TY - JOUR
T1 - Efficient modal parameter identification using DMD-DBSCAN and rank stabilization diagrams
AU - Wu, Chengyuan
AU - Yang, Zhichun
AU - He, Shun
N1 - Publisher Copyright:
© 2025 Elsevier Masson SAS
PY - 2025/6
Y1 - 2025/6
N2 - Modal parameter identification of aerospace structures, particularly improving the efficient using time-domain vibration responses, is currently a topic of great interest in structural dynamics. Data-driven Dynamic Mode Decomposition (DMD) technique in fluid dynamics has shown good potential for identifying the flow mode. In this respect, there have been a few studies on the application of DMD technique in structural dynamics. This paper proposes an efficient structural modal parameter identification method that combines the DMD technique and Density-based Spatial Clustering of Applications with Noise (DBSCAN) algorithm, called the DMD-DBSCAN method. Combined with the proposed rank stabilization diagram, the spurious modes produced by the DMD technique can be effectively cleaned by using DBSCAN algorithm to obtain the physical modal parameter. A numerical case of a composite wing model is presented to validate the proposed method, particularly when identifying the closely spaced modes. It has been shown that the rank stabilization diagram is more appropriate for DMD technique when compared with the solely existing sampling frequency stabilization diagram. An experiment on an aircraft model with response from three-dimensional optical technique is used to demonstrate the method's ability to deal with real complex structures. Compared with traditional methods, results show that the proposed DMD-DBSCAN method can accurately identify modal parameters of complex structures and has excellent operational efficiency because it is free from high order Hankel matrix construction. The proposed data-driven modal parameter identification method is efficient and promising for analysis of complex structures with large datasets.
AB - Modal parameter identification of aerospace structures, particularly improving the efficient using time-domain vibration responses, is currently a topic of great interest in structural dynamics. Data-driven Dynamic Mode Decomposition (DMD) technique in fluid dynamics has shown good potential for identifying the flow mode. In this respect, there have been a few studies on the application of DMD technique in structural dynamics. This paper proposes an efficient structural modal parameter identification method that combines the DMD technique and Density-based Spatial Clustering of Applications with Noise (DBSCAN) algorithm, called the DMD-DBSCAN method. Combined with the proposed rank stabilization diagram, the spurious modes produced by the DMD technique can be effectively cleaned by using DBSCAN algorithm to obtain the physical modal parameter. A numerical case of a composite wing model is presented to validate the proposed method, particularly when identifying the closely spaced modes. It has been shown that the rank stabilization diagram is more appropriate for DMD technique when compared with the solely existing sampling frequency stabilization diagram. An experiment on an aircraft model with response from three-dimensional optical technique is used to demonstrate the method's ability to deal with real complex structures. Compared with traditional methods, results show that the proposed DMD-DBSCAN method can accurately identify modal parameters of complex structures and has excellent operational efficiency because it is free from high order Hankel matrix construction. The proposed data-driven modal parameter identification method is efficient and promising for analysis of complex structures with large datasets.
KW - Data-driven
KW - Density-based spatial clustering of applications with noise
KW - Dynamic mode decomposition
KW - modal parameter identification
KW - stabilization diagram
UR - http://www.scopus.com/inward/record.url?scp=86000315621&partnerID=8YFLogxK
U2 - 10.1016/j.ast.2025.110112
DO - 10.1016/j.ast.2025.110112
M3 - 文章
AN - SCOPUS:86000315621
SN - 1270-9638
VL - 161
JO - Aerospace Science and Technology
JF - Aerospace Science and Technology
M1 - 110112
ER -