Vibration-based Damage Localization and Quantification Framework of Large-Scale Truss Structures

The use of structural health monitoring (SHM) systems on a regular basis is critical to achieve early damage detection, avoid unpredicted failures, and perform cost-effective maintenance planning. The main objective of this work is to present a model-based Damage Detection Framework for truss structural systems that uses output-only vibration measurements. Model-based methods provide much more comprehensive information about the condition of the monitored system than the data-driven and also allow the prediction of the location and level of damage. The measured vibration response of a healthy structural system under operational vibrations is employed to tune a parameterized FE model using state-of-the-art FE model updating techniques to obtain an optimal numerical model of the structural system. Based on the optimal FE model, a set of damaged FE models is generated for selected damage scenarios. A damage approximation approach that represents local damage with uniform stiffness reduction is also presented. In the Damage Detection Framework, the vibration data records for both the healthy and the damaged structure and the results from multiple numerical analysis on the healthy and the damaged FE models are used. The transmittance functions for the healthy and damaged states of the structure and the FE models are derived to calculate the damage indicators. Using these indicators, potentially damaged structural members are identified, grouped, and compared to finally locate the specific damaged member. The proposed framework provides both accurate damage localization and damage quantification using a limited number of sensors for unknown input excitation. Herein, the case study used is a laboratory steel truss bridge.

Automated summary

This paper presents a model-based Damage Detection Framework for truss structural systems, which utilizes vibration measurements to detect damage. The framework provides accurate damage localization and quantification, and can be achieved using a limited number of sensors for unknown input excitation.