• Plataforma de vídeos del IIT
  • Twitter
  • LinkedIn
  • Asociación de Ingenieros del ICAI
  • Intranet
  • English
Ir arriba
Información del artículo en conferencia

Global fault detection in adhesive joints using multivariate statistical analysis

Y. Ballesteros, R. Caro, J. Rodríguez, J.C. del Real-Romero, A. Zapico

Non-destructive testing (NDT) can be applied to detect and localize structural faults by using a signal with a wavelength smaller than the fault. Some NDT techniques like, e.g. Ultrasonic Testing, require analysing the whole structural component in small sections to detect the fault. Global Fault Detection techniques, e.g. Frequency Response Functions (FRFs) or Sound Pressure Level (SPL), require only a global measurement in the structural component in service. In previous work [1] it was shown how it is possible to perform Global Fault Detection techniques to detect adhesion failures in aluminum-acrylic adhesive bonded joints, where the FRF values were used as an input into an artificial supervised neural network that classifies the beams in four clusters depending on the damage type in the joint. Damage in structures causes small changes in the FRF resonances in a vibration test. In the present work, SPL signals obtained with the Impulse Excitation technique in 5 groups of aluminium-acrylic adhesive samples. Artificial failures were achieved in these groups during the bonding process. The studied groups were (G1, 100% no failure, G2, 10% unbonded, G3, 25% unbonded, G4, 50% unbonded, G5, 75% unbonded). Ten samples where achieved for each group, whose characteristic parameters; weight, length, width, etc., were registered. Based on the SPL results, the fundamental frequencies of each sample were obtained (six frequencies between 100 and 8000 Hz). Statistical data analysis were used to identify the group or cluster to which a sample belongs in function of their fundamental frequencies, also it can be classified as healthy or faulty. In addition it is possible to predict the amount of damage. The multivariate exploratory analysis has been used as statistical technique, in particular discriminant analysis. In this case the variables that separate groups depending on the fault type in the joint are the first frequency and the second one. Similarly, within each group it was examined how different variables (length, width, weight and thickness, etc.) affect the value of each frequency, using the statistical technique of multiple linear regression analysis.


2nd International Conference on Structural Adhesive Bonding - AB2013, Oporto, Norte (Portugal). 04 julio 2013

Fecha de publicación: julio 2013.



Cita:
Y. Ballesteros, R. Caro, J. Rodríguez, J.C. del Real-Romero, A. Zapico, Global fault detection in adhesive joints using multivariate statistical analysis, 2nd International Conference on Structural Adhesive Bonding - AB2013, Oporto, Portugal, 04-05 Julio 2013.


    Líneas de investigación:

IIT-13-087A_abstract