The aim of this book is: 1) to provide an introduction to conventional system
identification, model predictive control, and control performance monitoring,
and 2) to present a novel subspace framework for closed-loop identification, data-
driven predictive control and control performance monitoring.
Dynamic modeling, control and monitoring are three central themes in sys-
tems and control. Under traditional design frameworks, dynamic models are the
prerequisite of control and monitoring. However, models are only vehicles to-
wards achieving these design objectives. Once the design of a controller or a
monitor is completed, the model often ceases to exist. The use of models serves
well for the design purpose as most traditional designs are model based; it also
introduces unavoidable modeling error and complexity in building the model. If
a model is identified from data, it is obvious that information contained in the
model is no more than that within the original data. Can a controller or monitor
be designed directly from input-output data bypassing the modeling step?
This book aims to present novel subspace methods to address these questions.
In addition, as necessary background material, this book also provides an intro-
duction to the conventional system identification methods for both open-loop
and closed-loop processes, conventional model predictive control design, con-
ventional control loop performance assessment techniques, and state-of-the-art
model predictive control performance monitoring algorithms. Thus, readers who
are interested in conventional approaches to system identification, model predic-
tive control, and control loop performance assessment will also find the book a
useful tutorial-style reference.