Sensing of operational civil infrastructure systems is a key component in structural health monitoring (SHM) that allows civil engineers and decision-makers unprecedented levels of observation of the infrastructure systems. Identifying the unknown parameters of a structural model (in other words, calibrating a structural model) is achieved by observing the system’s behavior or response using a suitable collection of sensors. This aids professionals to appropriately model their systems, improve system designs, and optimize future operations and maintenance processes.
Traditional fixed-sensor setups – sensors which remain attached to specific locations of the structure – have enjoyed decades of successful field deployments and have proven valuable for measuring static and dynamic structural responses. However, with the rapid advancement in the field of wireless communication and robotics, there has been a surging interest in the use of mobile-sensor networks for SHM.
In mobile-sensor networks, a few sensors can be used to collect data containing dense spatial information along mobile-sensor paths compared to limited information obtained by fixed-sensor installations at certain locations as shown in Figure 1. With denser spatial observations, high resolution mode shapes can be extracted improving the accuracy of system identification and damage detection to a great extent. Additionally, mobile-sensor networks are easier to implement and can also prove to be cost-effective.
Currently, most system identification algorithms are not compatible with mobile-sensor data. The incompatibility is primarily attributed to the large proportion of structured missingness in the collected mobile-sensor data. Missing (or unknown) entries in data matrix inherently arises when a mobile-sensor traverses across space with time (as shown in Figure 2). Due to this very reason, many structural identification algorithms which deal with observations from fixed-sensor setups fail to deal with mobile-sensor observations.
In this context, our work is focused towards developing algorithms for structural system identification which can integrate mobile sensor data and provide reliable results.
