ABSTRACT: Uncertainty about the survivability and technological evolution presents its risks when designing sensors to use on infrastructure health monitoring systems. Fibre optic methods have been available for some time, and are reaching the technological maturity required to provide reliable long-term structural health monitoring solutions. By embedding fibre optic sensors in the structures at construction time, the engineers are able to detect deviations from the structure’s original design and expected dynamics. However, due to the long-term infrastructure’s life span and the relatively new fibre optic techniques, there are very few historical examples to analyse and use as proof for the adequacy of such techniques to the infrastructure’s Life-Cycle.
This study explores the opportunities associated when applying machine learning techniques to these technologies through a real case of one of the first Fibre Bragg Grating (FBG) sensorized highway bridge installed in Spain in the year 2000, the «Las Navas» bridge at the A-8 «Autovía del Cantábrico» highway in the north of Spain. The survivability of the fibre optic sensors, after eighteen years of embedded exposure, is a positive evidence of their suitability as traditional strain gauges solutions and the maintenance challenges they face. And, most importantly, the exciting new opportunities that the new machine learning analysis offer, helps evaluating the re-purpose capacity of the health monitoring systems fibre optic sensors as sustainable real time operational monitoring systems. The originally designed structural sensors are proved to be useful also to detect, count and classify operational traffic using the infrastructure indirectly and without detail physical modelling, by applying machine learning techniques that will help generalizing the use of these type of measurement sensors into the infrastructure’s world.
