Development of contact/non-contact nonlinear ultrasonic Lamb wave mixing technique for micro-damage detection and localization

Abstract

Early detection of damage in structures is important to ensure their structural safety. Nonlinear ultrasonic techniques are widely used to measure the quadratic nonlinearity that represents the third-order elastic constants of materials for damage detection. In addition, there are ongoing efforts to exploit both the third- and fourth-order elastic constants that describe the cubic nonlinearity to detect damage. This dissertation develops a nonlinear ultrasonic Lamb mixing technique to measure cubic nonlinearity and to detect and localize micro-damage. The proposed wave mixing technique generates three Lamb waves in the target structure and measures the nonlinear mixed waves produced by nonlinear cross-mixing of the Lamb waves. A theoretical model is developed to describe the generation of nonlinear mixed waves in a nonlinear elastic and homogeneous plate. In addition, a theoretical relationship between the nonlinear mixed components and the fatigue crack size is derived from the Paris-Erdogan and Nazarov-Sutin theories. Experiments were then conducted on aluminum plate specimens with micro-damage to validate the effectiveness of the proposed technique. The position of the micro-damage within the target structure was identified by spatially scanning the mixing zone. In particular, the proposed technique is more sensitive to micro-damage than existing nonlinear ultrasonic techniques like two-wave mixing and third harmonic production. The amplitude of these mixed components weaken in a noisy environment, necessitating noise elimination for a reliable crack detection. To overcome this, a novel hybrid method that incorporates a deep learning (DL) model with higher-order spectral analysis is proposed in this dissertation. The DL model based on long short-term memory (LSTM) takes an original ultrasonic time signal and outputs a reconstructed ultrasonic signal after noise reduction. Finally, random noise in the reconstructed signal is eliminated by trispectrum (TS)-based higher-order spectral analysis. In this dissertation, a non-contact nonlinear Lamb wave mixing technique based on laser line-array excitation was developed for microcrack detection in plate-like structures. Specifically, a pulsed laser with a line-array pattern (LAP) source was created to generate two narrowband Lamb waves with distinctive frequencies

then, a laser Doppler vibrometer (LDV) was used to measure the corresponding ultrasonic responses. The performance of the developed laser ultrasonic system was experimentally validated by applying it to aluminum specimens with microcracks. In this dissertation, the effects of the number of LAP on the wave mixing zone size, as well as the amplitude of the mixed components, are discussed. The obtained results indicate that the proposed system can locate and detect microcrack in a plate by scanning the wave mixing zone.