Uncertainty Quantification of Data-Driven Stochastic Subspace Identification Based on Dynamic Programming

Modal-parameter identification is a critical technique in structural vibration, with extensive applications in structural health monitoring, damage identification, vibration control, and design optimization. Among available methods, stochastic subspace identification (SSI) has gained wide attention for its accuracy in modal-parameter estimation and stability in implementation. However, the data-driven stochastic subspace identification method (Data-SSI) generates numerous intermediate variables during the identification of modal parameters in vibration systems, resulting in high computational costs when applied to uncertainty issues of vibrating structures. To address this issue, this study incorporates dynamic programming to redefine intermediate variables, refine the algorithmic formulation, and simplify the overall structure of the method. Building on this improvement, an efficient uncertainty quantification approach for modal parameters based on Data-SSI is developed. The proposed method is validated through comparisons with the traditional Data-SSI approach, demonstrating enhanced effectiveness, accuracy, and robustness in terms of identification results and computational efficiency under diverse operating conditions.