Acoustic Emission Source Characterization Through Direct Time-Domain Deconvolution

Hsu, N
Eitzen, D
Major Professor
Committee Member
Journal Title
Journal ISSN
Volume Title
Research Projects
Organizational Units
Journal Issue

While detected acoustic emission (AE) signals contain potentially useful information about the deformation source mechanisms of a structure under load, signal processing techniques such as threshold counting, RMS recording, energy measurement, peak detection, and spectral analysis often fail to extract such information unambiguously. The difficulty lies both in the inherent complexity of the deformation mechanism and in the lack of understanding of the source mechanism, the wave propagation details, and the physics of the sensor's mechanical-to-electrical conversion process. Instead of taking an empirical approach to establish the correlations between the detected AE and the observed possible deformation mechanism, we approach the problem by constructing a simple test system which consists of three main ingredients: a true displacement sensor (capacitive transducer), a simple structure (either a large block or a plate), and known theoretical impulse-response functions for specific sensor-source relative locations. We first establish the validity of these ingredients by testing with simulated AE of known step-function time dependency generated by breaking glass capillaries. Unknown sources are then introduced, one at a time, into the system for determination of their time functions. The time function at the source is determined by a deconvolution process from the known impulse response and the detected displacement. Furthermore, we show the existence of the inverse of the impulse-response function with respect to convolution for at least two extreme cases. Consequently, the source function can be obtained simply by convolving the detected signal with the inverse function. Applications to AE system calibration, sensor characterization, wave propagation studies, and brittle crack opening signature analysis will be demonstrated.