Increasing Probability of Detecting Acoustic Emission Sources Using Artificial Neural Networks

We study the possibilities of reducing the location error of acoustic emission (AE) sources for cases where AE sources are located near AE receiving transducers (AET). Artificial neural networks (ANN) were used to reduce the location error and increase the probability of detecting such sources when processing AE diagnostic data. The training of the ANN was carried out according to the parameters of location pulses recorded using a pulse generator on the surface of a steel strip. A broadband transducer installed at various distances from the receiving AET was used to generate pulses. After training, the ANN was used to process the results of locating AE sources recorded during rupture tests of a steel strip with stress raisers in the form of holes with a diameter of 5 mm located at a distance of 40 mm from the receiving AET. The probability of detecting AE sources using the ANN according to the parameters of location pulses in the areas of the stress raisers was , i.e., almost 12 times higher than the probability obtained in the case of using the standard algorithm.