Automated structural health monitoring based on adaptive kernel spectral clustering

Abstract

Structural health monitoring refers to the process of measuring damagesensitive variables to assess the functionality of a structure. In principle, vibration data can capture the dynamics of the structure and reveal possible failures, but environmental and operational variability can mask this information. Thus, an effective outlier detection algorithm can be applied only after having performed data normalization (i.e. filtering) to eliminate external influences. Instead, in this article we propose a technique which unifies the data normalization and damage detection steps. The proposed algorithm, called adaptive kernel spectral clustering (AKSC), is initialized and calibrated in a phase when the structure is undamaged. The calibration process is crucial to ensure detection of early damage and minimize the number of false alarms. After the calibration, the method can automatically identify new regimes which may be associated with possible faults. These regimes are discovered by means of two complementary damage (i.e. outlier) indicators. The proposed strategy is validated with a simulated example and with real-life natural frequency data from the Z24 pre-stressed concrete bridge, which was progressively damaged at the end of a one-year monitoring period.

A Sensor Data Fusion System Based on k-Nearest Neighbor Pattern Classification for Structural Health Monitoring Applications

“…When it is possible to ensure the proper performance of the sensors, damage identification tasks can be applied. On this topic, it is possible to find some strategies for damage detection, localization and classification, including robust detection [ 4 ], which considers the variations in the environmental conditions; or even the use of a robust regression technique to analyze data from an SHM system in order to distinguish between damage and environmental conditions [ 5 ], the development of a methodology to remove the environmental effects from the SHM data by using principal component analysis and Hilbert–Huang transformation [ 6 ] or the use of adaptive kernel spectral clustering that detects damage in its initial stage [ 7 ]. With respect to the use of machine learning approaches, several strategies have been explored.…”

Railway bridge structural health monitoring and fault detection: State-of-the-art methods and future challenges

“…The methods showed good performance in identifying failures when an FE model was adopted; however, when a real steel railway bridge was analysed, several false alarms were raised. Langone et al 137 presented an unsupervised adaptive kernel spectral clustering in order to monitor the health state of bridge infrastructure. The method was applied to a real-time monitoring process of a concrete bridge that was damaged artificially.…”

Enhanced Affinity for Spectral Clustering using Topological Node Features (TNFS)

“…Many real-world pattern recognition applications such as social community detection [21], document clustering [17], health analytics [13], etc., require clustering technique as a critical step for their realization. Since the data space corresponding to a real-world application, in general, contains complex structures, traditional algorithms• such as k-means or single linkage fail to produce useful results.…”