Structural identification of soft-sensor of MTBE unit

Abstract

The paper deals with the task of estimating methanol concentration in the process of methyl-tert-butyl ether production by soft-sensor. A comparative analysis of the methods for determining of the main informative input sets for the soft-sensor model is presented. The dynamic soft sensor modeling based on the use of convolution sums in order to take into account the time of measurement delay is used. The static and dynamic soft-sensors are evaluated with the use identification criteria. The performance of dynamic estimator is shown to be better than that of the static estimator.