基于多维振动响应GSC-TARMA模型的时变结构模态参数辨识

The problem of output-only identification of time-varying structures based on vector vibration response measurements is considered. A generalized stochastic constraints vector time-dependent auto-regressive moving average (GSC-VTARMA) model is presented, which is an extension form of the generalized stochastic constraints time-dependent ARMA (GSC-TARMA) model. In order to reduce computation complexity, an improved generalized stochastic constraints vector time-dependent auto-regressive moving average (GSC-TARMA*) model is subsequently proposed. The proposed model is validated by non-stationary vibration signals of a numerical system with time-varying stiffness and a laboratory time-varying structure consisting of a simply supported beam and a moving mass. The results indicate that the proposed GSC-VTARMA* model achieves same identification accuracy but less computation complexity to the GSC-VTARMA model, and achieves better identification robustness and higher data utilization to the GSC-TARMA model. Furthermore, the proposed model shows a similar identification accuracy and lower computational cost than the traditional FS-VTARMA model. Due to the recursive algorithm used by the GSC-VTARMA* model, its enhanced online identification capability has also been demonstrated.