Fault Detection Diagnosis Of Polymer Extrusion Processes

Project Title:

Fault Detection/ Diagnosis of Polymer Extrusion Processes

Project Description:

The target of these research activities is to establish suitable mathematical models for the process to maintain safety and productivity. Specific research areas include multivariate statistical process control and the development of novel concepts for detecting/diagnosing abnormal process behaviour in complex industrial systems. The involved methods include principal component analysis (PCA), independent component analysis (ICA) and its nonlinear extensions.
In particular, nonlinear extension of PCA (NLPCA) based on RBF networks and principal curves (Fig 1) has proved effective in handling the inherent nonlinearities in the data modelling.

NLPCA%20based%20on%20RBF%20networks%20and%20principal%20curves%20%28a%29%20mapping%20%28b%29%20de-mapping.png

Fig. 1 NLPCA based on RBF networks and principal curves (a) mapping (b) de-mapping