Robust optimal sensor placement for response reconstruction using output-only vibration measurements

Abstract

Vibration measurements taken from various locations in a structure are used to improve models and model-based predictions of output quantities of interest (QoI). Such improved models and predictions can be used to update the state of the structure, estimate model and parameter uncertainties, trace structural deterioration over time, update structural reliability and safety, prognose the remaining lifetime of structural components (e.g. fatigue lifetime) and eventually design cost-effective maintenance actions. Since measurements are affected by errors and limited in number, placement of the sensors is a crucial problem. In this study, the problem of response reconstruction in structures using output-only vibration measurements is investigated. A filtering technique for joint input-state estimation is used to obtain estimates of response quantities of interests and the covariance of the errors in these predictions. A robust optimal sensor placement framework is then presented to be able to obtain most reliable responses considering uncertainty in the model parameters. Copyright © 2019 European Safety and Reliability Association.