Multi-input Multi-output System Identification Using Impulse Responses

Abstract

This paper presents a new algorithm for multi-input multi-output (MIMO) system identification in the time domain using impulse responses. The algorithm is suitable for the on-orbit system identification of spacecraft using the responses to thruster impulse inputs measured by typical satellite on-board sensors. The Eigensystem Realization Algorithm (ERA) realizes a multi-input multi-output (MIMO) system using asynchronous impulse responses in the time domain. Our new method identifies the input and output matrices of a MIMO collocated system by applying a recursive least-squares iteration scheme to refine the matrices obtained from conventional ERA. In this manner, the input matrix is considered to be constructed by the mode shape vectors and the actuator sensitivity matrix. A numerical simulation of an actual spacecraft, the Engineering Test Satellite-VI (ETS-VI), is performed to verify the algorithm. The nominal dynamics model of ETS-VI, which has six rigid body modes and fourteen elastic modes due to large flexible solar panels, is excited by six body-mounted thrusters, and the translational velocities and attitude rates are measured simultaneously. Our method successfully identifies all of the fourteen natural frequencies, damping ratios and mode shape vectors, confirming its validity.