Adaptive Backstepping Control Based On Neural Network In The Application Of A Parallel Robot

As a new kind of robot,parallel robot has gradually become the hot spot in scholars with its features of big toughness,high bearing capacity,high accuracy,and good dynamic performance and so on.Low-DOF parallel robot has a better development and application prosperity for its few drive elements,low cost,simple and compact structure and high practicality value.It is practical significant in industrial production that study on its kinematics and control strategy.As the parallel robot is a nonlinear system of highly complex and strong coupling which exists uncertain factors such as parameter errors,un-modeled dynamics,measurement noise and external disturbance,the conventional control strategy is difficult to meet the control requirement.Backstepping design method is an effective means to deal with uncertainty,especially non-matched uncertainties which apply to uncertainty nonlinear systems that can be stated linearizable or transformed into parametric-strict-feedback form.Therefore,this paper proposed a new type of backstepping control method with neural network adaptive in studying on 2-DOF redundant parallel robot.With that method,RBF neural network is used to approximate non-deterministic function,backstepping control as a master controller output.The method does not need to establish an accurate mathematical model,and has strong robustness of the uncertain factors in the parallel robot.On the basic of analyzing position solution in forward and inverse,working space,branch model and path planning of the parallel robot,the paper construct the control system hardware platform with PC and multi-axis motion controller firstly. After that,does simulation to each branch of the parallel robot system using the control method of neural network adaptive backstepping.The simulation results show that the control method has high control accuracy and strong robustness,and can track the expected trajectory well.In the end of this paper,we use object-oriented software VC++ to design a control interface with its functions thoroughly,combined with multithreading programming idea and dynamic link library technology in windows system environment.The user interface thread is to accept user’ input,and worker thread calculate the complex algorithm and send the control value to the control card. So that parallel robot can complete the given trajectory in the working space.The experimental results demonstrate that the control scheme is feasible and effective.