Impedance-based structural health monitoring incorporating neural network technique for identification of damage type and severity

Abstract

Impedance-based structural health monitoring (SHM) has come to the forefront in the SHM community because of its practical potential for real applications. In the impedance-based SHM technique, it is very important to select the optimal frequency range most sensitive to the expected structural damage, and more quantitative information on the structural damages might be needed compared to the conventional damage index. Therefore, this study proposes an innovative neural network (NN)-based pattern analysis tool (1) to identify damage-sensitive frequency ranges autonomously and (2) to provide detailed information such as the damage type and severity. The importance of selecting the optimal frequency range was first investigated experimentally using a simply-supported aluminum beam. The performance of the proposed NN-based approach was validated throughout damage identifications of loose bolts and notches on a bolt-jointed aluminum beam and a lab-scale pipe structure. Finally, the proposed NN-based algorithm was embedded into a wireless impedance sensor node to detect real damage in a full-scale bridge. Overall, the proposed approach incorporating a wireless impedance sensor node was used to evaluate the damage type and severity in multi-type and multiple structural damage cases.