Availability of reliable process information in real time is key in any decision-making procedure. Thus, good industrial decision-support implementations require dealing with gross errors and consideration of process transients in order to get a set of measurements which will be coherent with the basic underlying process dynamics. This report presents dynamic data reconciliation methods and tools adapted to the requirements of industrial environments (large-scale systems and noisy/faulty data). Moreover, basic concepts in literature are extended to artificially increase system redundancy as well as to cope with time-varying parameter estimation. The procedure summarized in this report has been tested in the Lenzing case study.
Except where otherwise noted, this item's license is described as Attribution-ShareAlike 4.0 International