Peter Wojsznis

Robust predictive controller in object-oriented implementation

Abstract This paper presents a model predictive control (MPC) algorithm and its application in a robust MISO controller with constraints handling. The robustness is achieved by defining an objective function which penalizes output errors significantly more at the end of the prediction horizon than at the beginning. To suppress penalties on control errors in the immediate future, the patience factor is defined in terms of how many steps ahead control errors are not penalized. Constraint control is achieved by using a virtual setpoint concept. Based on this concept, the controlled variable setpoint is moved as required to handle constraints on auxiliary variables. A controller prototype has been implemented in C++. The user interface and the basic principles of predictive control programming are summarized.