One of the factors affecting the performance of the acoustic emission method used for real-time monitoring of structures is the interference of noise signals with the damage-related AE signals. Since the signals detected from the sensors are analyzed to obtain damage information, noise-related signals cause inaccurate inspection results. Especially, in the conditions where electrical or mechanical friction is prominent, noise causes big data and errors in damage detection
In this study, to solve this problem, it is aimed to develop an AE signal filtering model by artificial intelligence that can filter the noise-related signals. For this purpose, a wide range of AE data were filtered using conventional filtering techniques and noise- and damage-related signals were used to train artificial intelligence. The noise-related signals were successfully separated from the damage-related signals.