Continuous Modal Identification and Tracking of a Long-Span Suspension Bridge Using a Robust Mixed-Clustering Method
Min He, Peng Liang, Eugene J. OBrien, Xin Sun, Yang Zhang
Abstract
For long-term continuous structural health monitoring to be effective, a process of continuous modal identification that should preferably be automated is required. This case study paper describes a robust, fully automated approach for continuous modal identification and tracking. The approach is demonstrated on a long-span suspension bridge under operational conditions. Contributions are made in three stages: eliminating the spurious modes, extracting the physical modes, and estimating the precision. The proposed approach helps avoid any manual intervention, requires no manually tuned thresholds or prior assumption, and is robust. One week of field monitoring data are analyzed to validate the process. Modal tracking is conducted to show the stability of continuous analysis and to track the evolution of modal parameters. Parametric analysis is conducted to demonstrate the robustness. The case study shows that the proposed approach yields better results than alternative approaches and successfully identifies and tracks multiple closely spaced modes without any manual intervention.