Damage precise localization under varying operating conditions via the vibration-data-based Functional Model method: Formulation and experimental validation

Abstract

A robust to varying Operating Conditions (OCs) vibration-data-based damage precise localization method is postulated. The method is, for the first time, capable of providing precise damage coordinate estimates despite the presence of varying OCs. Moreover, it is based on only partial models of the dynamics and a few vibration sensors, without resorting on large-scale Finite Element Models or requiring knowledge of the OCs in its real time operational phase. It is founded upon a proper formulation of the broader Functional Model Based Method and uses Functionally Pooled AutoRegressive with eXogenous excitation (FP-ARX) models in which the scheduling vector incorporates both the damage coordinates and the OCs. Both excitation-response and response-only versions are provided. The method is experimentally validated via damage precise localization on a laboratory plate structure under varying boundary conditions using only two vibration sensors. A comparative assessment of the two versions is also provided based on hundreds of experiments. The results reveal high achievable accuracy, which is slightly superior for the excitation-response version, despite considerable variations in the boundary conditions.