A BAYESIAN BIAS UPDATING PROCEDURE FOR AUTOMATIC ADAPTATION OF SOFT SENSORS

SANGOI, EMMANUEL; SANSEVERINATTI, CARLOS I.; CLEMENTI, LUIS A.; VEGA, JORGE R.

COMPUTERS AND CHEMICAL ENGINEERING

PERGAMON-ELSEVIER SCIENCE LTD

Año: 2021 vol. 147

0098-1354

Straightforward bias updating procedures for online adjustment of soft sensors (SS) are of interest for many industrial processes in which quality and / or production variables cannot be measured online. This work proposes a simple Bayesian strategy for automatically updating the bias of a SS from: (i) online measurements of typical process variables and (ii) sporadic laboratory measurements of the critical variable to be estimated. The method continuously monitors the mean and standard deviation of the prediction error (i.e., the difference between the current laboratory value and the SS output), and produces a self-adaptation of the bias without requiring the adjustment of unknown parameters. The proposal was evaluated on the basis of simulated examples of an industrial continuous process for the production of Styrene-Butadiene Rubber (SBR). Results are promissory, since the estimates are similar to those obtained with classical methods where their parameters were adjusted ad-hoc for the investigated example. Moreover, the bias obtained with the Bayesian approach exhibits lower variability and therefore is preferable in the industrial practice.