Control theory is an interdisciplinary field of engineering and mathematics , dealing

with the behavior of dynamic systems . The input of a system is called a reference .
When one or more output variables of a system need to follow a certain reference over
time, a controller manipulates the input to the system to obtain the desired effect on the
output of the system ( feedback ). Feedback can be negative (self-compensatory
regulation) or positive ("snowball" or "vicious cycle " effect).

A control system is a set of devices responsible for managing, ordering, directing or

regulating the behavior of another system or variables, in order to reduce the
probabilities of failure and obtain the desired results (control the variables).

Control system is the set of devices that act together to achieve a control
objective . Industrial control systems are typically used in industrial production
processes to control equipment or machines. There are two common classes of control
systems, open loop systems and closed loop systems:

Open loop control system

It is that system in which the process only acts on the input signal and results in an
output signal independent of the input signal, but based on the first. This means that
there is no feedback to the controller so that it can adjust the control action. That is, the
output signal is not converted into an input signal for the controller.

These systems are characterized by:

 Be simple and easy to understand.

 Nothing ensures its stability in the face of a disturbance.
 The output is not compared to the input.
 Being affected by disturbances. These can be tangible or intangible.
 The precision depends on the previous calibration of the system .

Closed loop control system

They are systems in which the control action is a function of the output signal. Closed-
loop systems use feedback from an end result to adjust the control action accordingly.
Closed-loop control is essential when any of the following circumstances occur:

 When a process is not possible to regulate by man.

 A large-scale production that requires large facilities and man is not capable of
managing.
 Monitoring a process is especially difficult in some cases and requires attention that
man can easily lose due to fatigue or absent-mindedness, with the consequent
risks that this may cause to the worker and the process.

Their characteristics are:

 Be complex, but broad in number of parameters.

 The output is compared to the input and affects it for system control.
 Your feedback property .
 Be more stable to disturbances and internal variations.

There are linear dynamic systems and nonlinear dynamic systems.

If two solutions are known for a linear system , the sum of them is also a solution; This
is known as the superposition principle. In general, solutions from a vector space allow
the use of linear algebra and significantly simplify the analysis. For continuous linear
systems, the Laplace transform method can also be used to transform the differential
equation into an algebraic equation.

For discrete linear systems , the Z transform method can also be used to transform
the differential equation into an algebraic equation. Nonlinear systems are much more
difficult to analyze and often exhibit a phenomenon known as chaos , with totally
unpredictable behavior.

Co-design Control, Computing and Communication.

A host of challenges face the field of feedback control, especially when the new
environment that differs substantially from that of the last fifty years is rapid. In any case, rapid
advances in computing, communication, sensors and networking also give rise to
unprecedented opportunities for Real-Time Control Systems. The field of control systems
science and engineering is entering a golden age of unprecedented growth and opportunity
that will likely dwarf the advances spurred by the space program of the 1960s.

These growth opportunities are spurring enormous advances in computer technology,

materials science, sensor and actuator technology, as well as the theoretical foundations of
control and dynamic systems. Many of the opportunities for future growth lie at the
boundaries of traditional disciplines, especially at the boundary of computer science with
other engineering disciplines. Control systems technology is the cornerstone of the new
automation revolution occurring in areas as diverse as home appliances, consumer electronics,
automotive and aerospace systems, manufacturing systems. . .

As mentioned above, the new environment characterized by broad applications of

networked embedded devices, the traditional standard design pattern, allowing the separation
of different disciplines, will be the longest and most valid when applied in real-time systems
engineering. of control. To address this problem, principles and methods in the computing and
communication communities must be incorporated into the design and implementation of
control systems.

Feedback control theory and technology can, of course, both be adapted in computing
and communication systems to achieve better performance. This interaction is giving rise to
the emerging wave of information technology revolution, the convergence of computing,
communication. In the last half century, the fields of computing, communication and control
have achieved great successes in isolation.

