A hybrid model of Prandtl-Ishlinskii operator and neural network for hysteresis compensation in piezoelectric actuators

Article type: Research Article

Authors: Chen, Yuansheng^a | Qiu, Jinhao^{a; *} | Sun, Hongjun^a

Affiliations: [a] Key Laboratory of Mechanics and Control of Mechanical Structures, Nanjing University of Aeronautics and Astronautics, Nanjing, Jiangsu, China

Correspondence: [*] Corresponding author: Jinhao Qiu, Key Laboratory of Mechanics and Control of Mechanical Structures, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, Jiangsu, China. Tel.: +86 25 8489 1123; Fax: +86 25 8489 1123; E-mail: qiu@nuaa.edu.cn

Abstract: A hybrid model is proposed with the Prandtl-Ishlinskii operator and feedforward neural network. The hybrid model consists of three layers, where play operator is the transfer function of first layer. The training algorithm for model identification is derived from the backpropagation algorithm. With the measured data of piezoelectric actuator, a simulation is performed and shows that the hybrid model can precisely describe the asymmetric hysteresis loops. Incorporating the hybrid model into a feedforward controller, experiments indicate that the hybrid model exhibits a good performance for real-time control and the maximal tracking error is reduced by 76.75% with feedforawrd and PID controller.

Keywords: Hysteresis modeling, piezoelectric actuators, Prandtl-Ishlinskii model, neural network

DOI: 10.3233/JAE-131648

Journal: International Journal of Applied Electromagnetics and Mechanics, vol. 41, no. 3, pp. 335-347, 2013