Vibration based damage detection for a population of nominally identical structures via Random Coefficient Gaussian Mixture AR model based methodology

Vibration based damage detection for a population of nominally identical structures is characterized by considerable uncertainty which is caused by even slight dissimilarities among the population members, and is compounded with that of additional sources. In this work a response only and unsupervised Random Coefficient Gaussian Mixture AR model based methodology is postulated for tackling the problem. Its effectiveness is experimentally assessed via damage detection for a population of composite beams. The results indicate significant performance improvement over a corresponding Random Coefficient Gaussian method, yet similar to that of a Multiple Model based method. (C) 2017 The Authors. Published by Elsevier Ltd.