State Feedback Control and Kalman Filtering with MATLAB/Simulink Tutorials

Preface

Between PID Control and Model Predictive Control

Control systems are usually designed using one of the two types of mathematical models: transfer function model and state space model. The PID controller is an example of transfer function‐based design and is very popular in control system applications because of its simplicity in the design, analysis and implementation. However, the PID controller suffers from performance loss (mild or severe), when it is applied to higher order systems or systems with interactive dynamics. Among the second category of control systems designed using state space models are the model predictive controllers¹. The model predictive controllers are widely used in industrial applications and are suitable for systems with complex dynamics and interactions between many input and output variables. Based on optimization techniques in real‐time, the model predictive controllers have the advantage of maintaining optimality for control systems in the presence of operational constraints. As a result, process efficiency and profit margins may improve. On the other hand, the deployment of online optimization techniques with operational constraints causes complexity in real‐time computation and implementation, and in an industrial environment it also leads to higher cost of maintenance and commission of such a complex control system.

When we consider the two major groups of control systems, there is an opportunity for a compromise. This compromise exists ...