Data reconciliation for measurements
Résumé
For plant control improvement, coherency of information supplied by instrument lines and sensors must first be ensured. Because of the presence of random and possibly gross errors, the model equations of the process are not generally satisfied. The problem of how to reconcile the measurements so that they satisfy the model constraints and how to use the reconciled values to detect gross errors are considered in this article. The authors strive to summarize various aspects of data reconciliation with the double aim of presenting the state of the art and bringing out the major difficulties encountered in the field. For that, the different steps of the methodology are presented in the following order : data reconciliation techniques, variables classification based on the concept of observability, gross errors detection and localisation, measurement error variance estimation and sensor location.