Stable model identification of dynamics identification and controls experiments (DICE)

System identification is often a precursor to model-based control system de signs that assume the plant dynamics to be controlled are known. Many syste m identification algorithms, such as Observer Kalman Filter Identification (OKID), do not guarantee that the identified model will be stable and when applied to flexible structures that exhibit rigid modes, the identified mod els are often unstable. These unstable models can create problems for model validation and subsequent model order reduction for control design. By exp onentially curve fitting the unstable Markov parameters generated by OKID, this instability can be effectively identified and removed allowing stable models to be identified. This system identification technique is validated by computer simulation as well as experimentally using the Dynamics Identif ication and Control Experiment (DICE), a flexible spacecraft emulator desig ned to fly on the NASA space shuttle.