System Identification and Multivariable Control Design for a Satellite UltraQuiet Isolation Technology Experiment (SUITE)

Abstract

The main objective of the’Satellite UltraQuiet Isolation Technology Experiment’ is to attenuate the vibration in the braodband up to 100 Hz, and at 5-, 25-, and 100-Hz corner frequencies in a six-degree-of-freedom hexapod system mounted on a satellite. To design the controller for vibration isolation, the identification of the plant transfer function was necessary. The hexapod assembly model was identified using various band-limited test signals such as chirp, white Gaussian noise, and pseudorandom binary signals. Using the observed responses to these test signals, a 6 x 6 transfer function matrix relating six piezo-actuator input voltages and the six geophone-sensor output voltages was obtained in the discrete-time domain. Various model structures such as auto-regression with exogeneous inputs, auto-regression moving-average with exogeneous inputs, and Box-Jenkins (BJ) were used for system identification. The transfer functions obtained by a fifth-order BJ model were validated in the time and frequency domains. Their orders were matched with the order of the electromechanical model of the combined piezo-geophone system. Various multi-input multi-output control design methodologies such as linear quadratic Gaussian and H∞ were proposed for active vibration isolation up to 100 Hz. The simulation results using these controllers achieved 33- and 12-dB attenuation at 5- and 25-Hz corner frequencies, respectively.