基于系统输出的时变特征参数辨识

在子空间辨识方法的基础上给出时变特征参数辨识的新方法。将特征阻尼修正与数字滤波相结合，由输出信号估计特征频率和阻尼比。对于系统输出，假定其具有短时平稳特性，通过对响应信号在局部时问段上建模得到时变特征。使用分量能量指标指示信号中各振动分量的强弱，并通过能量指标在时频面上建立一种时频表示，以此反映系统的时频特性。对Hankel矩阵使用QR分解，在提高运算速度的同时不降低参数估计的精度。

Time-Varying Eigen-Parameter Identification Using Output Data Only

It is of great significance to identify time-varying eigen-parameters from system output only. On the basis of subspace identification technique, a new method has been proposed in the paper to estimate frequency and damping ratio, which adopts a modification procedure to estimate damping ratios that are otherwise distorted by data correlation on short time intervals. Moreover, digital filtering is applied to suppress the fluctuation of estimated parameters. To ensure the validity of subspace modeling on short time intervals, short-time stationarity is imposed on system output. Also in the paper, Component Energy Index is used to indicate vibration intensity of signal components and to establish a kind of time-frequency representation, which enhancs the exhibition of time-varying characteristics. To reduce computation complexity and guarantee the accuracy of estimation, QR decomposition is applied to decompose Hankel matrices. Simulation of a system with time-varying stiffness is presented and the effectiveness of the propose method is demonstrated.