PROJECT TITLE :

Fault Subspace Selection Approach Combined With Analysis of Relative Changes for Reconstruction Modeling and Multifault Diagnosis

ABSTRACT:

On-line fault diagnosis has been a vital task for industrial processes, which generally is taken once some abnormalities are detected. Reconstruction-based fault diagnosis has been drawing special attention as a sensible alternative to the ancient contribution plot. It identifies the fault cause by finding the precise reconstruction model (i.e., fault subspace) that may well eliminate alarm signals from a bunch of alternatives that have been prepared based mostly on historical fault knowledge. However, in follow, the abnormality could result from the joint effects of multiple faults, which therefore cannot be well corrected by single-fault subspace archived in the historical fault library. During this paper, an aggregative reconstruction-based mostly fault diagnosis strategy is proposed to handle the case where multiple-fault causes jointly contribute to the abnormal process behaviors. First, fault subspaces are extracted based mostly on historical fault information in 2 completely different monitoring subspaces where analysis of relative changes is taken to enclose the most important fault effects that are responsible for various alarm monitoring statistics. Then, a fault subspace choice strategy is developed to investigate the combinatorial fault nature that can type and select the informative fault subspaces by evaluating their significances in information correction. Finally, an aggregative fault subspace is calculated by combining the chosen fault subspaces, which represents the joint effects from multiple faults and works as the ultimate reconstruction model for online fault diagnosis. Theoretical support is framed and the related statistical characteristics are analyzed. Its feasibility and performance are illustrated with simulated multiple faults from the Tennessee Eastman benchmark process.