Damage Detection and Localization of a Loose Joint through Pattern Correlation of Measured and Computed Dynamic Properties

This article presents a novel procedure for detecting and locating damage caused by a loose bolted joint in a modular plate structure. The primary aim of this work is to locate the loose joint with minimum measured points and without precise tune of computational model. The proposed method is based on correlating patterns of modal property changes using simulated and measured data. These patterns combine the relative shifts in natural frequencies and the norms of relative changes in mode shapes, derived from a pre-computed database of finite element method (FEM) simulations for various potential damage scenarios. The experimental validation demonstrates that the procedure can effectively and accurately locate the position of a loose joint using only five accelerometers, despite using an untuned FEM model for prediction.