Identification of Acoustic Emission Sources by Pattern Recognition Techniques [chapter]

R. K. Elsley, L. J. Graham
1983 Review of Progress in Quantitative Nondestructive Evaluation  
Computer pattern recognition has been used to identify and separate acoustic emission (AE) signals that are similar in appearance but are due to different sources. Simulated joint specimens were tested in the laboratory in which a fatigue crack was grown from the edge of a central loading pin hole. The hardened steel loading pin produced fretting AE by its contact with the 7075 T65l aluminum plate specimens during cyclic loading. The fatigue crack produced AE due to crack growth and to crack
more » ... wth and to crack face rubbing during load cycling. The AE signals detected at two transducers mounted on opposite sides of the loading pin hole, at 2 in. and 4 in. from the fatigue crack, were .digitally recorded at a 5 MHz digitization rate. The waveform features that were extracted from these AE signals and used in the pattern recognition were derived from the frequency spectral content of the waveforms. Better than 90% separation of crack growth from crack face rubbing was achieved using frequency features of the waveforms from either transducer separately. Better than 95% separation of fretting from crack growth or crack face rubbing, separately or combined, was achieved using the ratios of the spectral energies detected at the two transducers.
doi:10.1007/978-1-4613-3706-5_28 fatcat:vmvl62s6orhvjhqsdlodybudau