The influence of frequency content on the performance of artificial neural network–based damage detection systems tested on numerical and experimental bridge data

The method herein proposed provides a novel perspective about data processing within structural health monitoring, which is essential for automated real-time monitoring and assessment of civil engineering structures. The low- and high-frequency contents of the forced vibration response of a structure are used to train and test artificial neural networks for the purpose of damage detection. In the context of several damage scenarios, the different versions of the networks are compared with each other with the aim of verifying which are the most efficient regarding novelty detection (one-class classification). The data related with the high-frequency response showed to contain more useful information for the proposed damage detection algorithm, when compared with the low-frequency response data (typically modal). In view of that, high frequencies should be given more attention in future research about their application in connection with structural health monitoring systems.

Dieses Werk wurde unter der Creative-Commons-Lizenz Namensnennung 4.0 International (CC-BY 4.0) veröffentlicht und darf unter den Lizenzbedinungen vervielfältigt, verbreitet, öffentlich zugänglich gemacht, sowie abgewandelt und bearbeitet werden. Dabei muss der Urheber bzw. Rechteinhaber genannt und die Lizenzbedingungen eingehalten werden.