On Stiction Compensation Methods for Practical Nonlinear MPC Implementations

Abstract

Stiction in actuators is a common issue within the process industry, which may considerably degrade the control performance in practice. In particular, the improved prediction capabilities of model predictive control (MPC) vanish if such actuator imperfections are neglected in actual implementations. In this work, the authors review the more recent approaches for stiction compensation in literature and propose two practical alternatives that are more suitable to use in large-scale nonlinear MPC problems. The proposals are illustrated and tested over a real industrial case study: the fiber humidity control in a medium-density fibreboard dryer.

Enhancing MPC formulations by identification and estimation of valve stiction

“…which include integral action via disturbance estimation) also show a similar oscillating behavior, unless a dynamic model of the actuator malfunction is taken into account in the controller formulation. Indeed, stiction compensation via MPC and other advanced control methods is an area of recent activity, as witnessed by other several works published in the last years [5,6,7,8,9,10].…”