Structural management and Value of Information analysis accounting for sensor data quality

Abstract

Structural health monitoring (SHM) can be used to assess the state of health of civil structures and infrastructures and acquire information that can support maintenance-related activities and post-disaster emergency management. Nevertheless, SHM outcomes may be susceptible to errors due to malfunctioning of the sensing system. The long-term benefit of SHM systems against the initial investment in sensing instrumentation is often quantified without considering the eventuality of faulty sensors. Inaccurate or missing sensor data, not accounted for when information from the SHM system is used to support decisions, can lead to the choice of sub-optimal maintenance actions, and associated economic losses. In the last two decades, Sensor Validation Tools (SVTs) have been proposed, which assess data quality before the SHM information is extracted to isolate and discard abnormal measurements. Nevertheless, automatic SVTs are still rarely implemented in real applications. Recently, a framework based on Bayesian decision theory has been proposed to quantify the benefit of using an SVT before it is implemented. The novel approach extends the traditional VoI to consider multiple ﾓfunctioningﾔ states of the SHM system with the final goal of quantifying the additional benefit obtained from SVTs. In this paper, this framework is demonstrated using a general example representative of different real situations. Uncertainties in the SVT results are accounted for to show that the adoption of an SVT enhances the overall benefit provided by an SHM system.