Premium
Enhanced stabilization diagram for automated modal parameter identification based on power spectral density transmissibility functions
Author(s) -
Afshar Mehrnoosh,
Khodaygan Saeed
Publication year - 2019
Publication title -
structural control and health monitoring
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.587
H-Index - 62
eISSN - 1545-2263
pISSN - 1545-2255
DOI - 10.1002/stc.2369
Subject(s) - modal , spurious relationship , algorithm , operational modal analysis , parametric statistics , cluster analysis , stability (learning theory) , control theory (sociology) , basis (linear algebra) , transmissibility (structural dynamics) , sensitivity (control systems) , modal analysis , engineering , mathematics , computer science , artificial intelligence , vibration , electronic engineering , finite element method , physics , statistics , acoustics , chemistry , structural engineering , geometry , machine learning , polymer chemistry , control (management) , vibration isolation
Summary Operational modal analysis based on power spectral density transmissibility (PSDT) functions is a useful tool to identify the modal parameters with low sensitivity to excitations. For pole extraction from the PSDT function, a proper parametric identification method such as the polyreference least squares complex frequency‐domain method or poly‐Max method can be used. Then, the poles are selected from a stabilization diagram (SD) with overestimating the system model order. Therefore, spurious modes can be identified that must be distinguished and removed from the system poles. To reach this aim, many techniques have been proposed and applied. In this paper, a new algorithm is proposed to enhance the performance of the SD for automated modal parameter identification based on the PSDT. The algorithm is composed of two main phases. In the first phase, the spurious modes are discriminated from the system poles on the basis of the conventional and supplementary stability criteria. On spurious mode omission, two new criteria named “pole criterion” and “coherence criterion” are introduced and applied as the supplementary stability criteria to make a more clear SD. Then, the extracted poles are categorized in the distinct clusters through a new strategy for comparing modes. In the second phase, a novel multiscreening algorithm is implemented for the automated identification of the system poles. Accordingly, the searching and averaging processes are followed between clusters, and the poles are screened to automatically identify the system poles on the basis of the numbers of their repetition in the SD via k ‐means clustering algorithms. Also, to improve the accuracy of the identification, the Hilbert transform is used in the construction of the PSDT functions. Finally, to validate and demonstrate the efficiency of the proposed method, a computer simulation and an experiential case study are considered.