Control System Lectures
Why Learn Control Theory
Understand systems to
o Switching power supply
o Automatic gain control
o Damping of vibrations in structure
o Tune PID controllers in industry
Open Loop vs Closed Loop Control
Control system: mechanism that alters the future state of a system
Control theory: strategy to select appropriate inputs
Open loop control system: input does not depend on output
Closed loop control system: input depends on output
o Error signal used
o Reference signal encodes desired output
System components
o Input
o Plant
o Output
Transfer Functions
Map from input to output
LTI: linear (homogeneity -> scaling and superposition -> additivity) and time invariant (shifted
input results in shifted output)
LTI allowable operations: scaling, integration, differentiation, addition, and subtraction
No real system meets LTI restrictions, but used because easier to solve
o Can only solve simple non-linear systems, but most of the time linearization used
Impulse function: infinitely tale and thin with area of 1
o Functions can be represented as series of impulse responses with outputs combined
Convolution integral:
o Reverse the input function, with effect from the last output sample in time having the
effect of the first impulse sample. There is also a discrete time equivalent
o
o Can determine the system response to any arbitrary input given the impulse response
o Produces complicated integrals with more than one subsystem
The Frequency Domain and the Fourier Transform
Steady-state behaviour can be represented with frequency-amplitude and frequency-phase
Information is not lost from time domain to frequency domain
Fourier and inverse Fourier map from the time domain to the frequency domain
May be more complicated to plot in the frequency domain, but easier to analyze
Convolution is the multiplication of Fourier transforms
Fourier transform of a shifted function is just 𝑒 −𝑗𝜔𝜏
Transform functions are not completely represented in the frequency domain, but instead in the
s-domain with frequency and exponential growth/decay
The S-domain and the Laplace Transform
1 1 𝜔
The Fourier transform of an exponential decay after t = 0 is 1+𝑗𝜔 = 1+𝜔2 − 𝑗 1+𝜔2
o The transform has both a real and imaginary component
o The components can be converted to give the following
The Laplace transform represents time domain signals with cosines/exponential functions
instead of only cosine with the Fourier transform
The s variable represents both the frequency and the exponential part of the signal
o The first expression is an exponential multiplied by a sinusoid
o 𝑒 𝑠𝑡 = 𝑒 (𝜎+𝑗𝜔)𝑡 = 𝑒 𝜎𝑡 𝑒 𝑗𝜔𝑡
o If sigma there is no exponential decay, then the Laplace transform is equal to the
Fourier transform (if sigma equal to zero is within the region of convergence <-
absolutely integrable)
The s variable is shown on a plane with exponential on the horizontal axis and cosine on the
vertical axis
o The Laplace transform is the Fourier transform of the product 𝑓(𝑡)𝑒 −𝜎𝑡
Stable system: impulse response goes to zero over time
Unstable system: impulse response does not go to zero over time -> system response grows
without bound
Step function -> sigma equal to -1, does no grow and decay with time
The Laplace transform can be represented with 3D phase-magnitude-exponential and phase-
magnitude-exponential plots, but can also be represented with the poles
Another transform decompose signals in sinusoids, exponentials, and square waves
