Predictive Control

1Brief History and Basic Principles of Predictive Control

1.1 Generation and Development of Predictive Control

Predictive control, later also called Model Predictive Control (MPC), is a kind of control algorithm originally rising from industrial processes in the 1970s. Unlike many other control methods driven by theoretical research, the generation of predictive control was mainly driven by the requirements of industrial practice. For quite a long time the industrial process control mainly focused on regulation using the feedback control principle. The well‐known PID (Proportional–Integral–Differential) controller can be used for linear or nonlinear processes, even without model information, and has few tuning parameters and is easy to use. These features are particularly suitable for the control environment in industrial processes and make it a “universal” controller that is widely used. However, the advantage of the PID controller is mainly embodied in the loop control. When the control turns from a loop to the whole system, it is difficult to achieve good control performance by using such a single‐loop controller without considering the couplings between the loops. Furthermore, a PID controller can handle input constraints but is incapable of handling various real constraints on outputs and intermediate variables. Particularly, when the control goal is promoted from regulation to optimization, this kind of feedback‐based controller seems powerless because it lacks understanding ...