This paper presents Successive-Linearization based Model Predictive Control (SL-MPC) paradigm to de-tumble and control CubeSats. The control problem is particularly challenging because of the high non-linearity of the attitude dynamics, magnitude uncertainty of initial angular velocity after the deployment, limited actuators authority. computation power available on-board and battery limitation. Today, mostly Bdot controllers are employed to address the control problem because tumbling spacecraft might have as high as 30 degrees/sec initial angular rate and Bdot controllers are able to minimize the rate error regardless of the attitude of the spacecrafts. Similarly, the de-tumbling problem is formulated such that the angular rates can be minimized. Additionally, the algorithm also has the ability to explicitly handle the input constraints in the system. This feature has the potential to take into account the magnitude of the Earth's magnetic field and avoid saturation of reaction wheels, which must be closely monitored to avoid failure of the mission. Furthermore, the algorithm is real-time implementable because the control problem is formulated as convex optimization problem such that deterministic convergence properties and finite time solution are guaranteed. The proposed SL-MPC algorithm has three distinctive features when compared to other real-time implementable convex optimization based algorithms. First, the nonlinear spacecraft attitude dynamic model is completely covered with the proposed SL strategy and those models are taken into account over the prediction horizon within the MPC framework, which allows the controller to include and preview the future dynamic models behavior to represent the nonlinearity in advance. Second, as small spacecrafts mostly use Earth's magnetic field and reaction wheels for de-tumbling and pointing purposes, the algorithm is not only able to formulate the limits of the actuators but also the variations of it. This feature is p
展开▼
