Skip to content
Tutorials Center
My learning

Topics

Lesson 1 of 37 ยท Getting Started

What Is a Control System?

Learn what a control system is, its parts (plant, sensor, controller, actuator), everyday examples, and how to simulate one in Python with NumPy.

A control system is anything that makes another system behave the way you want. A thermostat keeps a room at 21 degrees. Cruise control holds a car at 100 km/h. An autopilot keeps a plane level, and a drone's flight controller keeps it hovering in a gust of wind. In each case a device measures what is happening, compares it with what should be happening, and acts to close the gap.

Control theory is the mathematics behind that idea. It tells you how to model a system, predict how it will move, decide whether it can be made stable, and design a controller that meets a specification such as "reach the target in two seconds with no more than 10% overshoot". This track teaches it step by step, with small programs you can run and change.

What runs on this site. Every example here is plain Python (NumPy and SciPy) running in your browser through Pyodide. We simulate plants and controllers in software; there is no real motor, heater or drone attached. Where we use a library function we also show the idea written from scratch, so you can see what the library does.

The vocabulary

Every control problem has the same cast of characters.

  • Plant (process): the thing being controlled, such as a motor, a tank of water, or a room.
  • Output y: the quantity you care about, such as speed, level, or temperature.
  • Input u: what you can change to influence the output, such as motor voltage, valve opening, or heater power.
  • Reference (set point) r: the value you want the output to reach.
  • Disturbance d: things you cannot choose that push the output around, such as wind, load, or an open window.
  • Sensor: measures the output and reports it.
  • Controller: decides the input from the reference and the measurement.
  • Actuator: the hardware that turns the controller's decision into a physical action.

The parts of a control loop

Reference What you want the output to be.
Controller Decides what to do next.
Plant The system being controlled.
Sensor Measures the real output.

The controller compares the measurement with the reference.

1 / 2

Everyday examples

System Plant Output Input Disturbance
Home heating Room Temperature Heater power Open window
Cruise control Car Speed Throttle Hills, wind
Water tank Tank Water level Inlet valve Outflow demand
Robot arm Arm and motor Joint angle Motor voltage Payload weight
Insulin pump Body Blood glucose Insulin dose Meals, exercise

The same mathematics covers all of them. A temperature that creeps up slowly and a robot joint that swings quickly are both described by differential equations, and the controller design tools in later lessons work on both.

Why you need a model

You cannot design a good controller by guessing. A model is a set of equations that predicts how the output responds to the input. With one you can simulate the plant, test a controller safely on a computer, and use theory to prove properties such as stability. Chapter 2 builds models from physics; here we use the simplest useful one, a first-order plant:

Text
dy/dt = (-y + K u) / T

K is the gain (how far the output moves per unit of input in the long run) and T is the time constant (how quickly it gets there). A heated cup of water or a charging capacitor behaves like this.

Python Simulate a first-order plant

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

With a constant input of 1 the output rises toward K * u = 2. After one time constant (5 s) it has covered about 63% of the way, and after about four time constants it is within 2% of the final value. You will meet those numbers again in time-domain analysis.

What control is trying to achieve

Designers usually ask for some combination of these goals.

  1. Stability: the output settles instead of growing without bound. This is the minimum requirement, treated fully in Stability and Pole Locations.
  2. Tracking: the output follows the reference closely, with small steady-state error.
  3. Disturbance rejection: outside pushes have little effect on the output.
  4. Speed: the output reaches the target quickly.
  5. Smoothness: little overshoot and oscillation, and moderate actuator effort.
  6. Robustness: the design still works when the real plant differs from the model.

These goals pull against each other. A faster response usually means more overshoot and larger actuator effort. Control design is the art of balancing them.

A first look at control

Suppose the plant above is a heater with gain 2, and we want the output to equal a reference of 1. We could simply apply the input u = 1 / K = 0.5. That works if the model is perfect. Try a controller that instead looks at the error and pushes harder when the error is large:

Python Bang-bang control of a heater

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

This simple on/off rule, as in many household thermostats, keeps the output hovering near the reference using only a measurement. That is feedback: the control action depends on the result. The next lesson compares this with control that ignores the result.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. In a cruise-control system, what is the plant?

    The plant is the system being controlled, here the car. The set speed is the reference, the throttle command is the input, and the speed sensor measures the output.
  2. 2. Which signal is a disturbance for a room heater?

    A disturbance is an outside influence you do not choose. An open window removes heat and pushes the temperature away from the target.
  3. 3. A first-order plant has gain 2 and time constant 5 s. With a constant input of 1, what does the output approach?

    The steady-state output is the gain times the input, 2 x 1 = 2. The time constant only sets how fast it gets there.
  4. 4. Why do control engineers build a model of the plant?

    A model lets you simulate and analyse the system, including proving stability, before touching real hardware.

Summary

  • A control system makes a plant's output follow a reference despite disturbances.
  • The main parts are plant, sensor, controller and actuator; the main signals are reference, input, output and disturbance.
  • Models, usually differential equations, let you simulate and analyse before building.
  • Goals such as stability, tracking, speed and robustness conflict, so design is a balance.
  • Feedback means the control action depends on the measured output.

Next: Open Loop and Closed Loop