Premium
Model predictive optimal averaging level control
Author(s) -
Campo Peter J.,
Morari Manfred
Publication year - 1989
Publication title -
aiche journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.958
H-Index - 167
eISSN - 1547-5905
pISSN - 0001-1541
DOI - 10.1002/aic.690350408
Subject(s) - model predictive control , control theory (sociology) , controller (irrigation) , settling time , mathematics , flow (mathematics) , minification , optimal control , integrator , stability (learning theory) , mathematical optimization , internal model , volumetric flow rate , computer science , engineering , control (management) , control engineering , step response , mechanics , physics , computer network , geometry , bandwidth (computing) , artificial intelligence , machine learning , agronomy , biology
The infinite‐horizon, discrete‐time optimal averaging level control problem for surge tanks, with minimization of the rate of change of outlet flow as its objective, is formulated and a solution is presented. A finite moving‐horizon approximation is introduced and analytical solutions are obtained for two important special cases. These results provide a quantitative measure of the impact of a secondary objective, integral action, on flow filtering. The problem is then generalized to include nonconstant level and outlet flow constraints. A model predictive control formulation is presented which addresses the objectives of the generalized problem. The resulting controller minimizes the maximum rate of change of outlet flow, provides integral action, and handles constraints on the tank level and outlet flow rate. The proposed controller includes a single adjustable parameter that directly effects the trade‐off between the incompatible objectives of good flow filtering and rapid settling time. Examples are presented to demonstrate the properties of the model predictive controller. An implementation, involving imbedded feedback, is developed which guarantees internal stability of the model predictive scheme for open‐loop unstable processes (such as integrators).