ABSTRACT

Monitoring geotechnical structures such as landslides, levees, earth dams, foundation soils and the like is essential to develop early emergency alerts and prevent disasters. In this work, a soil movement monitoring system was developed by using inclinometers based on fiber Bragg gratings (FBGs) and pattern recognition techniques (which are rooted in the artificial intelligent field) to provide automatic assessment based on strain measurements within the context of geotechnical health monitoring. A prototype simulating a typical soil profile composed of different soil layers (different materials having dissimilar stiffness) and a network of nine inclinometers (each one instrumented with two FBGs) located at different points was developed in order to test and implement different pattern recognition techniques which allow automatically detecting landslides prior to a catastrophic event. Results showed that small changes in the inclinometers readouts can be detected and correlated with slight variations in the local strain field of the terrain.