State Constraints for Predictive Control with Air Vehicle Application

摘要

The object of a predictive controller is to move the states of a plant from some finite initial conditions to some finite and conditions in a time-optimal, or sub-time-optimal way. To do this the plant is driven in an on-off manner; the controller calculates when the plant drive direction should be switched to give suitable control. The constraint algorithms allow such control to proceed while simultaneously enabling any state or combination of states to be restricted to a prescribed level. Algorithms are outlined for putting hard constraints on the values of the state variables in a system controlled using predictive (fast model) techniques and are applied to an air flight trajectory problem. Two algorithms are described, the first dealing with restrictions on the N-1th state of an N'th order system and the second dealing with restrictions on the remaining state variables. To comprehend the constraint procedure, the principles of predictive control are briefly described with the aid of time history plots and phase plane portraits. Results are presented for constraints simultaneously applied to variables in a third order system and for trajectory problem. The constraint algorithms are independent of the predictive control strategy.