Multi-Input Multi-Output Time Domain Modal Parameter Identification Method Using ERA

In this paper, we introduce a multi-input multi-output time domain modal parameter identification method based on the Eigensystem Realization Algorithm (ERA), which significantly improves identification accuracy and system robustness. The method processes multiple input signals simultaneously and extracts multiple output parameters through state-space realization, providing comprehensive characterization of system dynamic response characteristics. Implementation typically involves constructing a Hankel matrix from time-domain response data, followed by singular value decomposition (SVD) for model order determination and system realization. Additionally, the method incorporates signal preprocessing techniques (e.g., filtering and correlation analysis) to eliminate noise and interference, thereby enhancing identification performance. Key algorithmic steps include: 1) Data organization into block Hankel matrices, 2) SVD-based system order reduction, 3) State-space model realization using the ERA formulation, and 4) Modal parameter extraction via eigenvalue decomposition of the realized system matrix. This approach demonstrates broad application potential in practical engineering scenarios, enabling engineers to better understand and analyze system dynamic characteristics while improving system reliability and performance through accurate modal parameter identification.