Title :
Adaptive Dynamic Matrix Control: Online Evaluation of the DMC Model Coefficients
Author :
Freedman, Richard W. ; Bhatia, Alok
Author_Institution :
Tulane University, Department of Chemical Engineering, New Orleans, LA 70118
Abstract :
Dynamic Matrix Control (DMC), an internal model control algorithm based on a linearized system model, has been used successfully as a process control algorithm for more than a decade. According to current literature, the model coefficients are evaluated by physically pulsing each input to the system, and by recording the responses of the various outputs. Since the model is a linear approximation of the system dynamics, its validity degrades as the system parameters change with time. This paper describes a method, called Online Coefficient Evaluation (OCE), of evaluating the DMC model coefficients without disturbing the system. When DMC is used in conjuction with OCE, the result is an adaptive control algorithm, called Adaptive Dynamic Matrix Control (ADMC), which is capable of automatically adjusting to time-varying system dynamics. An example of the application of ADMC is included.
Keywords :
Adaptive control; Automatic control; Chemical engineering; Chemical processes; Feedback; Heuristic algorithms; Process control; Programmable control; Robust control; Robust stability;
Conference_Titel :
American Control Conference, 1985
Conference_Location :
Boston, MA, USA