The creation of ENIAC, the first large-scale general-purpose electronic computer at the
University of Pennsylvania in 1946, is a symbol of the birth and evolution of modern
computing. The paper titled "A Mathematical Theory of Communication" by Shannon,
published in 1948, lays the foundation for information theory. The publication of Wiener's
book "Cybernetics" in 1948 symbolizes the naissance of control science, a discipline of which
the principles and methods have been used today in different engineering systems. The rich
theoretical and technological results achieved in these disciplines have formed a solid basis for
the development of a new unified theory of the system as a result of convergence. For
example, recent accelerated progress in computer technology, communications, and sensing
technology brings new opportunities for real-time monitoring.

Integrated processors and wireless networks will certainly become the characteristic
components of future Real Time Control Systems. The convergence of computing and
communication has given birth to a number of important technologies, of which the most
representative examples are the Internet and wireless technology.

On the other hand, a good number of academic conferences and journals covering
multiple disciplines such as computing, communication and control have been launched in
recent years. These phenomena indicate that the strict traditional boundaries between these
disciplines will become increasingly less rigorous. Instead, the dynamic interaction between
these disciplines will be a key feature of information technology.
 To promote the emergence of this multidisciplinary area, The Panel of the future.
Directions in Control, Dynamics and Systems (experts) have recommended that government
agencies and the control community should significantly increase research directed at control
integration (computing, communications and networking). Several institutions have started
initiatives. For example, the US Department of Defense has already made a large investment in
this area through the “Multidisciplinary University Research Program (MURI)” initiative. It has
been reported that the US Department of Defense plans to award research grants totaling
more than $150 million to academic institutions under this program over the five years starting
in 2006.

In the context of real-time control, there are intuitively two kinds of convergences:
integrated control and computing, and integrated control and integrated communication. The
integration of computing, communication and control offers a new methodology to implement
real-time control in dynamic environments. Following this methodology, it is therefore possible
to realize the co-design of computing, communication and control in control systems
engineering, which is in contrast to the traditional design pattern that separates control and
programming. The entire control system implementation process will never again consist of
these two separate stages. While controller design will take into account the limitations of
application platforms, well-designed control algorithms will be implemented by systems
engineers with the timing requirements of control applications in mind. In this way, a
fundamental process of dynamic interaction is established between computing,
communication technology and control systems, believed to be capable of contributing to the
optimization of system performance.

However, there are still doubts about how to build a more holistic theory that can
serve future advances in the convergence of computing, communication and control. Since
resource scheduling becomes the primary concern in this context, this area is often referred to
as co-design control and scheduling or integrated control and scheduling, encompassing both
the areas of integrated control and computing and integrated control. and communication.

In the last decade, there has been growing interest in control and programming co-
design. Some of the efforts made regarding integrated control and computing or integrated
control and communication, respectively, in the following two sections. The emphasis is placed
on:

1) the application of feedback programming to computer and communication systems

2) Co-design of programming and control of Real-Time Control Systems

 Integrated Control and Computing
Most of the existing jobs in the area of integrated control and computing can be
divided into two categories:

1)Computerized control systems: (In the first case, the applications considered are
usually general-purpose computer systems (i.e. non-control systems), and the key technology
used for resource management is Feedback control/programming.) As the name says, the
basic idea behind computerized control systems is the application of feedback control theory
and techniques to computer systems in such a way that the desired performance is achieved.

Feedback control provides the engineering community with a technology that allows
addressing uncertainty and improving the flexibility of systems. For example, when feedback
control is introduced, external and/or internal disturbances can be significantly improved by
computer systems. Thus also improving the determinism of the system during runtime.
Feedback control can also be used to compensate or reduce the negative effects of execution
platforms on system performance.

In particular, for computing systems operating in dynamic open environments,

feedback control of resource management could improve the flexibility of the system in a
context of uncertainty in resource availability, which is of vital importance to ensure the
satisfactory temporal behavior of the systems. Due to the inherent complexity of computing
systems, in most cases it is difficult to build mathematical models (precise enough) to
formulate their dynamics. Sometimes even selecting an appropriate type of model is extremely
difficult.

2) Embedded Control Systems ( embedded systems are designed to meet specific needs. ):
Embedded control systems considered in the context of control and programming are
generally composed of a set of digital control loops. The control algorithm within each control
loop is implemented separately, as a real-time task. Naturally, the shared resource of interest
is CPU time. Consequently, the most common goal of co-design is to optimize the overall
control quality of the system under CPU resource constraints.

