Design of NARMA L2 Neurocontroller for Nonlinear Dynamical System

K. Srakaew, V. Sangveraphunsiri, S. Chantranuwathana, and R. Chancharoen (Thailand)


Nonlinear Trajectory Control, Neural network.


This paper proposed a technique based on NARMA L2 neurocontroller to control a trajectory of a nonlinear plant. The NARMA L2 neurocontroller was first trained to cancel both the nonlinearity and dynamic of the system. Then, it was reconfigured to become a controller. Once the NARMA L2 neurocontroller suppresses both the nonlinearlity and dynamic behaviour, the closed loop system becomes implicit algebraic relation between the input and the output. Consequently, the system is able to perfectly follow a smooth reference trajectory even it is generated in realtime. This technique was successfully implemented to control the trajectory of a nonlinear disk plant to follow a sine sweep trajectory and also to follow the input device in realtime. Thus, it is a remarkable candidate for a trajectory controller for a nonlinear system.

Important Links:

Go Back