Session: 07-17-03: Machine Learning and Artificial Intelligence in Dynamics, Vibrations and Control

Paper Number: 147237

147237 - Deep Learning Enabled Approach for Nonlinear Response Modeling of Piezoelectric Energy Harvesters 

In the past couple of decades, piezoelectric energy harvesting has been extensively investigated to provide a sustainable power supply for small-scale devices, such as wireless sensors, medical implants and wearable devices. Linear piezoelectric energy harvesters (PEHs) have been well studied and understood. However, a linear PEH can only operate within a narrow bandwidth near the structural natural frequency to generate power at a meaningful level. In practical circumstances where the vibration sources are often featured with randomness and spread over a wide spectrum of frequency, the application of linear PEHs is significantly limited. As a solution to this issue, nonlinearities have been introduced to the system in order to broaden the operation bandwidth of linear PEHs. Based on the Duffing stiffness, monostable and multi-stable energy harvesters have been proposed. Meanwhile, nonlinearities could also help improve the energy harvesting efficiency from random vibrations owing to the stochastic resonance. Other alternative solutions for broadband energy harvesting include internal resonance based PEHs, magnetically coupled dual-beam PEHs and nonlinear energy sink-based PEHs. In addition to the nonlinearity introduced by structural designs, the system also has piezoelectric material nonlinearity, geometric nonlinearity, nonlinear coupling between the mechanical and electrical domains, and dielectric loss. Though significant progress has been made in the past two decades, accurate modeling of these nonlinear effects still remains difficult or usually involves parametric identification from experimental results. Thus, the motivation here is to explore and leverage deep learning techniques for modeling behaviors of nonlinear PEHs. To this end, we present a deep learning-based surrogate model for nonlinear response analysis of PEHs with high complexity and strong nonlinearity. The key concept is to establish a long short-term memory (LSTM) neural network to capture the dynamic properties of the piezoelectric energy harvesters from limited training data and infer the voltage and power given unseen excitations in a data-driven fashion without the need of conventional time-consuming numerical simulation or parametric identification. The LSTM overcomes the long-term dependence issue of recurrent neural networks (RNN). It has a similar overall architecture with repeating modules that accommodate sequence data. Each of the repeating modules now includes multiple layers and gates that allow the network to learn longer-term dependencies. Specifically, the set of gates are used to control when information enters the memory, when it is output, and when it is forgotten. In short, the LSTM network is capable of learning order dependence in sequence problems and capturing long-term, non-linear temporal dependencies between the input and out of a system, making it suitable for system dynamics modeling to build the relationship between the excitation and response. The performance of the proposed approach was successfully demonstrated through numerical data of a monostable PEH of Duffing stiffness nonlinearity and piezoelectric material nonlinearity. The results indicate that the proposed framework is a promising, reliable and computationally efficient approach for nonlinear response modeling of complex piezoelectric energy harvesters.

Presenting Author: Yabin Liao Embry-Riddle Aeronautical University

Presenting Author Biography: Dr. Yabin Liao received his B.E. degree in automotive engineering from Tsinghua University, Beijing, China, in 1999. After that, he received the M.S.E. degree in electrical engineering in 2004, and Ph.D. degree in mechanical engineering in 2005, both from Arizona State University in Tempe, Arizona, USA. Dr. Liao taught at Arizona State University between 2006 and 2017, and then joined Penn State Erie as an Assistant Professor of Engineering. After that, he joined Embry‑Riddle Aeronautical University-Prescott as an Assistant Professor of Aerospace Engineering in 2021. His research interests include smart materials, energy harvesting, structural dynamics, signal processing, and machine learning. Dr. Liao is a member of the Energy Harvesting Technical Committee under the ASME Aerospace Division Adaptive Structures & Material Systems (ASMS) Branch.

Authors:

Dubai Li Nanjing University
Yabin Liao Embry-Riddle Aeronautical University
Ruiyang Zhang Southeast University
Chunbo Lan Nanjing University of Aeronautics and Astronautics