Preditor de Smith Filtrado LPV para Processos Não Lineares

Abstract

In this paper, we propose a Filtered Smith Predictor (FSP) structure for nonlinear systems subject to dead time, considering the Linear Parameter Varying (LPV) framework. Our approach focuses on single-input, single-output (SISO) nonlinear processes represented by LPV models scheduled by exogenous variables and explicit dead time in the output variable. In particular, the control structure is designed using the same tuning methods as in the linear time-invariant (LTI) case, which is enabled by assuming that the scheduling parameters remains constant (for design purposes only). The tuning procedure is direct from the available LPV scheduling parameters during the implementation. Furthermore, specific tuning conditions are imposed to ensure internal stability of the scheme and established certain desired performance objectives. We compare our scheme with two standard LTI FSPs, an numerical simulations indicate better performance and overall robustness.