| Space-manipulator-based combined spacecraft dynamic modeling and post-capture control technology are the keys to the success of many space applications such as on-orbit assembly,space debris removal,asteroid capturing and on-orbit servicing.Therefore,many efforts have been made in this research area and it is one of the major orientations for the development of aerospace technology.Few attentions have been focused on the contact dynamics between the target and end-effector during the current research.However,for non-cooperative target capturing and operation missions,the end-effector relies on the static friction and torsion to realize the post-capture operation.Due to the limited contacting surface and grasping power,the maximum static friction can be provided by the end-effector may not be sufficient,which implies that the relative sliding motion between the target and end-effector is inevitable and must be taken into consideration.To fix this problem,the objective of this paper is to establish a complete new control framework for the non-coorperative target combined spacecraft by taking the unilateral contact constraint into consideration.In order to achieve the research objectives cited previously,the combined space post-capture dynamics and control technology is profoundly studied in this paper.The innovative points have been proposed on four aspects: the dynamic modeling,the relative sliding motion measuring and inertial parameter estimation,the post-capture controller design,and the Artificial-Neural-Network(ANN)-based controller adaptation.1)On the dynamic modeling aspect,the relative sliding motion between the target and end-effector is taken into consideration and the extra Degree-of-Freedom(ex-Do F)approach is adopted to model this sliding movement.Then the ex-Do F method is projected into the Lagrangian principle to obtain the dynamic model of the combined spacecraft formed by the target and space manipulator.2)On the relative sliding motion measuring and inertial parameter estimation aspect,the stereo-vision based motion sensor is combined with a laser-distance sensor to effectively extract the relative motion information between the target and end-effector.Furthermore,a recursive filtering algorithm is proposed based on the initial guess of the unknown inertial parameters of the target,which can efficiently estimate the target’s inertial parameters with the presence of the relative sliding motion.3)On the combined spacecraft post-capture control aspect,the relative sliding motion between the target and end-effector is considered.The classic reaction null space control approach is modified and implemented into the ex-Do F based dynamic model,and a pattern recognition algorithm is proposed based on the switch between different working cases.With these innovations,the new controller is established to carry out the accurate position and attitude control of the combined spacecraft while minimizing the relative sliding motion between the target and end-effector.4)On the ANN-based controller adaptation aspect,a new ANN framework is established based on the quantum-interference principle.It is proved that the new ANN framework can be seen as a Universal Approximator.Based on the new ANN activation function,an optimal learning rule is proposed to replace the classic DELTA learning rule,which can improve the performance of the ANN with high learning rate to follow high frequency signals.The new ANN is then integrated into the combined spacecraft post-capture control system to realize the error tracking and compensation control.The proposed methods and algorithms are verified by the theorecital analysis and numerical simulations.The simulation results show the correctness and reliability of the new methods,and the work presented in this paper stands as a potential solution for combined spacecraft post-capture control for future space missions. |
