|M.Sc Student||Zingerman Anna|
|Subject||Applications of Wavelets Entrophy in Nondestructive-Testing|
|Department||Department of Quality Assurance and Reliability||Supervisor||Dr. Phineas Dickstein|
In-service structural adhesive joints used in the aeronautical and automobile industries undergo a gradual degradation in their performance due to environmental factors. Water is one of the most prevalent and damaging environments.
The most promising non-destructive techniques for the evaluation of composite structures and composite adhesive joints are ultrasonic techniques. A difficulty arises from the fact that a variety of bulk material parameters and geometrical factors can be inferred from the characteristics of sound propagation in a bonded structure.
One approach for the quantitative acquisition of discriminatory information for the classification of bonded specimens is feature-based analysis: representative features are derived from the signals, by means of statistical tests, pattern-recognition and application of advanced signal analysis techniques.
In the last decades new transforms were developed, namely the Wavelet Transforms.
Wavelet decomposition introduces the notion of scale as an alternative to frequency and maps a signal into a time-scale plane.
The goal of this study was to examine the possibility of obtaining useful information or sensitive features from the presentations of the signals in the wavelet domain.
To achieve this goal, experimental ultrasonic signals from two different classes of composite adhesive joints were subjected to several families of Wavelet transforms. The joints had been immersed in circulating hot water and tested periodically both destructively and non-destructively.
The numerous applications of the various wavelet transforms in this study do not improve the capabilities of deciphering the condition of the joints. However the well known capabilities of the wavelet transforms data compression were demonstrated once again.
The entropy of the signals subjected to the wavelet transforms exhibit a certain pattern that does provide quantitative information as to the condition of the adhesive joint.
The entropy calculations based on the wavelets transforms of the signals agree with equivalent calculations of entropy-related values obtained through zero-crossing analysis of the signals.