Eventos Anais de eventos
COBEM 2023
27th International Congress of Mechanical Engineering
Machine Learning and Electromechanical Impedance Applied in the Structural Health Monitoring.
Submission Author:
Daniel Ferreira Gonçalves , MG
Co-Authors:
Daniel Ferreira Gonçalves, Fernanda Beatriz Aires de Freitas, Aldemir Ap Cavalini Jr, Roberto Mendes Finzi Neto, Valder Steffen Jr
Presenter: Daniel Ferreira Gonçalves
doi://10.26678/ABCM.COBEM2023.COB2023-1091
Abstract
The advancement of industrial systems, machine learning techniques to monitoring and damage diagnosis have attracted much attention and are widely used in engineering projects. The development of smart systems provides real-time analysis of large volumes of data and important insights to optimize, reduce costs and improve product quality. The objective of this study is to propose a smart system to assess structural integrity using electromechanical impedance. This model is constituted by supervised and unsupervised machine learning tools. The strategy adopted for the choice of resources for the data matrix and the data pre-processing step play an important role in the modeling process and, consequently, have a direct impact on the performance of the clustering and classifier algorithms. Primary a fitted Gaussian Mixture Model is applied to divide the samples into two main clusters: normal or damage. Then a Support Vector Machine classifier was used in the damage cluster to identify the different types of damage. Metrics such as Recall, Accuracy and F1 Score were used to evaluate the model performance. Preliminary results enable the use of the proposed model as a promising methodology for monitoring structural integrity.
Keywords
smart system, electromechanical impedance, Gaussian mixture model, Support vector machine, Damage diagnosis
