A method for robust identification and control of a flexible space structure

Abstract

The author suggests that the large space structure (LSS) control design problem, by its nature, has dimensions of interrelationship among the modeling, dynamics and control synthesis disciplines additional to those found in most other control design problems. The feasibility of applying robust identification and control techniques to the LSS control problem is explored. The identification module is a key subsystem. The identification method applies a canonical variate analysis (CVA) technique to selected input and measurement signals of the process to be identified, and thereby extracts model parameters. Model order is statistically determined using built-in criteria. By using such computationally stable algorithms as singular value decomposition, CVA identification is well-suited to LSS applications. The results presented, for three models, indicate clearly that CVA identification and model predictive control are capable of providing robust control of flexible structures. >