Fault-Tolerant Control of Full-Thrust Satellite Rendezvous and Docking Based on MPC and DNN

To effectively implement fault-tolerant control in full-thrust satellite attitude-orbit coupling, an innovative active fault-tolerant control method combining Model Predictive Control (MPC) and Deep Neural Networks (DNN) has been introduced. This approach accounts for uncertainties in the chaser satellite’s mass parameters, actuator saturation, faults, and unknown disturbances. Initially, nonlinear dynamic model of the attitude-orbit system, including a propulsion system fault model, was developed. Following this, a fault-tolerant control strategy based on MPC was proposed. To enable real-time detection of the thrust fault ratio factor, a data-driven fault diagnosis method utilizing DNN was devised. Fault-tolerant control was then realized by compensating for the identified fault factor. Finally, the simulation results indicate that the proposed method can effectively achieve autonomous docking of the satellite under various thruster fault scenarios.