Effect of Smaller Output Horizon in Neural Generalized Predictive Control

This paper deals with the effect of small prediction horizon in Neural Generalized Predictive control. A different strategy is proposed to overcome the problem of estimation at each sampling instant. In this method the parameter are estimated at a large sampling interval and control increments are calculated at a smaller sampling interval. Simulation studies are presented to show the merits of smaller prediction horizon over larger prediction horizon. Studies are presented to show the merits of smaller prediction horizon over larger prediction horizon and enable one to use variable output horizons, resulting in considerable saving of cost of simulation and computer time.