Research On Piezoelectric Impedance Technology And BP Neural Networks For Structural Health Monitoring

Recently, the study of piezoelectric impedance technology which could be widely applied in structure health monitoring field is very active. In this paper, the application status of the piezoelectric impedance technology and the Artifical Nenural Networks (ANN) is introduced, and the basic theory of them is ananlyzed. In the paper, a BP Nenural Networks is made on the base of the ANN and the piezoelectric impedance technology to ascertain the location of the damnification. The main contents are as following:1. The electromechanical coupling property is analyzed, and the relation between the mechanical impedance and electric impedance is deduced as well as the coupling expressions is gotten. In addition, the basic principle of piezoelectric impedance technology for structure health monitoring is analyzed.2. The basic principle of the ANN and BP arithmetic is studied. The BP Networks based on the piezoelectric impedance experiment is designed using the MATLAB 7 nenural networks toolbox.3. Experiments are made on the base of the piezoelectric impedance technology including the crack experiment and the bolts'looseness experiment. The health index is used for the result analysis. The results show that PZT is sensitive to the damnification which is near to it and not sensitive enough to the damnification which is far from it.4. A BP Nenural Networks is made using the experiment of the bolts' looseness on the base of the piezoelectric impedance. The results show that the networks we made can distinguish the location of the single looseness of the bolts, while for the double looseness of the bolts, it works with error but can obviously reduce the extension of the loosen bolts.