Material properties identification of a piezoelectric beam using inverse method

Abstract

This paper presents an inverse method for material properties identification of a piezoelectric beam (piezoelectric charge and relative dielectric coefficients) using a wavelet-based neural network as an inverse tool. The identification analysis is carried out by using two approaches. In the first approach, i.e. sensor mode analysis, the input data for wavelet-based neural network training are measured voltages at several specific points on the beam's top surface resulting from the applied beam tip deflection. In the second approach, i.e. actuation mode analysis, the input data are values of the beam tip deflection caused by applying voltage on the beam's top surface. In this study, the input parameters employed to train the wavelet-based neural network are obtained using the finite element method. The identification results are compared with those of some conventional neural networks including radial basis function and multilayer perceptron. The results show that the proposed neural network is an efficient tool in the material properties identification problem.

Hysteresis modelling and compensation for piezoelectric actuator using Jaya-BP neural network

“…Zhao et al 14 proposed an Elman neural network-based hysteresis model for the PZT actuator. Nematollahi et al 15 suggested a wavelet-based neural network for inverse identification of piezoelectric beam. Qian et al 16 introduced a hybrid least-squares support vector machine and Bouc-Wen model to model the ratedependent hysteresis of piezoelectric actuators.…”