Trajectory Tracking Control Of 6-PSS Parallel Robot

The parallel robot mechanism is simple,has strong bearing capacity and has unique advantages in structure.It is a research hotspot in recent years and is widely used in industrial production,motion simulation and vibration platform.The parallel robot does not require much work space,but it requires high precision and stability of the control system.In practical application,parallel robots are disturbed by their own coupling relationship and external environmental factors,and the traditional control strategy has the problem that the parameters are not easy to adjust,so it is difficult to meet the control requirements of nonlinear systems.Hence,this paper takes 6-PSS parallel robot as the research object to seek a simple,reliable and robust neural network control strategy.Firstly,the structure of 6-PSS parallel robot is expounded,the three-dimensional model of robot is drawn,the dynamic coordinate system and the fixed coordinate system are established,and the kinematics model is established by using the principle of coordinate transformation according to the geometric relationship of the fixed length rod of the branch chain,and the Jacobian matrix is obtained by the inverse kinematics solution.In order to ensure the parallel robot to complete the motion task continuously and smoothly,the trapezoidal trajectory planning is used to make the driving units of each branch chain accelerate the motion in the start stage and slow down the motion in the stop stage,and the moving pair displacement trajectory of the six branches is obtained,tracking time optimization from 0.5s to real-time tracking,which lays the foundation for the trajectory tracking control.Secondly,considering the coupling relationship of 6-PSS parallel robot and the influence of external interference on the control system,PID traditional control parameters are fixed and can not meet the dynamic demand for control.The decentralized control strategy is adopted to control the parallel robot branch chain accurately and stably to achieve the high precision control effect on the whole robot.The robot branch chain controller is designed by combining the RBF neural network with the PID control strategy.The controller input is theexpected motion trajectory of each branch chain.The RBF neural network dynamically adjusts the PID parameters according to the error values of the expected trajectory and the actual trajectory,realizes the adaptive control of the branch chain and improves the dynamic characteristics of the system.Finally,considering that the network parameters can not be guaranteed to be the global optimal solution during the training of RBF neural networks by gradient descent method,two intelligent algorithms are introduced to train RBF neural networks.The RBF neural network parameter value is taken as the particle initial position of the particle swarm optimization algorithm,and the particle position is updated during the continuous iteration process,and the global optimal solution is obtained to transfer to the RBF neural network PID controller.Using multi-population genetic algorithm,multiple sub-populations simultaneously train the initial value of the RBF neural network.The error of the RBFNN-PID controller is taken as the fitness function,and the migration operator is added.The optimization process is accelerated by manual selection to obtain the global optimal solution.PSO and multi-population genetic algorithm use the controller error as fitness function to adjust the optimization search mode,which has strong robustness,control can be restored within 0.05 s after being interfered,and the anti-interference is stronger,improves the ability of RBFNN-PID controller,and improves the trajectory tracking control effect of 6-PSS parallel robot.