DUAL-RATE DIGITAL ADAPTIVE CONTROL TECHNIQUES

Abstract

In this thesis, a dual-rate digital adaptive control technique is proposed and applied successfully to the design of an adaptive digital Smith Predictor for processes having long dead time. The dual-rate adaptive control scheme makes it feasible to use a larger sampling interval for on-line parameter estimation to reduce computation load as well as to improve robustness, while keeping a smaller sampling interval for control to maintain near continuous-time properties of a digital control system. The increase in complexity in designing a dual-rate adaptive digital Smith Predictor is quite modest. Three design modifications are required, namely the additional digital low-pau filtering or data averaging of input-output measurements, conversion of the identified slow model parameters to those of a fast model having the same frequency as the controller and matching of the d.c. term, and they are simple to incorporate. Model conversion can be done simply and accurately by means of the pole-zero mapping technique.