For structural health monitoring (SHM) of civil structures, one needs to install sufficient sensors for measuring structural responses and influential environmental/operational (E/O) factors. Due to various reasons such as total budgets, weather conditions, structure locations, and monitoring target and duration, it may not be feasible to install all potential sensors. In order to devise and implement an affordable SHM program on large-scale civil structures, this paper proposes a new methodology for verifying the sufficiency of contact-based E/O sensors installed in long-span bridges by benefiting machine learning and spaceborne remote sensing. The main premise of the proposed methodology lies in the fact that structural responses obtained from some products of remote sensing allow civil engineers to investigate the sufficiency of contact sensors and also analyze the impacts of measured and unmeasured E/O factors. Using structural displacement responses obtained from remote sensing and limited measured E/O data from contact-based sensors, a regression model developed from a supervised artificial neural network is designed to evaluate the sufficiency of contact E/O sensors using the R-squared metric under three scenarios. Real-world long-span bridges are considered to testify the proposed methodology using displacement responses and air temperature data. Results demonstrate that the methodology presents an effective and practical strategy for affordable SHM programs.
Loading....