Time-domain identification of low-order models for flexible structures

A new approach to system identification is proposed that applies cubic smoothing splines to the system Markov parameters calculated by Observer Kalman filter Identification (OKID). Applying this smoothing procedure early in the identification process allows models that are both low order and accurate to be directly identified. Experiments with the daisy spacecraft emulator demonstrate significant computational savings and numerical robustness by avoiding the model overspecification and reduction process.