Research On Identification,Kinematic Planning And Control Strategy In Capture Of Non-cooperative Target By Space Manipulator

With the increase of human space activities recent years,the occupation of orbit resources caused by space failure spacecrafts and space debris and the threat to normal service spacecraft become more and more serious.Considering the limitation of the orbit resources,especially the near earth orbit resources,measures should b e taken to effectively clean up the failed spacecrafts and space debris.In recent years,it has been basically mature to use onboard space manipulator and other means to achieve on orbit acquisition,takeover and maintenance of cooperative spacecrafts.While the acquisition of non-cooperative spacecrafts has gradually become a research hotspot.Therefore,identification of mass characteristic parameters,on-line optimal control of attitude planning with obstacle avoidance and acquisition strategy based on service spacecraft with space manipulator in the process of capturing non-cooperative space targets are carried out.Firstly,aiming at space flexible manipulator system with freee floating base,the kinematics model and rigid dynamics model are established.Based on the rigid dynamics model,the dynamics model of space manipulator with flexible joints inclueded is established according to the Lagrange equation.Then,the dynamics model of space manipulator with flexible arms included is established based on the vibration mode,and the dynamics model of the space manipulator including both flexible joints and flexible arms is established according to dynamics model estalished before.Then,a non-contact incentive strategy based on mass block is proposed to identify the mass characteristic parameters of non-cooperative target.Space manipulator is applied to spray mass block to the non-cooperative target and achieve firm adhesion to it.Hence,parameters such as velocity and angular velocity of the non-cooperative target are changed by mass block.By observing the moving state of the non-cooperative target,the position of center of mass and the ratio of principal-axes-inertia of the non-cooperative target are identified by the distributed robust Kalman filter algorithm.Based on the principle of the conservation of linear momentum,the identification result of the mass of the non-cooperative target is obtained,and the influence of the parameter error of the identification process on the accuracy of the identification result is analyzed.Simulation results show that the strategy can identify the mass parameters of the target better to improve the identification accuracy by adjusting ejected mass.The proposed method can obtain the mass characteristic parameters of the target in advance before capturing,and then the capability of the service spacecraft can be predicted to ensure the safety and feasibility of the whole capture mission.Then,aiming at the problem of obstacle avoidance in the attitude maneuver planning process of the spacecraft,the distance optimal path attitude planning algorithm with spherical geometry constraints(SGC)and the time optimal attitude planning algorithm based on shape function method are proposed respectively.In SGC method of attitude maneuver planning algorithm,the mathematical model of obstacle area between OOS and target position is established respectively,and the SGC algorithm is proposed in detail.Simulation results show that the SGC algorithm of attitude path planing is superior in distance index,and with the expansion of the size of obstacle area,the safety distance of the planned path from obstacles increases simultaneously.In the algorithm of time optimal attitude planning path based on shape function,Bezier shape-based function and finite Fourier series shape-based method are used to establish the continuous attitude planning path described by Euler angle,and Legendre-Gauss discrete points are used to discretize the continuous attitude planning path.Geometric constraints and attitude dynamics constraints are set for each discrete point along the trajectory.Gauss pseudospectral method(GPM)is applied to make a comparision.Simulation results show that the shape function method can quickly realize the attitude maneuver planning path with geometric constraints and attitude dynamics constraints,and the Bezier shape function method is better than the FFS shape-based method in the index of calculation time and accuracy.At last,aiming at the non-cooperative target acquisition problem with free floating space manipulator(FFSM),the DRL control algorithm is applied to train FFSM.In this paper,a deep deterministic policy gradient(DDPG)algorithm including pre-training(PT)strategy is proposed,and model of reward function including indexes of time,distance and control torque is established,which can effectively improve training efficiency of FFSM.Simulation results show that the proposed control algorithm can effectively reduce the relative speed of the capture collision,and improve the attitude stability of the system during the capture process.