Control and Communication Integration

In the area of integrated control and computing, the resource that requires our
attention is the computational resources (capacity), while, in systems for integrated control
and communication, the resource that requires the most special attention will be the
communication resource. , for example, network bandwidth. As done in the previous section,
two categories of related work are identified in the area of integrated control and
communication: network control systems and wireless control systems.
 The area of network control is quite broad, involving a variety of aspects of network
communications. The most representative application of wireline network control theory and
technologies is “network congestion control”, for example, end-to-end rate control of IP
networks and queue management in the routers. In history, the design of congestion control
algorithms in traditional networks, often performed by systems engineers, is rarely associated
with feedback control theory.

Classical control techniques are successfully applied for active queue management,
where a proportional controller and a PI controller are designed and analyzed. A derived
component is also added to the PI controller, thus forming the more general PID framework.
To improve the stability and robustness of the system, the authors present the adaptive
control approach to design an active queue manager using fuzzy logic control, which offers
better performance than the PI control based approach. Compared to wired networks,
resource management in wireless environments could be much more difficult. This is mainly
because wireless networks transmit data through radio channels, which mostly characterizes a
high level of non-determinism. The channel capacities of wireless networks depend largely on
many factors such as transmission power, bandwidth and time slot size, and are thus
potentially variable, which contrasts with wired networks which always Link capacities will
remain fixed.

1) Network Control Systems: Closed network control systems (through wired networks) are the
natural result of the integration of control and communication. In modem control applications,
networked control systems rely on both traditional control networks and general-purpose data
networks such as Ethernet and the Internet to transmit data and exchange information. It can
be classified into three categories:

(1) Controller design-based approaches: The key to controller design-based approaches is the
design of control algorithms that are robust to network delays induced by packet losses.

(2) Network design-based approach: Network design-based approaches are primarily

concerned with how to improve the quality of service of the network in such a way that the
timing requirements of the control systems are met.

(3) Control and network co-design:

2)Wireless control systems: Recent years have witnessed the emergence of wireless
networking technologies that are increasingly more mature, the availability of a large number
of devices using them are increasingly commercialized and cheaper. . All these factors make it
possible to carry out real-time control over wireless networks. In fact, the field of wireless
control is continuously attracting attention from academia and industry for its incomparable
advantages over traditional wired networked control systems, such as better support for node
mobility, easier installation and maintenance, and lower costs.

However, it is very difficult to achieve effective control based on wireless networks

(e.g. WiFi, ZigBee and Bluetooth) which are much less reliable than wired networks. The
impact of varying wireless channel fading on control performance is studied, where the
authors suggest that controller parameters should be dynamically adapted with respect to
channel conditions.

Classic Control Systems Architecture

Until well into the 20th century, the only analytical tools that the control specialist had
were the use of ordinary differential equations together with algebraic criteria to
determine the position of the roots of the associated characteristic equation. Applying
the Routh and Hurwitz criterion, the engineer determined the stability or not of the
systems, but for this the mathematical model had to be obtained by operating through
differential equations. This involved hard work. Furthermore, it should be noted that the
Routh and Hurwitz criterion does not offer information on how to improve the stability of
the system.

From a theoretical point of view, Control Engineering begins to consolidate when the
knowledge acquired in signal amplification problems is transferred and applied to
industrial control problems. These studies lead to the so-called Classical Control
Theory, in which the Laplace and Fourier Transformation methods and the external
description of the systems were used as mathematical tools.
The theory developed for process control, from the classical and modern point of view,
has its essential basis in the knowledge of the dynamics of the process that is desired
to be controlled. This dynamics is normally expressed using ordinary differential
equations, and in the case of linear systems, the Laplace transform is used to obtain a
mathematical representation that relates the signal to be controlled and the input signal
to the system. This mathematical relationship is known as the transfer function.

From classical control theory, considering the simplest case of a linear system with one
input and one output, the dynamics can be represented as in the following figure.

