Spectral Correlation Method for Fatigue Crack Detection Based on Nonlinearly Mixed Ultrasonic Waves

Abstract

Abstract Recent developments in the nonlinear ultrasonic wave mixing method have enabled its use in detecting early-stage fatigue cracks. In this method, two input waves at distinct frequencies are applied to a structure with a fatigue crack (i.e., a nonlinear structure). The interaction between these waves gives rise to nonlinearly mixed harmonic components at the sum and difference of the input components. However, the amplitude of the mixed harmonic components generated by the fatigue crack becomes weak in a noisy environment. To overcome this issue, the conventional power spectrum method is used to extract the nonlinear components; however, it is unsuitable for analyzing nonlinear and nonstationary signals. This study presents an advanced signal processing method, called the spectral correlation method, to overcome this problem. The proposed method considers not only the mixed harmonic components in ultrasonic waves, but also the spectral correlation between these components. The uniqueness of the proposed method lies in the following: (1) a single transducer is used, instead of two individual transducers, to generate two input Lamb waves sequentially, (2) the spectral correlation of the mixed harmonic components is insensitive to noise components; (3) tone burst inputs are used as in spectral correlation, and a good contrast between intact and fatigue crack conditions is obtained when compared with the sinusoidal inputs, (4) the mixed components extracted from the spectral correlation method are more reliable than those obtained using conventional methods, and (5) the proposed approach can be applied to detect fatigue cracks. A theoretical model is presented to establish a relationship between the spectral correlation coefficient and the quadratic nonlinearity parameter. Experiments were conducted on aluminum plate specimens with fatigue cracks to validate the performance of the spectral correlation method. The experimental results showed that the nonlinearly mixed harmonic component extracted using our method is more effective for fatigue crack detection than that extracted using the conventional power spectrum method.