The reinforced concrete structures require constant monitoring to ensure their durability and safety of the structures. Thus, it is of great importance to monitor and identify the damage in structures. From non-destructive testing methods, acoustic Emission (AE) is a well-known method used to locate cracks, and damages and provide information about the performance of the material or a structure.
This study aimed to determine the damage progress within a reinforced concrete beam by incorporating all AE datasets in a hybrid deep neural network. For this, a hybrid AE dataset consisting of three damage states of AE activities obtained from a four-point bending test of a reinforced concrete beam was developed. For the development of the hybrid model, a 1-dimensional convolutional network and a 2-dimensional convolutional network were combined and all AE data containing AE parameters, 1D waveforms, and scalograms extracted from continuous wavelet transform (CWT) were used to train the model and validated against a testing dataset. From the results, it was found that the model was trained well enough with a large amount of hybrid AE dataset, and the performance accuracy in training and testing was found 97.6% and 95% respectively, leading to the conclusion that the hybrid model has the potential to predict damages in reinforced concrete with great confidence.

Artificial intelligence; deep learning; acoustic emission; damage estimation; automated monitoring