A dual-rate adaptive digital smith predictor

Abstract

The limitations of a single-rate adaptive digital Smith Predictor are discussed. The combination of using a larger sampling interval for on-line parameter estimation and a smaller sampling interval for control, leads to the development of a dual-rate adaptive system. The necessary design modifications for the dual-rate system, namely the additional digital filtering of measurements, the conversion of slow model parameters to those of the fast model and the matching of the dc term, are simple to incorporate. The dual-rate system gives superior responses for set-point following and load rejection with a smaller control interval, while it improves the robustness of the model parameter estimation in the presence of unmodelled dynamics with a larger estimation interval. The superior performance of this new design is substantiated by computer simulation and experiments on a heat exchanger pilot plant. © 1989.