This figure represents the block labeled "Process" or "Plant", which is the system that
you want to control. Two signals arrive to this system, one labeled "Control input" which
will be the signal generated by the controller to be designed and the signal labeled
"Uncertain input" which can represent any undesirable signal external to the system
and which is Also known as "disturbance" or "noise". Finally, the "Output" signal, which
will be the signal that you want to behave in a certain way. The output signal is also
known as a controlled signal.

Classical control theory makes extensive use of the concept of transfer function (or
transmittance). The analysis and design are carried out in the s (Laplace) domain
and/or in the frequency domain. Modern control theory, which is based on the concept
of state space , extensively uses vectormatrix analysis. Analysis and design are
performed in the time domain.

Classical control theory generally provides good results for one-input, one-output
control systems. However, classical theory cannot handle multi-input, multi-output
control systems.
 Computer Control Architecture

The complete digital control system can be considered divided into two parts:

 An analog part, which contains the physical laws of the system itself, since
most real systems work in an analog way. Sensors and actuators are commonly
added to this.
 A digital part in which we find the control laws. These laws must therefore be
treated as a discrete system.
 An analog-digital interface that allows communication between the two
parties.
Analog-Digital Converter

An analog digital converter mainly performs two operations:

 Sampling of the signal, that is, its temporal discretization, with a fixed period that is generally
denoted as “T”. A discrete signal is a signal that has a value only at a few precise moments.
These time instants are defined by the sampling period, which is a constant time value that
separates two of these instants. An ideal discretizer can be seen as a switch that turns on every
second, for a very short instant of time, and is off the rest of the time.

 The quantification of the signal, that is, the discretization of the values that can be used so
that the signal is compatible with the representation format of computers. For example, a
computer that works with 8 bits can process only 256 different values. The information that
enters and leaves this will be sequences of numbers such as: 123, 135, 34, 0, 255, etc.

Digital-Analog Converter

A digital-analog converter, on the other hand, aims to obtain again the previously discretized
analog signal exactly equivalent for all time “t”. In the general case, the perfect reconstruction
of the analog signal is impossible, since different functions in time can have the same discrete
value

The operation of a discrete control system then depends on:

 The sampling frequency, which has an influence on the precision of the measurement
and the delay added by the zero-order blocker.
 The precision of the numerical system, which can adopt only precise numerical values,
therefore has an influence on the quantization error (Amplitude Quantization Error).

Communication is carried out discontinuously but uniformly distributed over time. The control
stage is implemented in a device of a discrete nature that is only capable of receiving and
generating discrete signals. Therefore, analog-digital and digital-analog conversion devices are
necessary that act as an interface between the analog world of the process under control and
the digital world where the control law is implemented.
The most common way to implement discrete control is to use a regular sampling period (T),
which means that the sampling moments of the variables of interest and the moments of
action of the controller on the plant will be equally spaced in time and will coincide. with
integer multiples of the sampling period.

In a more real situation there should be a difference of one sampling period between the
capture of a sample and the application of the control action generated from the information
it provides, since the calculation of the action will require a non-zero time. . This delay with
respect to the ideal situation will have to be taken into account in the design of the control law
so that it does not have a significant influence on the system's performance. In any case, in a
conventional discrete control strategy the instants of sampling and application of actions are
perfectly determined and distributed over time in a regular manner.

The discrete implementation of control has undeniable advantages, among them the
possibility of implementing complex control strategies that could not be achieved with
continuous components.

However, it has the drawback that the control-plant communication is interrupted periodically.
Only during sampling moments does the system behave as a true feedback control structure,
leaving the plant without control during the entire inter-sampling period.

Other times, it is the time cost of executing the control algorithm or the A/D and D/A
conversions that determines this limit.

A final possibility is that the limitation on the sampling frequency is imposed by the
communication link, through which the information is transmitted between the controller and
the plant.
 Distributed Control Architecture

History of Distributed Control Systems (DCS)

• SCDs were created to replace single-loop controllers and process computers that had
a single central processor.

• The availability of the first microprocessors made it possible for Honeywell to be the
company that launched the first Distributed Control System in 1974, the TDC 2000
system (Total Distributed Control). It was so successful that in a short time it became
the most widespread system in the process industry.

