Pose Synchronized Planning And Control Method For Spacecraft Close Proximity With Multiple Constraints

On account of dynamic time-varying,pose coupling features in the noncooperative target proximity mission with multiple constraints and uncertainties,simultaneously considering the requirement for autonomy,rapidity and robustness in maneuver guidance and control,the pose synchronized planning and control method is investigated in this thesis mainly around the aspects of constraints handling,performance index selection and on-line application feasibility,and the contents are detailed as follows:The characteristics of close proximity mission to noncooperative target are analyzed along with the mathematical descriptions of constraints enforced by system dynamics,sensor pointing,maneuver safety,control limitations and berthing or docking conditions.In view of performance index including energy consumption,target's light-of-sight(LOS)angle and time of flight,the problem of pose synchronized planning with multiple constraints are formulated and then solved employing pseudospectral method.Results indicate that the related solution trajectories carry the risk of constraint violation especially for time-optimal planning,which can be mitigated by incorporating LOS angle into the compound index.Additionally,the computational performance evaluation reveals significant fluctuations in time consumption,which implies that the pseudospectral method is inadaptable for the dynamic situation of close proximity within a short duration.Given the fast and stable on-line application requirement,the convex programming method with initial guess insensitivity and global convergence guarantees is invoked to solve synchronized planning for close proximity maneuver.For non-convex constraints in the problem,convexified attitude motion model in terms of Modified Rodrigues Parameters is derived for the affine relaxation of system dynamics;then the path constraints enforced by sensor field-of-view(FOV),collision avoidance and control ability limitations are relaxed using second-order cone,rotating hyperplane and lossless convexifications respectively.Considering the overrelaxations or unbounded biases occur during convexifications of these constraints with high-dimensional states and controls,a dynamic trust regions constrained sequential convex programming(SCP)method is proposed to improve the convergence efficiency and initial guess adaptability for iteratively solving the energyefficient problem.In view of the requirement of excellent visual relative navigation during the maneuver,the time integral of LOS angle is added into the performance to restrain the angular distance from the sensor optical axis,which leads to significant nonconvexity and makes direct SCP solving intractable.Accordingly,the LOS angle associated performance is redefined based on the local convexity of the cosine function and relaxed using the composite function gradient,then the convex subproblem with energy and LOS angle compound performance is constructed for iteratively solving.For the problem with free final time,a scaling factor variable is introduced to normalize the system dynamics with respect to time,and the nonconvexity further increases due to the augmented decision variable;to facilitate the feasibility during early iterations,virtual control variables are incorporated into the convex relaxation of system dynamics to develop a improved SCP method for solving the more general problem with free final time and compound performance index,consequently extending the application scope of convex programming method.To guarantee the control performance of the pose maneuver with uncertainties,perturbations,disturbances and noises,a double sliding surface control method is presented for rapid and stable trajectory tracking based on piecewise affine models obtained from sequential convexifications,and the selection conditions of control gains are provided through stability analysis of the closed-loop system.Further in view of time delay in relative navigation,with the paradigm of model prediction,a receding horizon control method using SCP is proposed with on-line trajectory generation and commands update,resulting the implicit close-loop guidance and control for close proximity.Finally,mathematical simulations are performed to verify and evaluate the proposed methods,revealing the rapidity,effectiveness and stability of SCP method for multi-constrained guidance and control in noncooperative target close proximity missions.