An efficient methodology to select high-performance M-estimators for robust data reconciliation

LLANOS, CLAUDIA E.; SÁNCHEZ, MABEL C.

COMPUTERS AND CHEMICAL ENGINEERING

PERGAMON-ELSEVIER SCIENCE LTD

Año: 2023 vol. 176

0098-1354

Within the framework ofChemical Engineering, many exponential M-estimators have been proposed andtested. As usual, extensive simulation procedures are run to compare theirperformances. This work presents a deterministic methodology that reduces thecomputational endeavor of the comparative analysis. It becomes an effectivetechnique for proposing and selecting high-performance M-estimators. Thestrategy uses the weight functions’ standardization to provide a common basisfor the posterior application of Equivalent M-estimators and Tails Discrepancydefinitions. In this way, identical M-estimators are figured out and those thatbehave likewise are identified. Whichever robust estimation procedure appliedto estimate process variables and parameters can take advantage of themethodology results. For validation purposes, its conclusions are contrasted tothe traditional performance measures obtained solving robust datareconciliation problems for two chemical processes.