• The architecture of a DCS is made up of multiple processors, each of which controls a

process unit of a plant, so that in the event of a failure it is only that part that is left
without control

• Distributed control systems have an optional redundant configuration. The TDC2000

control system had one backup controller for every maximum of eight active
controllers.

• SCDs initially designed for analog process control quickly evolved into hybrid systems
that also handled digital input/output signals (early 1980s)

Why did they arise?

Direct digital control (DDC) during that era suffered from a substantial problem: THE POTENTIAL
danger of a failure in a single digital computer that controlled or executed MULTIPLE PID control
loops, functions that should never be stopped. Digital control brought many advantages, but it was
not worth it if there was the risk of the operation completely stopping (or catastrophically failing)
followed by a hardware or software failure on a single computer.

Distributed controls are intended to solve this concern by having multiple computers, each
responsible for a group of PID loops, distributed throughout the facilities and linked to share
information between them and with the operation consoles. Now there was no longer the worry of
having all the links on a single computer. The distribution of computers or controllers also organized
the signal cabling, since now hundreds or thousands of instrument cables only have to reach the
distributed nodes, and not all the way to the centralized control room. Only network cables had to be
linked to the controllers, representing a drastic reduction in cabling needed. Furthermore, distributed
control introduced the concept of REDUNDANCY in industrial control systems: where digital signal
acquisition and processing units were equipped with a "spare" to automatically take control of all
critical functions. in case a primary failure occurs.

Definition
A Distributed Control System consists of the link, through a communications network, of various
physically distributed nodes, equipped with process capacity and linked to sensors and/or
actuators. These systems are characterized by the fact that the control process takes place in
these nodes in a coordinated manner. Communication networks oriented to the link of these
nodes are also known as communication buses or multiplexed networks. A node is an
autonomous processor with its own hardware: processor (CPU), memory, clock oscillator,
communications interface, and interface to the subsystem it controls.

On the contrary, in a centralized control system there is a single controller where all the input
signals to be sampled converge, are processed by carrying out all the necessary control
algorithms and all the necessary output signals are generated. Centralized systems give rise to
expensive and heavy point-to-point wiring (from each sensor or actuator to the centralized
system) and the use of analog networks (4-20mA) both for the connection of sensors dedicated
to capturing input signals and for the activation of indicators.

Initially, the main reason for the migration from centralized systems to distributed systems was
the need for simplification and standardization of wiring, based on the philosophy of replacing
copper (expensive point-to-point wiring) with silicon (intelligent nodes linked by a serial bus over
low section twisted pair). However, there are additional reasons why a distributed system is
preferable such as shorter design time and lower operation and maintenance costs.

Within distributed systems, the automotive world is a reference sector since it is a market that
requires a large volume of low-cost, reliable components capable of functioning in aggressive
environments. The communications bus used in distributed systems embedded in the
automobile is the CAN bus ( Controller Area Network ) and the study of this bus and the field
buses that take it as a basis is part of this course.
The following figure shows a typical architecture of a Distributed Control System (DCS):

Each rack contains a processor to implement all necessary control functions, with individual input
and output (I/O) cards to convert signals from analog to digital or vice versa. Processor redundancy,
network cable redundancy, and even I/O card redundancy are implemented to prevent component
failure. DCS processors are usually programmed to perform a self-check routine on their redundant
system components to ensure the availability of spare equipment in the event of a failure.

If there is even a total size on one of the control racks, only the PID loops of this single rack will be
affected, no other loops in the system. On the other hand, if the network cables fail, only the flow of
information between these two points would be damaged, the rest of the system continues
communicating information normally. Therefore, one of the "laws" or key characteristics of a DCS
is its tolerance to serious faults: regardless of hardware or software failure, the impact on process
control is minimized by the design.
 Hierarchical control architecture
The hierarchical control structure is derived from a simplification of the control system. When
control is model-based, it is aided by a series of process models that can range from a very
general abstraction to a detailed description. Higher levels of control correspond to long time
scales and control low frequency events, while lower levels correspond to short time scales
and control high frequency events. The control system is implemented with a hierarchical
control structure at three levels. Each level focuses on solving different control problems with
different objectives and even different time scale