Session: 07-19-01 Machine Learning and Artificial Intelligence in Dynamics and Vibrations I

Paper Number: 69392

Start Time: Tuesday, 06:40 PM

69392 - An Artificial Neural Network Model for Flexoelectric Actuation and Control of Beams 

In recent years, the development of smart materials provides many opportunities for the application of vibration control. The use of flexoelectric materials has shown great promise in the vibration behavior of nanostructures. Based on the converse flexoelectric effect, the inhomogeneous electric field induces internal stress in flexoelectric materials, which can be applied to precision actuation and vibration control of flexible structures. An atomic force microscope (AFM) probe placed on top of a flexoelectric patch can generate a nonuniform electric field with an excitation control voltage, which leads to stress in the flexible patch. Because of high stress concentration of a single-actuator, the influence can be alleviated by placing multiple actuators on the structure. The presented study introduces a new intelligent methodology to actuate and control the flexible cantilever beam. According to the position of the flexible electric actuator and the displacement data of the tip of the cantilever beam, a neural network model was established to optimize the position of the multiple actuators. Artificial neural network is a feasible computational model for distributed vibration control position optimization. The use of the multiple flexoelectric actuators excited by an AFM probe for active control of the beam is discussed. When it is trained with several sets of actuator position coordinates and corresponding tip displacement data of beam, the neural network can recognize the relationship of actuator position and tip displacement and forecast the tip displacement of the beam accurately. This neural network algorithm is applicable to the prediction of future response of the cantilever beam actuated by one or more flexoelectric actuators. In the case study, for the first three modes, the tip displacement of the beam actuated by three actuators, four actuators, and five actuators are predicted by using the neural network under resonance conditions where only one mode participates, and the validity of the neural network model is proved by comparing with the theoretical calculation results, while with very much reduced computational effort and higher effectiveness. Then, by using the neural network, the displacement data generated by all possible combinations of actuator positions are predicted. Finally, the optimal position of the flexible electric actuator can be obtained for modes1,2 and 3. At this time, flexoelectric actuators can make the beam produce the maximum displacement with the smallest excitation voltage. The neural network algorithm provides a fast and effective way to solve the distributed vibration control optimization problem of flexoelectric intelligent structures.

Presenting Author: Fan Mu Nanjing University of Aeronautics and Astronautics

Authors:

Yu Pengcheng Nanjing University of Aeronautics and Astronautics
Fu Xiaogang Shanghai Aerospace Control Technology Institute
Fan Mu Nanjing University of Aeronautics and Astronautics