Research On Posture Control Of Pneumatic 3-UPU Parallel Robot

The development of robotics is an important symbol and embodiment of the country's high-tech level and industrial automation.3-UPU robot is a kind of parallel robot with few degrees of freedom,which has the advantages of simple structure,strong bearing capacity,high micro precision and fast moving speed.In this paper,a pneumatic 3-UPU parallel robot is designed,the robot has the advantages of low cost,high power quality ratio,safety and environmental protection,and the ability to operate in a complex space.It has great research value in the future.Firstly,the control system and mechanical structure of the pneumatic 3-UPU parallel robot are designed.The virtual prototype of the robot is assembled in Solid Edge,and the physical prototype of the robot is built.The kinematic analysis of the robot is carried out,including the analysis of the degree of freedom of the robot mechanism,the solution of the inverse and forward solutions of the robot's position,the simulation solution of the Jacobian matrix and the analysis of singularity of the robot structure,and the simulation of the robot's workspace.Secondly,the mathematical model of valve-cylinder system is established for a better understanding of the characteristics of the controlled object.The third-order transfer function of the pneumatic proportional position system is obtained through the valve port flow equation of the proportional valve,the flow continuity equation of the air filling cavity and the exhaust cavity,and the cylinder piston force balance equation.Thirdly,considering that the pneumatic proportional position system is a typical nonlinear system,the traditional PID controller is difficult to achieve good control effect.With the strong nonlinear function approximation ability of neural network,the RBF neural network supervisory controller is designed by using the feedforward and feedback control method,and the control simulation is carried out;at the same time,due to the partial simplification in the modeling process,the model of the controlled object is inaccurate,and partial format model-free adaptive controller can identify the pseudo-gradient of the system through the input and output data of the controlled object,then realize the control.Therefore,a partial form model-free adaptive controller is designed,and control simulation was carried out.Finally,the control performance of RBF neural network supervisory controller and partial format model-free adaptive controller is verified based on MATLAB platform.The simulation results show that the control performance of the two controllers is better than that of the PID controller.The steady-state accuracy of the robot's pose is 0.01 mm,and the dynamic tracking error is less than 0.1mm.