Abstract:Output-only modal identification plays an important role in the structural health monitoring of large-scale structures. In recent years, blind source separation (BSS) has achieved great success in structural modal identification. Sparse component analysis (SCA), which is one of the most popular methods of BSS, has the capability to handle nonstationary excitation and underdetermined problems. In the process of SCA, clustering number, which is equal to the number of active modes, plays an important role in the estimation of modal matrix, in which the hierarchical clustering algorithm is used. However, the clustering number is always unknown in the clustering step, which makes application inconvenient. To fill this gap, an improved SCA method, equipped with a process of estimating the clustering number, is proposed in this paper. After transforming the signals into time-frequency (TF) domain, the single-source-points (SSPs) detection process is applied to pick out the TF points at which only one mode makes a contribution to the responses. The clustering technique is preceded by a preprocessing step to determine the clustering number. The key idea is that the clustering number is equal to the number of columns in the modal matrix, which is reflected in the number of lines in the scatter plot of two observations. A normalization method is proposed to distinguish the clusters clearly. The number of clusters is acquired through statistical analysis of the normalized vectors. After obtaining the modal matrix, the smoothed zero-norm algorithm is used to recover the modal responses in order to extract natural frequencies and damping ratios. An experimental cantilever beam and a three degree-of-freedom (DOF) numerical system with closely spaced modes were used to verify the effectiveness of the proposed method. The results showed that the improved SCA could detect the number of active modes for the beam and the numerical system. Full-scale data measured from the Green Building located at the Massachusetts Institute of Technology (MIT) campus and the Tianjin Yonghe Bridge were analyzed to verify the effectiveness of the proposed method in practical applications.

Automatic stochastic subspace identification of modal parameters based on hierarchical clustering method

Summary of methods eliminating spurious modes in automatic modal parametric identification

Automatic Modal Parameters Identification and Uncertainty Quantification Based on Block-Bootstrap and Multi-Stage Clustering under Ambient Excitation

Autonomous Modal Parameter Extraction Based on Stochastic Subspace Identification

Modal energy-based distinguishing spurious modes obtained from stochastic subspace identification

Automatic Analysis of Stabilization Diagram Using Iterative Genetic-Fuzzy Clustering Method

An improved stochastic subspace identification for operational modal analysis

Improvement of Electromechanical Mode Identification from Ambient Data with Stochastic Subspace Method

Automatic identification of structural modal parameters based on density peaks clustering algorithm

Clustering Number Determination for Sparse Component Analysis during Output-Only Modal Identification

Sparse Component Analysis with Optimized Clustering for Underdetermined Blind Modal Identification

Automated Modal Parameter Estimation Using Correlation Analysis and Bootstrap Sampling

Study on the method for stochastic subspace identifying structural modal parameters based on two-stage stabilization diagram

Fully automated precise operational modal identification

Time-Varying Modal Parameters Identification by Subspace Tracking Algorithm and Its Validation Method

Subspace System Identification Methods for Structural Modal Parameters

ANALYSIS HOW STOCHASTIC SUBSPACE IDENTIFICATION BRINGS FALSE MODES AND MODE ABSENCE

Performance Comparison for Data-Driven and Covariance-Driven Stochastic Subspace Identification Method

A Correlation-Based Bi-Partition Hierarchical Clustering Method for Mode Identification of Multimode Processes

Automatic modal identification via eigensystem realization algorithm with improved stabilization diagram technique

Machine Learning-Based Stochastic Subspace Identification Method for Structural Modal Parameters