An Artificial Neural Network Model for Multiple Piezoelectric Actuation of Plates

Abstract

With the increasing use of piezoelectric smart structures in various fields, the search for further optimization of the devices is inevitable. In this work, both artificial neural network (ANN) models and finite element models were used to explore the effect of piezoelectric patch size and numbers on the first order natural frequency and transverse displacement of a plate. The research objective is to investigate the efficiency and factors influencing the actuation of thin plate structures under single/multi-channel piezoelectric control conditions. A Finite element model built with COMSOL was used to analyze the effect of structural parameters, including the number of channels and other key parameters on the control of the main natural frequencies of the thin plate structure. With the obtained data from finite element simulations, an ANN model was used to predict the dynamic response of the plate, including the first-order natural frequency and displacement amplitude for finite element verification, so that structural optimization design can be achieved. A well-trained neural network model can quickly and efficiently predict the displacement amplitude and natural frequency of a piezoelectric driven rectangular plate. This study provides a convenient and effective method for predicting the dynamic response of a piezoelectric drive plate considering the number of piezoelectric patch channels and size effects.