A Predictive Approach for Adaptive Inferential Control

To improve the regulatory performance of the adaptive inferential control system, a cautious on-line parameter estimation algorithm and an adaptive predictive model for unmeasured load disturbances are proposed. The cautious on-line parameter estimation algorithm is used to provide reliable model parameter values in the presence of frequent changes in unmeasured disturbances. Then, the future effect of load disturbances is predicted through the use of an autoregressive model based on the estimated load distrubances. Simulation results demonstrate that better regulatory performance can be obtained with a proper selection of the prediction horizon.