题目：Design of Piezo-electric Energy Harvesters for Next Generation AIoT
时间：2024年10月16日 13:30-15:00
地点：suncitygroup太阳新城官网 F310会议室
邀请人：高秋华 助理教授（振动、冲击、噪声研究所）
Biography
Dr. Elena Atroshchenko holds a PhD in Civil Engineering from University of Waterloo, Ontario, Canada (2010). At present, she is a Senior Lecturer at School of Civil and Environmental Engineering at the University of New South Wales, Sydney, Australia. Prior to her appointment at UNSW, she was an Assistant Professor at Department of Mechanical Engineering at University of Chile, Santiago, Chile. Dr. Atroshchenko’s expertise is in computational mechanics, numerical modelling and optimization, and scientific machine learning. Her research focuses on such areas as piezo-electric energy harvesting, design of meta-materials and meta-structures, physics informed neural networks.
Abstract
In this seminar, we will introduce the concept of Simultaneous Energy Harvesting and Sensing (SEHS) system, where a single piece of hardware, a Piezo-electric Energy Harvester (PEH) is used for two objectives: harvesting energy from the source vibration and using the produced voltage signal to acquire information about the vibration source. In particular, we are interested in the design of SEHS for structural health monitoring of bridges. To simulate a healthy and damaged bridge response under passing vehicles we use a vehicle bridge interaction model solved with the finite element method. Subsequently, bridge acceleration serves as input to the PEH model to estimate the produced voltage. PEH model is a based on a bimorph Kirchoff-Love (KL) plate attached to a vibrating base. The system is solved using isogeometric analysis. In order to assess structural state of the bridge, convolutional variational auto-encoder (CVAE) is used. Since in real life, labelled data is usually not available, CVAE is trained on voltage data from a healthy bridge only, which can be characterized as unsupervised learning. Next, we perform a bi-objective optimization of a PEH with respect to energy harvesting performance and sensing accuracy, and show that in some cases, there is a Pareto front between these two objectives and in some cases, it is possible to define a unique device with the optimal performance in both objectives.
Finally, we discuss the extension of this study to a meta-structure based design of PEH and explore the possibility to use multi-variate voltage signal from various points inside the structure to enhance the sensing accuracy.
