An Automated Procedure for Continuous Dynamic Monitoring of Structures: Theory and Validation

Introduction: Stability diagrams are a helpful tool for operational modal analysis to obtain the physical modes of a structure. These modes can be defined by visualizing stable columns formed by consistently identified modes over a range of model orders. Extracting these modes manually becomes an obstacle if continuous identification of modal parameters is required. Materials and methods: In this paper, a procedure is configured to automatically interpret the stability diagrams constructed with the identification results of the SSI-COV/ref algorithm. This procedure is based on some methodologies found in the literature, which follow three stages. First, a stability diagram cleaning stage is defined where modal validation criteria and partitioning clustering algorithms are used to detect spurious modes. Second, a mode grouping stage based on a hierarchical clustering algorithm is implemented to form sets of modes that share similar modal information. Finally, a selection stage is applied to define representative modal parameters from the set of physical modes. Results: The proposed procedure is validated by simulating a beam-type structural model with ten degrees of freedom affected by ambient temperature functions. Natural frequencies computed for the DT140 and DT220 datasets collection with the frequency-domain decomposition method agree with the computed ones with the proposed procedure, presenting MAC coefficients higher than 0.97. A total of 192 datasets are simulated, and the acceleration responses are polluted with two noise levels, SRN = 40 [dB] and SNR = 20 [dB]. Conclusions: For the analyzed beam, the modal tracking results showed that the procedure could perform continuous identification automatically. The variations in the natural frequencies are correlated to the variations in the ambient temperature functions.