http://eprints.gla.ac.uk/613/1/Comp_Chem4.pdf

A self-validating control system based approach to plant fault detection and diagnosis

One problem facing the process industries is process monitoring in the presence of random and biased instrumentation errors. One method is to reconcile the data to close a constraining steady-state model before calculating indices such as heat transfer coefficients. Another problem is identifying the steady-state and process variable values due to instrument noise. Procedures have been developed for online steady-state identification and for data reconciliation and error detection. Both procedures can use the same set of easily calculated statistics. The procedures are illustrated with a simple example.

J.E. Albers

Papers

Online data reconciliation and error detection