Detection of internal defects of material on the basis of performance spectral density analysis

O. Krejcar, R. Frischer

Journal of Vibroengineering, Vol. 12, Issue 4, 2010, p. 541-551.
Received 20 September 2010; accepted 9 December 2010; published 31 December 2010

Copyright © 2010 Vibroengineering This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Creative Commons License
Abstract. The developed approach for nondestructive diagnosis of solid objects is described in this article. This is accomplished by means of software analysis of oscillatory spectra (possibly acoustic emissions,) which is formed while running the monitored device or in unexpected situations. The principle of this method is based on the analysis of spectrum from received signal, its subsequent processing in MatLab and following sample comparison in Statistica program. The last step (comparison of samples) is the most important because it enables determination with some certainty the actual condition of the examined object. The processed samples are currently compared only visually. On the other hand, in applying this approach they are subject to the analysis with the assistance of neural network (Statistica program). If correct and high-quality input data are provided to initial network, it is capable of analyzing other samples and identifying the actual condition of certain object with success rate of around 70% (minimum 70%). The instructed neural network is then able to determine whether any critical condition occurred (e.g. escape of gas from burst pipe, loosened screws in critical places, etc.). Data from accelerometer (microphone) are evaluated with the assistance of MatLab program and a special newly defined filter is implemented. This filter ensures extraction of relevant data. Analogue signal is digitized with the help of special NI I/O module PCI 6221. The correlation of the spectrum and the condition of examined material (its internal defects) are obvious after implementation of FFT. Finally, it is possible to detect defects or upcoming hazardous conditions of examined object/material by using only one device which contains HW and SW parts. This kind of detection can lead to significant financial savings in certain cases (such as continuous casting of iron, which saves hundred of thousands of USD)

Keywords: FFT, Power Spectrum, MatLab, Statistica, Defect .