Process monitoring in principal component subspace: part 2. Fault identification and isolation study

Principal component analysis (PCA) finds wide applications in process monitoring and product quality control. The objectives of this part of the two-part paper are (i) development of fault identification method based on the fault reconstruction results obtained in part 1 paper and (ii) analyses the fault isolatability issue to ascertain if a fault could be uniquely identified. Both the aims are restricted in the principal component subspace (PCS) as in part 1. It is shown the isolatability of the candidate fault set can be explored in prior using only normal process data. Then a reconstruction based faults identification index is proposed. Again the obtained results are illustrated and verified by simulation studies on a double-effective evaporator.