Eventos Anais de eventos
COBEM 2023
27th International Congress of Mechanical Engineering
A NEW APPROACH USING TRANSMISSIBILITY AND KERNEL PRINCIPAL COMPONENT ANALYSIS TO DETECT DAMAGE FOR A NONLINEAR STRUCTURE WITH UNCERTAINTIES
Submission Author:
Wellington de Lima Nogueira , SP
Co-Authors:
Wellington de Lima Nogueira, Eloi Figueiredo, Samuel da Silva
Presenter: Wellington de Lima Nogueira
doi://10.26678/ABCM.COBEM2023.COB2023-2401
Abstract
Structural health monitoring (SHM) has emerged as a promising tool for detecting and managing damages in structures and systems. However, SHM becomes more challenging when uncertainties and the intrinsic nonlinear behavior of systems and structures are taken into account. This is because nonlinear phenomena can be mistakenly identified as damage when classical SHM techniques, which rely on linear metrics, are employed. Therefore, this study aims to address this issue by detecting early-stage damages in systems with nonlinear behavior while considering operational and environmental uncertainties. To convert these vast amounts of data into meaningful information, approaches based on the Statistical Pattern Recognition (SPR) paradigm are utilized. In this regard, a proposed method employs measures of transmissibility obtained from an experimental nonlinear beam over multiple days of measurement as a consolidated output. Kernel Principal Component Analysis (KPCA) is then utilized to extract and classify features. For the feature extraction phase, the KPCA algorithm is used to reduce the dimensionality of the transmissibility measurements and for the feature classification phase a new transmissibility should be mapped onto the high kernel space. The vibration data utilized in the study were acquired from an experimental setup consisting of a beam constructed by connecting Lexan layers. To simulate the propagation of damage with nonlinear quadratic behavior in the response, a breathing crack was intentionally introduced into the structure. Furthermore, to introduce additional cubic nonlinearities, two fixed steel masses were attached to the free end of the beam, which interacted with a magnet. The results of this study demonstrate the effectiveness of the proposed approach in damage detection, outperforming traditional Principal Component Analysis (PCA) techniques. This approach presents a promising solution for detecting damages in complex systems characterized by multiple nonlinearities and uncertainties. By improving the reliability and efficiency of structural health monitoring, it can find broad applications across various industries.
Keywords
Structural Health Monitoring (SHM), nonlinear behaviour, Transmissibility measurement, Kernel Principal Component Analysis
