2016 Volume 59 Issue 3 Pages 150-160
Today, the primary identification methods in time domain for spacecraft are based on singular value decomposition (SVD), such as the eigensystem realization algorithm (ERA) or stochastic subspace identification (SSI), which requires significant computation time. However, some control problems, such as self-adaptive control, need the latest modal parameters to update controller parameters online. To improve computational efficiency, the fast approximated power iteration (FAPI) recursive algorithm, which avoids SVD, is applied as an alternative method to identify the time-varying frequencies of large flexible satellites. Moreover, an improved recursive form is also proposed to obtain the time-varying input matrix in a state-space model by rewriting the relation of the input and output data. In numerical simulations, the time-varying model of the Engineering Test Satellite-VIII (ETS-VIII) is established. The results illustrate that this recursive algorithm can implement time-varying parameter online identification and it has a better computational efficiency than the SVD-based methods.