Vibration monitoring via spectro-temporal compressive sensing for wireless sensor networks

The reliable extraction of structural characteristics, such as modal information, from operating structural systems allows for the formation of indicators tied to structural performance and condition. Within this context, reliable monitoring systems and associated processing algorithms need be developed for a robust, yet cost-effective, extraction. Wireless Sensor Networks (WSNs) have in recent years surfaced as a promising technology to this end. Currently operating WSNs are however bounded by a number of restrictions relating to energy self-sustainability and energy data transmission costs, especially when applied within the context for vibration monitoring. The work presented herein proposes a remedy to heavy transmission costs by optimally combining the spectro-temporal information, which is already present in the signal, with a recently surfaced compressive sensing paradigm resulting in a robust signal reconstruction technique, which allows for reliable identification of modal shapes. To this end, this work outlines a step-by-step process for response time-series recovery from partially transmitted spectro-temporal information. The framework is validated on synthetic data generated for a benchmark structure of the American Society of Civil Engineers. On the basis of this example, this work further provides a cost analysis in comparison to fully transmitting wireless and tethered sensing solutions.