The application of a neural network to classify the acoustic emission waveforms emitted by the concrete under thermal stress

In this article Acoustic Emission (AE) measurement results for five different compositions differing in compression strength are presented. Thermal stresses occuring in concrete samples during their cooling after heating up to 150$^{\circ}$C in controlled conditions have been the source of AE signals. The influence of structure of frequency spectra of recorded AE signals is described. An automatic recognition procedure of the recorded AE waveforms using neural network is discussed and the details of the learning process of the neural network are shown.