Predictive Control

2Some Basic Predictive Control Algorithms

Based on the three methodological principles of predictive control, a variety of predictive control algorithms can be formed by adopting different model types and optimization strategies as well as feedback correction measures. In this chapter, we introduce some typical predictive control algorithms based on these basic principles, with different model types, aiming at illustrating how the predictive control algorithm can be developed by concretizing these principles. The predictive control algorithms are introduced here for unconstrained SISO (Single‐Input Single‐Output) systems. Predictive control algorithms for MIMO (Multi‐Input Multi‐Output) systems and for constrained cases will be discussed in later chapters.

2.1 Dynamic Matrix Control (DMC) Based on the Step Response Model

Dynamic Matrix Control (DMC) [1] is one of the most widely used predictive control algorithms in industrial processes. Early in the 1970s, DMC was successfully applied to process control in the oil refinery industry [2]. DMC is based on the step response of the plant, and thus is suitable for asymptotically stable linear systems. For a weakly nonlinear system, step response can be tested at its operating point and then DMC can be adopted. For an unstable system, the DMC algorithm can be used after stabilizing the system by a traditional PID controller.

2.1.1 DMC Algorithm and Implementation

DMC algorithm consists of three parts: prediction model, rolling ...