Hybrid Two-Stage Identification-Based Nonlinear MPC Strategy for Satellite Attitude Control

Abstract

The control reliability of model predictive control (MPC) for satellite attitude is inextricably linked to the accuracy of the prediction model describing the satellite dynamics. In contrast to most existing work, which uses mechanism models as prediction models for MPC design, this article proposes a novel nonlinear MPC (NMPC) strategy based on the multivariate radial basis function-based autoregressive model with exogenous inputs (M-RBF-ARX model). To sufficiently learn the satellite dynamic characteristics, a hybrid parameter identification algorithm is presented for the M-RBF-ARX model, which consists of two identification stages: particle swarm iterative identification and multivariate hierarchical multi-innovation stochastic gradient identification. Derived from the identified M-RBF-ARX model, a hybrid two-stage identification-based NMPC strategy is proposed using sequential quadratic programming as the optimization algorithm. To overcome the possible model mismatch problem caused by uncertainty in satellite parameters and external disturbances during on-orbit control, an online parameter correction module is introduced. A simulation study is conducted to verify the feasibility of the proposed strategy in satellite attitude control.