A system identification approach to the characterization and control of a piezoelectric tube actuator

Abstract

This paper addresses the sensorless control of piezoelectric tube actuators to avoid the expense and practical limits of displacement sensors in nanopositioning applications. Three electrical signals have traditionally been used to estimate displacement: the piezoelectric voltage, the voltage induced in sensing electrodes and the voltage across a sensing resistor. In this work, the piezoelectric voltage was employed to estimate displacement; the use of this signal does not necessitate drift removal like the sensing voltage, and its superiority over the induced voltage is shown in this paper. The piezoelectric voltage is the actuating signal, so a feedforward architecture based on an inverse model is used for sensorless control. Inspired by internal model control (IMC), a filter together with the inverted model of the system, derived using system identification techniques, was used as the feedforward controller. The fixed-slope-input effect is illustrated as a prominent source of control error in tracking triangular references, then an additional nonlinear control command is proposed to address this effect and improve the control performance.