Eventos Anais de eventos
COBEM 2019
25th International Congress of Mechanical Engineering
Localization of a harmonic excitation load in a bi-clamped beam using finite element method and artificial neural network
Submission Author:
Otavio Duarte Zotelli Boaventura , SP , Brazil
Co-Authors:
Otavio Duarte Zotelli Boaventura, Daniel Henrique de Sousa Obata, Matheus Proença, Amarildo Tabone Paschoalini
Presenter: Otavio Duarte Zotelli Boaventura
doi://10.26678/ABCM.COBEM2019.COB2019-0629
Abstract
When designing, monitoring or evaluating the health of a structure its often critical to know which operational loads it is subjected to. However, sometimes, these loads cannot be measured directly, so estimating them using the structure response may be the only option. There are 3 categories of methods to do it: deterministic, probabilistic, and artificial intelligence based. In this work the third kind of methods was used: an Artificial Neural Network (ANN) trained to find the position of a harmonic concentrated force acting on a beam clamped at both ends. Through the Finite Element Method, the beam's response was simulated, features were extracted from few points, and used to train the ANN. After testing several configurations, the ANN was able to find the excitation position given attributes derived from velocity of any 4 nodes within a certain range, and mean absolute error of 0.2\% was reached.
Keywords
Load reconstruct, Force Estimation, Structural Health Monitoring (SHM), Redes Neurais Artificias, artificial neuronal network
