Uncertainty-Aware Adaptive Control of Metro-Induced Environmental Vibration Under Unknown Excitation

Metro-induced environmental vibrations impair the serviceability of adjacent vibra-tion-sensitive buildings and their durability through long-term accumulation, posing a challenge to the sustainable use of existing buildings. Due to uncertainty in the struc-tural model and an unknown, non-stationary vibration source, structural vibration control remains challenging. To address these problems, this paper proposes an uncer-tainty-aware method that combines system identification by a generalized extended Kalman filter with unknown input and the linear-quadratic-Gaussian (LQG) active control, for buildings under metro-induced excitation. First, the time-varying structur-al parameters are augmented into the state vector and identified with the state and the base excitation, the latter being recovered under an adaptive weight driven by the posterior covariance to suppress noise amplification. The identified state is then fed back, while the parameters enter the gain synthesis in proportion to their posterior confidence, to keep the control force smooth and maintain the control performance under uncertain parameter estimates. Finally, the noise covariances and the regulator weighting matrices are fixed by a two-stage salp swarm search, thereby improving the identification accuracy and the control performance. The proposed method is exam-ined on a nine-story frame structure under unknown metro excitation, with model mismatch and measurement noise, to verify its effectiveness.