Semi-supervised bridge indirect structural health monitoring using Isolation Distributional Kernels
- Publisher:
- ACADEMIC PRESS LTD- ELSEVIER SCIENCE LTD
- Publication Type:
- Journal Article
- Citation:
- Mechanical Systems and Signal Processing, 2025, 226
- Issue Date:
- 2025-03-01
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
This paper addresses the problem of structural health monitoring (SHM) involving the observation of bridges using periodically sampled acceleration responses from an instrumented vehicle as a method to detect changes in its structural integrity. It applies Isolation Distributional Kernels (IDK) with a finite and computational feature map to obtain accurate classification and severity quantization using probability distributions. This is achieved through the kernel expansion of the data's dimensionality, providing an implicit feature map framework for detecting abnormalities in higher-order space. The signals first undergo preprocessing with frequency filtering, averaging, and scaling. The unknown and benchmark signals are then compared to calculate an individual similarity score. The performance of this method is assessed through numerical simulation using a half-car model and a simply-supported beam, where various bridge states, including healthy and different percentages of damage severity, are considered. These findings are then experimentally validated from data generated from a scaled bridge within a laboratory. The proposed framework has three major advantages compared to traditional modal and non-modal frameworks. First, a new application of a novel IDK approach can successfully use drive-by monitoring for progressive damage assessment. Second, the approach works with a limited dataset due to the minimal training required. Third, the model is more efficient, obtaining linear runtime complexity as the sample size is increased with reduced pre-processing requirements compared to comparative methods. These benefits of the IDK approach to indirect SHM successfully overcome the challenges of large-scale data acquisition in drive-by monitoring, providing an accurate and computationally effective solution.
Please use this identifier to cite or link to this item: