SUPERVISED ADAPTIVE PREDICTIVE CONTROL USING DUAL MODELS

Abstract A new adaptive control method based on supervision is presented. The algorithm only updates the control estimates when new relevant information is obtained. The technique is based on a dual-model approach. The first model acts as the control model and the second model is the supervisor which determines when the control model should update its estimated parameters. Simulation results using a CSTR illustrate the set point tracking capabilities of the algorithm. Experimental studies on a heat exchanger showcase how the algorithm successfully suppresses output bursting.

[1]  Michael Athans,et al.  Robustness of continuous-time adaptive control algorithms in the presence of unmodeled dynamics , 1985 .

[2]  Eduardo J. Dozal-Mejorada,et al.  Stability of supervised adaptive control , 2007, 2007 46th IEEE Conference on Decision and Control.

[3]  K. Narendra,et al.  Bounded error adaptive control , 1980, 1980 19th IEEE Conference on Decision and Control including the Symposium on Adaptive Processes.

[4]  Robert R. Bitmead,et al.  Directional leakage and parameter drift , 2004 .

[5]  B. Ydstie.  Transient performance and robustness of direct adaptive control , 1992 .

[6]  D. Mayne,et al.  Design issues in adaptive control , 1988 .

[7]  B. Erik Ydstie,et al.  Adaptive control with selective memory , 2004 .