Predictive Control of Position Servo System using DSP

Generalized predictive control (GPC) developed by Clarke et al. is a powerful control method. It is known that GPC has a strong robustness to the parameter mismatch, model order, dead time, etc. In this paper, the authors implement this method to a servo system using DSP (Digital Signal Processor). The predictor used here predicts the future trajectory points. The control input is then determined by minimizing the norm of future errors (i.e. the differences between future trajectory points and reference trajectory points) over a future control horizon. The algorithm is easy to be implemented in a computer, especially in a DSP system. In the experiments, DSP is used to calculate the voltage to be applied to the servo motor. The experimental results show that accurate trajectory following is possible in spite of the simplification imposed to the predictor model and ignorance of the payload.