Self-tuning prediction and control

An attempt is made to summarize the vast amount of self -tuning algorithms that have recently appeared in the control literature. This is done by correlating the different methods into a generalized self-tuning predictor. This unifying algorithm contains several typos of self-tuners as special members using appropriate design filters. This generalized algorithm can be used to implement the different strategies. The presented method is based essentially on the prediction point of view. The self-tuning aspect is entirely embodied in the predictor while a fixed parameter controller is used. The controller consists of feedforward from the reference input and feedback from the predicted system output (instead of the real system output). The coupling of the self-tuning predictor with the fixed parameter controller makes the whole control system self-adaptive. The self-tuning system does not suffer from opening or closing of the control loop during its operation nor from alternating between several controllers. This is a real advantage enhancing its practical applicability. It opens the possibility of using a self-tuner in a commissioning mode during the initial stages of the parameter estimation