A hybrid approach for damage detection of structures under operational conditions

This paper presents a hybrid stochastic/deterministic optimisation algorithm to solve the target optimisation problem of vibration-based damage detection. The use of a numerical solution of the representation formula to locate the region of the global solution, i.e., to provide a starting point for the local optimiser, which is chosen to be the Nelder-Mead algorithm (NMA), is proposed. A series of numerical examples with different damage scenarios and noise levels was performed under impact and ambient vibrations. Thereafter, an experimental study of three cantilever beams with several different damage scenarios was conducted. To test the accuracy and efficiency of the optimisation algorithm, its results were compared to previous procedures available in the literature, which employed different solutions such as the genetic algorithm (GA), the harmony search algorithm (HS) and the particle swarm optimisation (PSO) algorithm. The performance of the proposed optimisation scheme was more accurate and required a lower computational cost than the GA, HS and PSO algorithms, emphasising the capacity of the proposed methodology for its use in damage diagnosis and assessment In addition, the methodology was able to handle incomplete measurements and truncated mode shapes, which is of paramount importance for dealing with operational conditions in long-term structural health monitoring (SHM) applications. (C) 2013 Elsevier Ltd. All rights reserved.