The characteristics of impact-echo testing instrument which applied the sound-pressure compression method

摘要

Impact-echo testing instrument that designed with artificial neural networks is presented. Successive impact-echo signals are transformed into frequency spectra, and their intensities &e compressed with taking a human's hearing characteristic into consideration. Experiment was made using bottles with and without cracks of the mouth. The identifying result with the compression rate 0.6 is better than 92 percent correctly, against the result with liner 84.5 percent.