Scribd
Upload
76 views

MIT16 06F12 Lecture 10

This document describes the three main components of PID control - proportional, integral, and derivative. It uses an example transfer function to demonstrate how each component affects the closed-loop pole locations and system response. Proportional control increases the natural frequency but reduces damping. Derivative control increases damping without affecting natural frequency. Integral control is needed to eliminate steady-state error for systems with integrators in their transfer functions. The document walks through designing a PID controller to meet specifications for rise time, overshoot, and zero steady-state error for a given plant transfer function. Mirror loop feedback of the derivative term is recommended to avoid additional overshoot.

Uploaded by

Pello Errasti Irureta
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
76 views

MIT16 06F12 Lecture 10

This document describes the three main components of PID control - proportional, integral, and derivative. It uses an example transfer function to demonstrate how each component affects the closed-loop pole locations and system response. Proportional control increases the natural frequency but reduces damping. Derivative control increases damping without affecting natural frequency. Integral control is needed to eliminate steady-state error for systems with integrators in their transfer functions. The document walks through designing a PID controller to meet specifications for rise time, overshoot, and zero steady-state error for a given plant transfer function. Mirror loop feedback of the derivative term is recommended to avoid additional overshoot.

Uploaded by

Pello Errasti Irureta
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

16.

06 Principles of Automatic Control

Lecture 10

PID Control
A common way to design a control system is to use PID control.
PID = proportional-integral-derivative
Will consider each in turn, using an example transfer function
Gpsq

s2

A
` a1 s ` a2

Proportional (P) control


In proportional control, the control aw is simply a gain, to that u is proportional to e:

kp

G(s)

u kp e
For our example, the characteristic equation is

0 1 ` kp Gpsq
kp A
1 ` 2
s ` a1 s ` a2
2
0 s ` a1 s ` a2 ` kp A
1

The resulting natural frequency is

a
a2 ` kp A

So in the example, increasing kp increases the natural frequency, but reduces the damping

ratio.

Plot of pole location vs kp :

Im(s)

open loop
pole location

increasing kp
-a1/2

Re(s)
closed-loop
pole location

Derivative (D) control


To add damping to a system, it is often useful to add a derivative term to the control,

uptq kp eptq ` kD e9ptq


or
vpsq kp Epsq ` kD sEpsq
pkp ` kD sqEpsq
KpsqEpsq
What is the characteristic equation?

0 1 ` KpsqGpsq
pkp ` kD sqA
1 ` 2
s ` a1 s ` a2
2
0 s ` pa1 ` kD Aqs ` pa2 ` kp q
2

So increasing kD increases the damping ratio without changing the natural frequency, for

this example.

For kp xed, kD varying, plot of closed-loop pole location is:

Im(s)

pole position for kD = 0


pole position for kD > 0

Re(s)

increasing kD
NB: For other Gpsq, results may vary.

Sometimes, its better to place derivative feedback in the feedback path:

kp

G(s)

kD(s)

Why? We get the same pole locations, but no additional zeros to cause additional overshoot.
Another way to think about this is that we want the derivative eect on y, because that
adds damping, but we dont want to dierentiate the reference.

Integral (I) control


Especially if the plant is a type 0 system, we may want to add integrator to controller to
drive steady-state error to zero:
V psq pkp ` ksI

P
I
3

`kD sq Epsq

Example:
Gpsq

s2

1
`s`1

Suppose we want a system that


1. Has rise time above tr 1s
2. Has peak overshoot of Mp 0.05
3. Has zero steady-state error to step command
Lets do one piece at a time:

G(s)

kp

Characteristic equation is
0 s2 ` s ` 1 ` kp
So can only change n (and indirectly, ) with kp . for tr 1, need
1

1.8
n 1.8
n

So lets take kp 2 for simplicity. Then


3
kp G
2
s `s`4
1 ` kp G
0.25, Low
T

To get Mp 5%, need 0.7. So add derivative control. Characteristic Equation is


0 s2 ` p1 ` kp qs ` 1 ` kp
The desired polynomial is
0 s2 ` 2.8s ` 4
So take kD 1.8.
4

If PD control is in forward loop,


T

s2

1.8s ` 3
` 2.8s ` 4

and the peak overshoot will be 16%, not 5%. So instead, use control structure

G(s)

(*)

1.8s

pq mirror loop feedback


With this structure, we have:

tr 1.06s
Mp 4.6%
ess 0.25
So lets add integral control:

kv+kI/s

G(s)
kDs

Take kI 0.25 (trust me!)


Then
T

s3

3s ` 0.25
` 2.8s2 ` 4s ` 0.25

Response sort of meets specs:


5

0.8

0.6

0.4

5%

0.2

0
0

10

20
t

30

The response has a long tail, due to slow pole poles are at:

s 1.37 1.40j
s 0.065
slow pole causes long tail

40

MIT OpenCourseWare
http://ocw.mit.edu

16.06 Principles of Automatic Control


Fall 2012

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.

You might also like