An Efficient Approach for Detecting Bolt Loosening in Periodically Supported Structures

Abstract

Periodic structures consist of a large number of repeated units, which are joined together in an identical manner to form the whole system. It has been revealed in the literature that periodic structures have a variety of interesting dynamic properties including the well-known vibration band-gap phenomenon. To our best knowledge, most of research works related to periodic structures to date focus on the structural vibration control and optimization design by employing such unique dynamical characteristics of periodic structures, however, there are very few studies been carried out for the damage diagnosis with respect to this special type of structural system. By utilizing the measured modal parameters, this paper reports a probabilistic methodology for detecting bolt loosening on a periodically supported beam-type structural system, which intends to represent typical pipeline structures endowed with bolted flange joints in industry. Firstly, the entire periodic structural system is modeled by using the spectral element (SE) method combined with the transfer matrix approach. Then, a model-based damage detection approach is presented to identify the statistic characteristics of uncertain parameters with the most plausible class of models selected with suitable complexity of model parameterization. The validity and efficiency of the proposed methodology is verified through numerical case studies conducted for characterizing the connection status in a pipeline model.KeywordsPeriodic structureBolt loosening detectionParticle filterBayesian approach