Two-phase dual-type optimising control algorithm for uncertain plants with active output constraints

Abstract

An algorithm for steady-state optimising control of technological processes within a multilayer control structure is considered in the paper, applicable to the situation when only an approximate and rather rough steady-state process model is available and disturbance estimates are uncertain. It is known that in such cases a single process model optimisation can lead only to very suboptimal results. The algorithm performs iterative set-point optimisation (improvement) using the model and process measurement information to cope with the uncertainty. It belongs to the ISOPE class (Integrated System Optimisation and Parameter Estimation). Further, a case with active constraints on process outputs vital for economical efficiency is considered, with an additional dedicated feedback correction controller used to keep the constraints satisfied. A two-phase dual-type ISOPE algorithm derived for the considered control structure is presented in the paper. It removes the major drawback of the available dual algorithm — inefficiency of the initial phase. Simulation results on a case study example of distillation columns in series are presented, for both new and previous algorithms.