Monitoring and fault detection in processes with multiple operating modes, transitory phases and start-ups using principal component analysis

This paper presents a global monitoring and fault detection approach considering the different operation points, start-ups and transitory states that can appear during plant operation. Stationary states have been monitored using the classical PCA approach and start-ups, while grade transition due to, for example changes in the set-point, are monitored using batch and semi-batch PCA-based methods. These methods have been adapted to deal with this type of problems. Thus, theoretical aspects of PCA and unfolded PCA will be presented and explained. The method developed has been tested in a laboratory pilot plant, using different protocols and grade transitions to examine the method's possibilities with good results.