This paper describes an experimental study to extract the dynamic characteristics of a two-storey reinforced concrete (RC) frame structure using piezo-electric ceramic (PZT) patches. PZT patches were embedded in the structure at the time of construction. Basically two techniques were applied to monitor the health of structure, the global dynamic technique and the local electro-mechanical impedance (EMI) technique. Global dynamic technique, which is based on frequency changes, is effective in low frequency range only. Due this limitation, initial damage/hair crack can not be detected by the global dynamic technique. On the other hand, EMI technique acts at higher frequency range and is very sensitive to detect the initial damage/hair cracks. The lower natural frequencies of the frame structure were determined experimentally using global techniques. The two-storey R.C. frame was modeled using ANSYS 9.0 to determine the frequencies numerically. Experimental results were compared with numerical results, which were found to be agreeable. Initial cracks were detected by the EMI technique. Severity and location of damage can be also determined with the help of these parameters. Inputs were chosen from these parameters to train an artificial neural network (ANN) whose outputs were the severity and the location of damage. Thus, complete monitoring can be done by the combination of global vibration and EMI technique using PZT patches. This integrated approach can be used for damage/crack detection at very early stage. This approach is very sensitive and cost effective to predict the incipient damages in civil structures.
Deep convolution neural network for crack detection on asphalt pavement
