Structural Health Monitoring Driven by Big Data and Machine Learning

Structural Health Monitoring (SHM) is a process of deploying an array of sensors or a sensor network permanently on a structure and successively assessing the structural condition and health status by implementing various diagnostic and prognostic methods in line with the monitoring data. It helps issue early warnings on structural damage or deterioration prior to costly repair or even catastrophic failure. With the rapid development in advanced sensing, signal processing, data transmission and management techniques, impressive SHM practices on critical infrastructure systems appear across different countries over the past two decades. Machine Learning (ML) has currently emerged as one of the principal theoretical and practical approaches for designing ‘machines’ that learn from data acquired by an SHM system and inferring the structural condition. Because a long-term SHM system successively accumulates monitoring data over time, the evolutionary tracking of structural condition using vast amounts of SHM data would be accomplished in the context of Big Data (BD). This lecture will introduce innovative sensors and various ML methods (e.g., deep neural networks, sparse Bayesian learning, transfer learning) in compliance with BD for SHM applications in civil and railway engineering. SHM paradigms for long-span bridges, high-rise buildings and high-speed rail are illustrated. 

The speaker will also give a brief introduction to MSc and PhD Programmes in CEE department, as well as some academic colleagues in his Unit and the focused research areas.