Vibroacoustic modulation method with arbitrary probe wave frequency based on machine learning algorithms

Abstract

ABSTRACT The vibro-acoustic modulation (VAM) method assesses the integrity of a structure using modulation of a high-frequency probe wave and a low-frequency pump wave and measuring the modulation index. A large body of studies found that the modulation index could be used for the defection of contact-type defects in different materials. However, it was also found that the performance of this method is dependent on the probe wave frequency selection. To enhance the robustness of this method, researchers made efforts to modify the VAM method raising the complexity of the equipment and related operator skills, creating needs for extra preparation or procedures. All of these limits VAM potential for industrial application. To develop a robust, efficient, and fast VAM-based non-destructive testing method, the machine learning algorithm was employed. The algorithm utilises amplitude of the multiple sidebands of the modulation spectrum as feature variables and the structural integrity as target variables for training and testing the machine learning model. The experimental result shows that the fatigue crack could be detected with an arbitrary probe wave frequency using the applied machine learning approach.