdi Shahbazian
Abstract
A new approach based on the Partial Least Squares (PLS) and Wavelet Transform is presented for the industrial process monitoring. A different scheme for applying PLS for multiple faults diagnosis is used in this approach. Because of multi-scale nature of the variable measurements in the most of industrial processes the Discrete Wavelet Transform (DWT) is applied to extract the multi-scale features of these measurements. Comparison of the ability of this Multi- Scale PLS (MSPLS) algorithm with the PLS to diagnosis the multiple faults in the Tennessee Eastman process (TEP) benchmark, demonstrates the efficiency of the proposed approach and indicates that this MSPLS algorithm can be useful for process monitoring and detection and diagnosis multiple faults.
I3M Scientific Sponsors
I3M Industrial Sponsors
I3M Media Sponsors
