Trajectory Planning And Control For Vibration Suppression Of Flexible Manipulator

Compared with traditional manipulators,flexible manipulators have the advantages of light-weight,low energy consumption,and high operational speed.However,the light-weighted material also causes vibration during the movement of flexible manipulators.Vibration not only effects operational accuracy and speed in a negative way,but also accelerates the fatigue damage of materials,shortening their service life.Thus,researching how to suppress the negative influence of flexible manipulators' vibration when they are performing given tasks is a significant task.Trajectory planning and control are two main methods to solve this problem.In terms of trajectory planning for flexible manipulators,the existing researches only focus on vibration suppression for simple planar flexible manipulators.However,when flexible manipulators work in three-dimensional space,the vibration of flexible links becomes more complex;in addition,when flexible manipulators have rigid-flexible coupling phenomenon,the complex dynamics would increase the difficulty of vibration suppression.According to these problems,based on particle swarm optimization,a trajectory planning algorithm is proposed to achieve vibration suppression for three-dimensional rigid-flexible manipulators.Initially,based on Absolute Nodal Coordinate Formulation and Lagrange Principle,the dynamic model of a rigid-flexible three-dimensional manipulator is established in this paper.Then according to the vibration principle,fifth order polynomials are adopted as trajectory functions for each joint.After that,this thesis deduces the sufficient and necessary conditions for acceleration constraints of each joint,develops an objective function for trajectory planning,and transfers trajectory planning problem into an optimization problem.Finally,via solving this optimization problem by particle swarm optimization,the desired trajectory is produced.The simulation demonstrates the effectiveness of the proposed algorithm to suppress the residual vibration.In terms of control of flexible manipulators,the existing methods are all dependent on dynamics,i.e.the controller design process is based on the analysis of flexible manipulators' dynamics.When the configuration of flexible manipulators is complex,the dynamics has serious coupling and underactuated problems.This will cause obstacles to analyze dynamics,and therefore the existing approaches will no longer apply.According to this problem,a dynamics-independent control method is proposed in this thesis,i.e.the controller design method is independent on dynamics of flexible structure.Initially when considering the rigid movement of flexible manipulators,a position-tracking controller in the joint space and a trajectory-tracking controller in the operational space are designed separately.The neural networks are applied as contemporary controller to solve the vibration problems.Finally,a training algorithm is proposed to train the aforementioned neural networks based on reinforcement learning.The simulation demonstrates that when conducting tasks in the joint space,the proposed control method can track the desired position and suppress the vibration simultaneously;when conducting tasks in the operational space,the proposed control method can track the desired trajectory accurately.Besides,the proposed control method is a general approach and can be expanded to other flexible manipulators.