Dead-time compensation through direct input measurement in systems with unknown time-varying delay

Abstract

In systems with deterministic and constant delay Smith predictor offers a very efficient method for compensating dead-time. However, it is vulnerable to model uncertainties and parameter changes, in particular to delay variations. In this paper, we provide an alternative solution for compensating dead-time in systems with uncertain and time-varying delay. Two compensator structures are proposed - one for linear systems with known nominal delay (and unknown variation pattern), and another for nonlinear plants and/or systems with unknown delay. The proposed solutions retain conceptual simplicity of the traditional Smith compensator and are applicable to a large variety of control problems as demonstrated in case studies.