Multitask-based Trajectory Planning And Anti-disturbance Control Of Redundant Space Manipulator

Redundant space manipulator performs one task in most current research,whereas the spatial activities are developing like on-orbit assembly and base construction.Therefore,the tendency of performing multiple tasks is inevitable,in order to improve efficiency and save energy.The basic prerequisites for performing multiple tasks are the best task ordering and the optimal manipulator movement.However,the methods about performing one task are not applicable to multiple tasks.Therefore,this dissertation studies the multitask-based trajectory planning of redundant space manipulator.Efficient and fast planning are realized in each case of optimal end-effector path,optimal maneuvering time and optimal energy cost.Besides,an anti-disturbance control strategy is developed,since actual motion of redundant space manipulator is usually affected by parameter uncertainty,white noise and other factors.The main work and results are followed.(1)Multitask-based trajectory planning of redundant space manipulator for optimal end-effector path.The problem is decomposed into the sequential order problem of task points and the multiple point-to-point trajectory planning problems(MPTPs).First,task points are optimally ordered using a hybrid particle swarm optimization algorithm.The constructed objective functions consider attitude variation and path length of end-effector.The control parameters of algorithm are adaptive.Second,the trajectory planning between adjacent task points,namely MPTPs,is converted into an optimization problem.The manipulator redundancy is considered.The constructed objective function considers minimum base attitude disturbance.An improved particle swarm optimization algorithm(IPSO)is then used to solve the optimization problem.The updating mechanism of population in IPSO is improved,the exploration and the exploitation capabilities are thus balanced.Simulation results verify the proposed method.The end-effector accurately visits each task point.Besides,minimization of attitude variation,path length and base attitude disturbance is realized.(2)Multitask-based trajectory planning of redundant space manipulator for optimal maneuvering time.The problem is converted into an optimization problem.The manipulator redundancy is considered.The constructed objective functions consider maneuvering time and base attitude disturbance.An improved genetic algorithm(IGA)is proposed to solve the optimization problem.Three types of genes are simultaneously optimized in IGA.Therefore,each chromosome contains a task-points order,a manipulatorconfigurations order and an unknown function coefficient for joint trajectories.The three parts are separately encoded.The crossover and the mutation are also separately performed in the three parts.Task-points ordering and manipulator trajectory between adjacent task-points are simultaneously optimized.This method avoids low accuracy in decomposition method due to the predefinition of end-effector.Heavy calculation burden is also avoided,because of repeatedly performing algorithm and frequently calculating Jacobian matrix.Simulation results verify the proposed method.The end-effector accurately visits each task point.Besides,minimization of maneuvering time and base attitude disturbance is realized.(3)Multitask-based trajectory planning of redundant space manipulator for optimal energy cost.The problem is converted into an optimization problem.The manipulator redundancy is considered.The constructed objective functions consider joint-torque accumulation and base-disturbance accumulation.The optimization problem is then solved using an IGA.The update mechanism of population is improved,by optimizing the initial population,the encoding mechanism,the crossover probability and the mutation probability.Therefore,the efficiency of IGA is improved.Simulation results show that the convergence speed of IGA is improved,and the quality of solution is guaranteed.The end-effector accurately visits each task point.Besides,minimization of energy cost,maneuvering time and base attitude disturbance is realized.(4)Anti-disturbance control of redundant space manipulator using non-linear model predictive approach.The sequential quadratic programming method is employed to obtain the predictive control sequence at each sampling time,based on the non-linear model of space robot.The first element in sequence acts on the real system.The proposed method avoids frequently changing models and repeatedly calculating matrix parameters in traditional methods.A penalty function of state variables is constructed for last predictive domain in sequence.Besides,an inequality constraint about the state variables is established using the linear quadratic regulator method.The asymptotic stability of the closed-loop system is finally guaranteed.Simulation results show that the redundant space manipulator moves from initial state to desired state accurately,considering disturbance factors like physical parameter uncertainty,white noise and control input delay.