HIERARCHICAL MODEL PREDICTIVE CONTROL SYSTEM BASED ON ACCURATE MANIPULATION VARIABLE PREDICTION

A hierarchical model predictive control system has a decentralized control system having a plurality of PID controllers each having a plurality of control modes for a PID calculation of a manipulation variable for controlling a process variable of the control object, and for controlling the control object on the basis of an obtained manipulation variable; and a model predictive controller for reading each of a control mode and control information of the PID controllers at a predetermined control cycle, for predicting a process variable future value of the control object by using a dynamic characteristic model of the control object on the basis of the control mode and control information read from the PID controllers, for sequentially calculating the manipulation variable about each of the PID controllers in the manner that a performance function with respect to a deviation and a manipulation variable between a predicted process variable future value and a process variable reference value added in advance becomes a minimum value, and for transmitting a calculated manipulation variable as a set point value corresponding to the control mode to the corresponding PID controller. The PID controller performs a PID calculation corresponding to the control mode on the basis of a set point value supplied from the model predictive controller or a set point value set by an operator.