Control theory

Control theory is a special area of science and engineering that helps machines and systems work better. It uses math and engineering ideas to make sure machines do exactly what we want them to do, without making mistakes or taking too long. For example, it helps robots move smoothly, planes fly safely, and factories run without stopping.

To make this happen, engineers use something called a controller. This controller watches what the machine is doing and compares it to what we want it to do. If there is any difference, the controller makes changes to fix it. This way, the machine stays on the right path and works just as planned.

Control theory began in the 1800s when scientists first studied how to balance machines like steam engines. Over time, more smart people added to this knowledge, making it better and more useful. Today, control theory helps not just machines, but also ideas in money, science, and many other areas where feedback is important.

History

See also: Control engineering § History

Control systems have been around for a very long time. One of the earliest examples is the centrifugal governor, which was used to control the speed of windmills. In 1868, a scientist named James Clerk Maxwell studied how these governors worked and discovered that delays in the system could cause problems, like shaking or instability. His work sparked more interest in understanding how to control systems.

During World War II, control theory became very important for guiding airplanes and other machines. New ideas helped make control systems better and more reliable. Today, control theory is used in many areas, from guiding spacecraft during the Space Race to helping computers make decisions. The goal is always to create systems that can stay stable and work well, even when faced with changes or challenges.

Open-loop and closed-loop (feedback) control

Control theory helps us manage systems so they behave the way we want. Sometimes, we can control a system without checking its status—this is called open-loop control. For example, a timer set to water a garden doesn’t check if the soil is already wet; it just turns on the water at a certain time.

But often, it’s better to watch the system and make changes as needed. This is called closed-loop or feedback control. Imagine driving a car: you look at the road and steer to stay on course. If you drift to the left, you turn the wheel right to correct it. Feedback control lets systems adjust themselves to reach the right goal, even if things change.

Classical control theory

Classical control theory is a way to manage and guide systems that change over time. It helps us create rules so that a system can reach the state we want it to be in. The goal is to make sure the system gets there quickly, without going too far past the target or stopping in the wrong place, and stays stable. This is done by using a special part called a controller, which adjusts the system as needed to keep it on the right path.

Linear and nonlinear control theory

Control theory has two main parts. The first is linear control theory, which deals with systems that follow simple rules where the output matches the input in a steady way. These systems can often be studied using special math tools like the Laplace transform and frequency response.

The second part is nonlinear control theory, which covers more complex real-world systems that don't follow those simple rules. These systems need more advanced math and computer simulations to understand and control them. Sometimes, these systems can be made simpler by using techniques that turn them into linear systems for easier study.

Analysis techniques – frequency domain and time domain

There are two main ways to study and design control systems: the frequency domain and the time domain.

In the frequency domain, we look at how the parts of a system change with frequency. This method turns complicated math problems into simpler ones, but it only works for certain types of systems.

In the time domain, we study how the system changes over time. This method is useful for more complex, real-world systems and can handle situations that the frequency domain cannot. Modern computers make it easier to work with these time-based models.

System interfacing

Control systems can be grouped based on how many inputs and outputs they have.

• Single-input single-output (SISO) – This is the simplest type, where one output is managed by one control signal. Examples include cruise control or an audio power amplifier, where the input audio signal controls the loudspeaker.
• Multiple-input multiple-output (MIMO) – These are used in more complex systems. For instance, large modern telescopes like the Keck and MMT use many segments, each controlled by an actuator. Their shape is adjusted by a MIMO active optics system to correct for changes caused by temperature, movement, and air disturbances. Systems like nuclear reactors and even human cells can be modeled as large MIMO control systems.

Classical SISO system design

Classical control theory mainly focuses on designing systems with one input and one output (SISO). Analysis can be done using differential equations, the Laplace transform, or frequency domain methods. Many systems are treated as having a second-order response. Classical controllers, especially PID controllers, are often chosen for their simpler implementation in industrial applications, though they may need tuning after installation.

Modern MIMO system design

Modern control theory works in the state space and handles systems with many inputs and outputs (MIMO). This approach overcomes some limits of classical theory and is used in complex designs like aircraft control. Modern theory includes nonlinear, multivariable, adaptive, and robust control. Important figures in this field include Rudolf E. Kálmán and Aleksandr Lyapunov.

