Experimental study of event-based neural network control on parallel manipulator

In the paper, the event-based switching controller (ESC) is utilized to achieve better tracking performance compared with the neural network controller in circumstance of parameter uncertainty and unknown disturbance for the parallel manipulator. The ESC optimizes the choice of the neural weights by combining the prior knowledge of the system dynamics and the estimation of the system parameters. To implement the controller, a general method of computing the system regression matrix for the PM is proposed and the stability proof is given in circumstance of unbounded disturbance. The ESC is tested by the simulated Delta manipulator and the experimental 5R testbed. The results show the effectiveness of the proposed controller.