On-orbit Identification Methods For Time-varying Dynamic Parameters Of Large Flexible Spacecraft

The full-scale identification experiment of dynamic parameters of large flexible spacecrafts usually can not be implemented on the ground. When the finite element method is adopted to carry out the dynamic analysis, the model simplification and computational error will make it difficult to obtain the exact values of model parameter. In this case, the on-orbit identification of spacecraft parameters should be considered to evaluate the system dynamic parameters accurately and reflect the on-orbit operational situation of spacecraft structure. Moreover, some conditions, such as appendage motion and fuel consumption, could lead to dynamic parameters changing of spacecrafts. Therefore, the on-orbit identification of time-varying dynamic parameters of spacecraft can not only verify the proposed mathematical model but also provide reference basis for control system design and control parameter modification of the spacecraft, which has important engineering application in practice.In this dissertation, the author focuses on the identification of time-varying dynamic parameters (modal parameters and state-space model parameters) of large flexible spacecraft during on-orbit operation. The identification of periodical time-varying dynamic parameters, the recursive identification of time-varying modal parameters and state-space model parameters, and the time-varying dynamic parameters of the spacecraft closed-loop system, are studied respectively. The main contents of this dissertation are given as follows:(1) The on-orbit identification of periodical time-varying modal parameters and state-space model of spacecraft is studied in Chapter 3. Considering the influence on dynamic parameters due to the rotation of large flexible appendages, the on-orbit identification of the periodical time-varying modal parameters and corresponding state-space model parameters of the flexible spacecraft is implemented using the periodic time-varying subspace approach instead of the common repeated experiments method. Numerical results demonstrate that the periodical subspace approach is effective in the identification of time-varying dynamic parameters of spacecraft.(2) The recursive algorithms for the on-orbit identification of time-varying modal parameters is studied in Chapter 4, and an improved recursive subspace algorithm, to increase the identification efficiency, is then proposed. Firstly, the computation cost of the common on-orbit identification method based on singular value decomposition (SVD) is large. Therefore, the Projection Approximation Subspace Tracking (PAST), Approximated Power Iterations (API) and Truncated Window Approximated Power Iterations (TW-API) recursive algorithms, which are based on the signal subspace projection theory, are employed to identify the time-varying modal parameters of spacecraft. Then, the applicable condition and computational efficiency of the three recursive algorithms are summarized and compared, respectively. Moreover, an improved TW-API method is developed to increase the computational efficiency of on-orbit identification. Comparing with the traditional TW-API method, the improved algorithm guarantees the computational accuracy and significantly reduces the computation time of on-orbit identification by simplifying the matrix recursion procedure in data processing. Particularly when the order of system model is high, the advantage of computational efficiency is much more obvious. In numerical simulation, the computational efficiency between the improved TW-API algorithm and periodical subspace method are compared. The simulation results illustrate that the recursive subspace algorithms can identify the time-varying modal parameter of spacecraft effectively and have a higher computational efficiency than the classical identification method based on SVD. Finally, the wavelet de-noising technology is applied to the input and output data to improve the identification accuracy of recursive algorithm in lower signal-to-noise ratio.(3) The recursive algorithms for the on-orbit identification of time-varying state-space model parameters is studied in Chapter 5, and a novel recursive form for identifying the time-varying input matrix is presented. In contrast with the frequently-used repeated experiments method, the proposed recursive form is derived from the signal subspace projection theory. The time-varying input matrix of system is obtained from the new signal subspace matrix by reconstructing the relation between the input and output data, and then the complete state-space model parameters are determined. Comparing with the existing identification algorithms based on repeat experiments, the proposed approach, without using the SVD, can significantly reduce the computation time.(4) The on-orbit identification of time-varying dynamic parameters for the spacecraft closed-loop system is studied in Chapter 6, and a novel recursive form to identify and verify the output feedback gain parameters is presented. By constructing the augmented matrix of the system, the least squares method is employed in this recursive form to identify the time-varying feedback gain matrix. The new recursive form avoids the non-uniqueness problem of identified matrix parameters when the least squares method is directly used in the process of matrix inversion. The identified gain matrix parameters can be applied to verify the designed controller model parameters. The simulation results show that the proposed recursive form is effective in the identification of the time-varying output feedback gain matrix parameters of the spacecraft. In addition, the applications of the aforementioned periodical subspace method and recursive subspace algorithms are also addressed to provide important references for the identification of time-varying dynamic parameters in the closed-loop spacecraft systems.