Advanced SHM using computer vision and machine learning

The research presented in this paper aligns to the digital transformation of Civil Engineering and specifically Structural Health Monitoring (SHM) Systems. SHM can provide valuable information on the structural capacity and changes in structural performance, generally as an indication of damage. The applications of many SHM systems are currently limited by structure type, access for fixing of sensors, light levels and maintaining power supplies. This paper investigates the use of computer vision systems for SHM to ensure the safety and resilience of our civil infrastructure. Computer Vision is a new method of SHM which operates by recording motion pictures of a target area, or feature, on bridges and civil infrastructure. The development and validation of a contactless deflection monitoring system which tracks features to sub pixel accuracy is presented. The image is also pre-filtered for changing light levels in the environment and due to crossing freight. Machine learning is also used to identify events which provides useful data on real loading. The results of this research confirm the suitability of these systems for information to accurately determine the health of bridges.