Preface

Model Predictive Control (MPC) is not only the name for a special kind of control algorithms or controllers based on these algorithms, but also the name for a powerful thought in control theory. MPC has a very special originating process; because its industrial application appeared much earlier than its theoretical description, it could solve process control problems without precise theory. However, from the perspective of its essence, this special originating process indicates high coherence between MPC and natural thinking manner of humans, bringing out control problems to be solved. To first predict using a model, and then provide a control law, considering the predicted result, seems to be the most comprehensible, convenient and understandable way in controlling.

However, limited by modeling and optimization method during the early days, MPC could only be used in process industry, using a local linear model and a large sample period. In fact, this poor situation lasted for decades following MPC's birth. In some researchers' minds, MPC still remains a different name for advanced process control. The advantages of MPC were underestimated seriously during those years.

While we are aware that the real world is much more complex than the linear, timeinvariant, lumped-parameter and deterministic model in classical MPC, if we want to have better control performance, these factors must be taken into account. The advantages of MPC can only be demonstrated entirely and clearly if MPC can handle much more kinds of system models.

Fortunately, during the recent years, the rapid development of computational science and technology led to the second "boom" of MPC. Nonlinearity, stochastic character, robustness and many other factors started to be considered. Efficient applicable MPC algorithms have been established using modern computational techniques, such as the genetic algorithm. Theoretical achievements have also been obtained with the efforts of many control scientists. Applications of MPC can now be found in almost all engineering domains.

To start with, this book will introduce the basic structure and the historical development of MPC, for readers who are not so familiar with the topic. Some distinctive examples of recent MPC use will then be presented, both in the theoretical

#### X**II** Preface

and the engineering domain, to illustrate the frontiers of this field. This special structure can help the readers who want to acquaint themselves with MPC in general, while readers who want to study MPC in one particular direction can also get helpful guidance in this book.

The book's authors from around the world appreciate the contributions made by researchers before them, and bear in mind the quote 'we stand on the shoulders of giants'. We would also like to thank all the people who helped us greatly in the writing process, including our colleagues and friends, and especially the zealous managers and editors at InTech. Finally, we thank all of our family members, you are always our ultimate love and help.

> **Tao Zheng**  Hefei University of Technology, China

X Preface

guidance in this book.

always our ultimate love and help.

and the engineering domain, to illustrate the frontiers of this field. This special structure can help the readers who want to acquaint themselves with MPC in general, while readers who want to study MPC in one particular direction can also get helpful

The book's authors from around the world appreciate the contributions made by researchers before them, and bear in mind the quote 'we stand on the shoulders of giants'. We would also like to thank all the people who helped us greatly in the writing process, including our colleagues and friends, and especially the zealous managers and editors at InTech. Finally, we thank all of our family members, you are

**Tao Zheng** 

China

Hefei University of Technology,

**Introductory Chapter** 

**Model Predictive Control:** 

The name 'Model predictive control' exactly indicates the three most essential characters of this kind of controllers, a model can be used to predict the future behaviour of the system, the prediction based on above model and historical data of the system and online optimal

Any model that could be used to predict the future behaviour can be the system model in

MPC itself has no special request on the choice of model, the only need is that the model could predict the future behaviour of the system, no matter how we get the system model and how we obtain the future output by the model. But many researchers still classify MPC into different types by their models, since different model usually lead to quite different optimization method in solution of control law. Because all MPC have the same basic structure, the optimization method may be the most important part of a novel MPC algorithm indeed, and it also can determine the algorithm's practical applicability in industry. In Certain Meaning, the develop history of MPC is mainly the develop history of

When MPC was invented in 1970s, limited by the modelling and computational method, the scientist and engineers often use simple models, such as discrete time linear model (Richalet *et al*., 1978, Culter *et al*., 1980, Rouhani *et al*., 1982 and Clarke *et al*., 1987), to build MPC, while using this kind of models could already satisfy the requirement on control performance in process industry of that days. Later, based on modern control theory, a lot of MPC based on linear state-space system model is proposed (Ordys *et al*., 1993, Lee *et al*., 1994). These mentioned references can also help the readers of this book to understand the basic characters thoroughly if they still have problems after reading this short guidance, because these references were work of the precursors, who paid special attention to explain

But, nonlinearity, constraints, stochastic characters and other complex factors exist naturally

control based on above prediction and certain control criterion.

**1. Introduction** 

**2. The predictive model** 

the predictive model of MPC.

what MPC's essential properties are.

in the physical world, especially in control engineering.

MPC, and it is usually called predictive model.

 **Basic Characters** 

*Hefei University of Technology,* 

Tao Zheng

*China* 
