Article révisé par les pairs
Résumé : This paper deals with the experimental application of modal filters for automated damage localization using dynamic strain measurements. Previously developed for damage detection, the extension of modal filtering to damage localization consists in splitting a very large network of dynamic strain sensors into several independent local sensor networks. An efficient signal processing coupled to control charts allows a fully automated data-based damage localization once the modal filters are initialized. The method is tested experimentally on a small clamped-free steel plate and a 3.78 m long steel I-beam, both instrumented with a network of cheap piezoelectric patches to measure the dynamic strains. A removable damage is introduced at different positions by means of a small removable damage device. For both applications, the method can successfully detect and locate all damage cases considered, showing the potentiality of the method for field applications.