Topics in control theory

Stability

The stability of a system with no input can be described using special rules. For simple systems, stability means the output stays calm no matter what input is given. For more complex systems, stability has its own rules too.

To be stable, a system's poles—which are special points in its math description—must stay in certain areas. In simple terms, this helps make sure the system doesn’t behave unpredictably or wildly over time.

Controllability and observability

Main articles: Controllability and Observability

Before choosing how to control a system, it’s important to know if we can actually control it and see its inner workings. Controllability means we can guide the system to a desired state using the right signals. Observability means we can figure out the system’s internal state just by watching its outputs.

If parts of the system aren’t controllable or observable, they might not respond to our control efforts. Sometimes, adding more tools like sensors can help solve these problems.

Control specification

There are many ways to control a system, from general methods to ones made for specific uses like robots or airplanes. No matter the method, the system must always stay stable. Sometimes we also want the system to behave in a certain way, like moving smoothly or quickly.

System identification

To control a system, we first need to understand how it works. This is called system identification. We can do this by testing the system and using the results to build a math model. Even with a good model, real-world systems can change, so some advanced methods update the model while the system is running to keep control accurate.

Analysis

We can study how strong a control system is by looking at its behavior over time and using special graphs. For simpler systems, we check things like how much the system can be pushed before it breaks. For more complex systems, we use different control methods that include these strengths.

Constraints

Real systems have limits. Sometimes a controller might try to do something impossible, like moving too fast. Special control methods help make sure the system stays safe and works well within these limits.

System classifications

Linear systems control

Main article: State space (controls)

For complex systems with many inputs and outputs, we can use math to place important points where we want them. This often needs computers and might not always work perfectly, especially when we can't measure every part of the system.

Nonlinear systems control

Main article: Nonlinear control

In fields like robotics and the aerospace industry, systems often behave in complex ways. Sometimes we can simplify these systems to use easier methods, but other times we need new ideas just for them. These new ideas can include special ways to control the system using knowledge from math and theories about stability.

Decentralized systems control

Main article: Distributed control system

When many controllers work together to manage a system, it is called decentralized control. This helps systems cover larger areas. The controllers talk to each other through messages to work as a team.

Deterministic and stochastic systems control

Main article: Stochastic control

Some control problems have random changes from outside the system, which we call stochastic. Other problems do not have these outside changes and are called deterministic.

Main control strategies

Every control system needs to stay stable. For simple systems, this can be done by placing special points in just the right way. More complex systems use theories to keep things stable, no matter what happens inside them.

There are many ways to control systems. One way tries to use the least amount of fuel or energy. Another way works even if the system isn’t exactly as expected. Some control methods deal with random changes or use smart computers to adapt to new situations. Others use networks of devices to manage bigger, more complicated systems.

People in systems and control

Main article: People in systems and control

Many people, both past and present, have helped shape the field of control theory:

• Pierre-Simon Laplace created the Z-transform in his work on probability theory, which helps solve certain control problems. The Z-transform is related to another math tool named after him, the Laplace transform.
• Irmgard Flugge-Lotz worked on special control methods and used them in automatic aircraft control systems.
• Alexander Lyapunov started the study of stability in the 1890s.
• Harold S. Black came up with the idea of negative feedback amplifiers in 1927 and built stable versions in the 1930s.
• Harry Nyquist created a way to check if feedback systems are stable in the 1930s.
• Richard Bellman developed a method called dynamic programming in the 1940s.
• Warren E. Dixon was a control expert and teacher.
• Andrey Kolmogorov and Norbert Wiener worked together on a filtering method in 1941.
• John R. Ragazzini brought digital control and the Z-transform into control theory in the 1950s.
• Lev Pontryagin introduced important ideas about maximum and bang-bang control.
• Pierre-Louis Lions expanded methods for studying best control strategies.
• Rudolf E. Kálmán led the way in a new way to look at systems and control, and developed the Kalman filter for making good estimates.
• Ali H. Nayfeh contributed greatly to control of nonlinear systems.
• Jan C. Willems introduced new concepts that helped in studying complex systems.