Precision control of a piezoceramic actuator using neural networks

This paper is to study the precision control of a piezoceramic actuator (PA). Due to the inherent hysteresis nonlinearity, the PA always causes position error in the open-loop system and instability in the closed-loop system. To remedy this problem, a new control method combining the feedforward and feedback controllers is proposed to improve the dynamic performance of the PA. In the feedforward controller design, the hysteresis nonlinearity of the PA is modeled by using Preisach model first. A database of input/output history and a neural networks architecture treated as the inverse function of Preisach model are also utilized in the feedforward controller. In the feedback controller design, a PI controller is used to regulate the error between the command input and system output. In experiment, the command of square wave, sinusoid wave and triangular wave is taken as the tested signal to validate the excellent performances of the proposed controller.