Modelling and Neural Control of Satellites with Flexible Appendages

摘要

This paper is concerned with the design and comparison of attitude neural controllers for satellites provided with large flexible appendages. Two different control architectures are analysed: a predictor-corrector and an adaptive-critic control system. Using the capabilities of the neural networks to approximate nonlinear dynamics systems, the neural controllers are accomplished in order to orient the satellite toward the desired target while minimizing the strain energy of the panels. The results are given in term of performance comparison and control activity.