Dynamic Modeling And Control Of A 2-DOF Redundantly Actuated Parallel Robot

In comparison with serial robots, parallel robots have very good performances in terms of rigidity, accuracy and ability to manipulate large loads. In order to avoid or reduce singularity, and optimize motor torques, actuation redundancy is introduced into parallel robots. However, due to the closed-loop existing in parallel robots, the motions of these mechanisms are rather complex. Meanwhile, it is hard to control parallel robots, because of the MIMO system with time-varying, strong coupling and nonlinear dynamic properties. In this paper, a 2-DOF redundantly actuated parallel robot was taken as the object of study. The study contents mostly included kinematic analysis, electromechanical coupling dynamic modeling, control system design, and experimental test.The effects of assembly configuration on the velocity and acceleration performance were studied. All types of assembly configuration were found by the forward position analysis and the inverse position analysis. Then the velocity performance index and acceleration performance index were introduced into the performance analysis of the configuration. The Jacobian matrix and the Hessian matrix were established by adopting the influence coefficient matrix. According to the given velocity and acceleration performance index, the corresponding performance atlas was plotted. Considering the workspace analysis, the singular configuration, the velocity performance and the acceleration performance, the optimal type of assembly configuration was pointed out. So the blindness of mechanism design can be reduced and the efficiency of optimization can be raised. And it is beneficial to realize high precision trajectory tracking control of parallel robots.Dynamic deals with the relationship between the input torque of motors and the displacements, velocities, and accelerations of the parts of the system. In order to develop an effective controller for optimal trajectory tracking performance, the dynamics of a mechanism must be thoroughly analyzed. The steps of establishing the electromechanical coupling dynamic model were as follows:Firstly, by using the Lagrange method we got three open-loop two-bar mechanisms, by loading constraints we got the dynamic model of the parallel mechanism. The modeling process was simplified and the practical performance of the models was enhanced by using the mature dynamic model of the open-loop two-bar mechanism. Secondly, the dynamic model of the permanent-magnet AC servomotor was formulated using the field orient control principle. Thirdly, two dynamic models were integrated via torque pass, so we got the electromechanical coupling dynamic model. Lastly, through dynamic analysis, we got the trajectory tracking results of the end-effector, the angular displacement of the three active joints, and the motors'current.The nonlinear proportional integral differential (PID) is insensitive to the change of system parameters, and the diagonal recurrent neural network (DRNN) control has the nonlinear approach ability. So considering both of the advantages of the nonlinear PID and DRNN, we designed a new compound intelligent controller. Comparing the new compound intelligent controller with the traditional PID controller and the diagonal recurrent neural network self-tunning PID controller, we found that the new compound intelligent controller had the best performance not only for the circular trajectory with initial errors, but also for the symmetric trapezoidal trajectory. The new compound intelligent controller overcomes deficiency of the traditional single controller, and has a good robustness, so it is helpful to improve the tracking performance.In order to deal with external disturbance, sliding mode control schemes were introduced into trajectory tracking control of the parallel robot. The terminal sliding mode control not only guarantees the existence of the sliding phase of the closed-loop system, but also guarantees that the tracking error converges to zero under limit time. In order to deal with modeling error and external disturbance, the neuro-sliding mode control was derived. The RBF neural networks were used to approximate the discontinuous part of control gain in a classical sliding mode controller. Through the numerical simulation, the effectiveness of the sliding mode control methods was verified.The experiments on the dynamic properties and trajectory tracking control of the 2-DOF redundantly actuated parallel robots were carried out. Tests of vibration and frequency characteristic were finished. The electromechanical coupling dynamic model was verified by the symmetric trapezoidal trajectory tracking control experiment. On the basis of traditional PID controller, velocity and acceleration feed-forward control was introduced to improve the trajectory tracking performance.The work of this dissertation will lay a solid foundation for further application study of redundantly actuated parallel robots, and will give reference to theory and application study of other types of parallel robots. This will help parallel robots operate with high quality.