Enhanced Event-Based Identification Procedure for Process Control

Abstract

An enhanced event-based identification procedure for process control has been developed based on the information obtained from the oscillations that an event-based sampler introduces in the feedback loop. The describing function analysis is used to explain the basis of the method because the event-based sampler behaves as a static nonlinearity. Features of the method are (a) the event-based procedure does not require a priori process information, (b) noniterative algorithms are sufficient to derive the process parameters, (c) only one test is needed, and (d) it allows identifying the process at a user-specified phase angle in the third quadrant. The method is presented for estimation of most common transfer functions found in chemical and process industry: first and second order, as well as integrating processes with non-minimum-phase dynamics. Furthermore, the procedure can be extended to estimate models of any structure with five or more parameters.