To address the susceptibility of modal parameter identification to interference and instability under complex environmental excitation, this paper proposes a modal parameter identification method referred to as RVMD-SSI that combines Reduced-order Variational Mode Decomposition (RVMD) with covariance-driven Stochastic Subspace Identification (SSI). The method adaptively determines RVMD parameters using particle swarm optimization, constructs a comprehensive evaluation index incorporating correlation coefficients, energy proportion, and energy entropy to extract effective intrinsic mode functions, and subsequently determines SSI model order based on RVMD results to achieve robust modal identification. Numerical simulation results demonstrate that under various noise types and signal-to-noise ratio conditions, RVMD-SSI achieves frequency errors below 0.12% and damping errors below 2.5%, with accuracy and efficiency surpassing VMD-SSI and MVMD-SSI. Laboratory model experiments indicate robust identification stability across different excitation environments, with coefficients of variation and MAC values outperforming conventional SSI. Regarding damage identification, the method exhibits notable sensitivity to changes in structural dynamic characteristics, though its capability to detect subtle damage such as single-member stiffness reduction remains constrained. Field measurement data over two months confirm engineering effectiveness, with coefficients of variation for each modal frequency below 0.013 under complex marine conditions. These findings suggest the proposed method holds promise for providing reliable technical support for offshore platform health monitoring.
Feng et al. (Tue,) studied this question.