Extended Self-tuning Generalized Predictive Control with Computation Reduction Focused on Closed-loop Characteristics

Abstract

Abstract This paper proposes a design method of extended self-tuning generalized predictive control (GPC) with computation reduction focused on closed-loop characteristics. The authors have extended GPC by coprime factorization and proposed the extended controller for constructing a strongly stable system. Moreover, the proposed controller is able to be designed to make the same steady state output as pre-designed system's steady state output even if feedback loop is cut. Although self-tuning controller is one of the control methods for systems with uncertainty, there is a problem that the computation of self-tuning GPC increases as design engineer takes long prediction horizon in the design parameters. Therefore this paper considers computation reduction for extended self-tuning GPC focused on closed-loop characteristics. The validity of the proposed method is shown by numerical simulation.