A New Methodology based on First Passage Times for Structural Health Monitoring in Civil Engineering

Structural Health Monitoring refers to the process of implementing a damage identification strategy for infrastructures. As the global population continues to grow, urbanisation expands and the age of existing infrastructure is also increasing. Therefore, the field of Structural Health Monitoring has gained in popularity. A wide variety of methods have already been developed with vibration-based methods being among the most prevalent.

The concept of early damage detection is crucial. In the absence of apparent damage to a structure, the common and basic technique of visual inspection, which is employed for damage detection, is ineffective. The objective of detecting damage as early as possible is to reduce the costs and time required for repairs, as well as to enhance the safety and reliability of existing structures.

This thesis presents a novel method, located within the vibration-based methods. Unlike the modal-based methods, such as those tracking the degradation of eigenfrequencies, the proposed approach hinges on the concept of First Passage Time. This concept is applied for the first time to Structural Health Monitoring and refers to the time required by a dynamical system to reach a particular state for the first time, while starting from a known initial condition.

Although the mathematical study of First Passage Times has been well-established, there has been a lack of efficient algorithms for computing First Passage Times from experimental data. Therefore, a new and optimised algorithm is developed. As the First Passage Time is the keystone of this thesis, this algorithm benefits from fast computation time to extract the First Passage Times from any given time signal. For any random signal, First Passage Time is a random variable. The distribution of First Passage Times is shown to be a good candidate for the early damage detection. A novel methodology is proposed. It is based on the pre-processing of the input data and on the comparison of the distributions of First Passage Times. The latter relies on a new proposed statistical test, which is based on the sampling distribution of First Passage Times, and various existing two-sample tests, such as the Kolmogorov-Smirnov and the Anderson-Darling tests.

Finally, the sensitivity of the methodology is initially evaluated through numerical examples before being applied to two experimental setups: a small-scale laboratory test under control conditions and a large-scale outdoor test submitted to environmental effects. It is demonstrated that the First Passage Time is sufficiently sensitive to detect minor structural changes, thereby enabling damage detection at an early stage.