An experimental evaluation of long range predictive control

The generalized predictive control is proposed as an important contender for high performance servo applications. A simple formulation of this approach is presented and its implementation for accurate trajectory control of a servo system with a nonlinear load is described. The single joint of an industrial manipulator is used as a model in the experimental setup. The system is based on a MC68000 microprocessor which implements the control algorithm and a PWM amplifier driving a servomotor with a nonlinear load. The control algorithm aims for the minimization of a performance index which is chosen as the norm of the tracking error over a future control horizon. Experimental results are presented and discussed.