一种改进的航天器时变模态参数递推辨识方法

针对基于预测器的递推子空间辨识(RPBSID)方法在估计系统的状态变量时计算量较大的问题，提出一种改进的RPBSID方法并应用于航天器的时变模态参数辨识。与原算法相比，改进后的方法在求解状态量时不需要逐个时刻构建相应的Hankel矩阵，而是利用仿射投影算法(APA)实现状态量的递推估计，从而减少了辨识过程中的数据量。在此基础上，利用该状态变量递推得到时变系统的状态空间模型和模态参数。在数值仿真中，建立带有大型挠性附件的卫星动力学模型，分别考虑系统模态参数线性变化、突变和周期改变的情况，利用改进的RPBSID方法对结构的时变频率和阻尼比等参数进行了辨识。理论分析和数值仿真的结果表明这种改进的方法不仅能够有效地辨识系统的时变模态参数，而且与原方法相比具有更高的计算效率。

An Improved Recursive Identification Method for Time Varying Modal Parameters of Spacecraft

 The recursive predictor-based subspace identification (RPBSID) algorithm bears heavy computation burden in estimating the system state variables. In this paper, an improved RPBSID method is presented and applied to identify time-varying modal parameters of the spacecraft. Compared with the original algorithm, it does not need to construct the corresponding Hankel matrix in each time instant for the improved method when solving the state variables. The affine projection algorithm (APA) is used to implement recursive estimation of the state variables, thereby reducing the amounts of data in the identification process. Moreover, the state-space model and corresponding modal parameters of the time-varying system are determined recursively by the identified state variable. In the numerical simulation, the dynamics model of a satellite with large flexible appendages is established. Three cases that may factors the modal parameters to vary with time, such as the linear variation, step change and periodical variation are investigated. The improved RPBSID method is applied to identify the structural time-varying frequency and damping ratio parameters. Compared with the original algorithm, the results of the theory analysis and numerical simulation illustrate that the proposed method not only can effectively identify the system time-varying modal parameters, but also features higher computational efficiency.