Enhanced data-driven Damage Detection for Structural Health Monitoring Systems

Abstract

In structural engineering, it is essential to monitor the operation condition of an aging structure. Thus, damage detection is widely used for structure monitoring. The aim of this work is to propose an adaptive kernel PLS based GLRT chart to improve the detection of damage in civil structural systems. The proposed technique aims to integrate the advantages of the adaptive nonlinear input-output model (kernel PLS) with those of GLRT chart. This technique will be tested using a simulated benchmark structure through the surveillance model variables. The technique based on adaptive representation is found to be more effective over the conventional technique.