Long Range Non-Linear Predictive Control

Self-tuners based on linear models have shown their ability to achieve adequate control for special class of processes, such those which are slowly time-varying and weakly non-linear. However, many practical applications are highly non-linear and embarking on designing a linear self-tuner could be a fruitless exercise. In this research the dynamic process to be controlled is described by a non-linear model (namely the Hammerstein model) and a long-range, incremental and non-linear predictive self-tuner is obtained accordingly. Various simulation examples indicated that the long range and non-linear predictive controller gives a superior performance in comparison with linear self-tuners as long as the controlled process is adequately represented by Hammerstein model.