Scribd
Upload
49 views

Digital Control Part 2

The document discusses mapping poles between the s-plane and z-plane when designing digital controllers. It provides examples of how poles in different regions of the s-plane map to locations in the z-plane, and how this affects system stability and response. An example is then given of using state feedback to design a digital controller for a servomotor system by first placing poles in the z-plane to meet a specified settling time, and then calculating the corresponding state feedback gain. Simulation results demonstrate the improved closed-loop response.

Uploaded by

Zhraa Nor
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
49 views

Digital Control Part 2

The document discusses mapping poles between the s-plane and z-plane when designing digital controllers. It provides examples of how poles in different regions of the s-plane map to locations in the z-plane, and how this affects system stability and response. An example is then given of using state feedback to design a digital controller for a servomotor system by first placing poles in the z-plane to meet a specified settling time, and then calculating the corresponding state feedback gain. Simulation results demonstrate the improved closed-loop response.

Uploaded by

Zhraa Nor
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 18

Digital

Control: Part 2

ENGI 7825: Control Systems II


Andrew Vardy
Mapping the s-plane onto the z-plane
• We’re almost ready to design a controller for a DT
system, however we will have to consider where
we would like to position the poles
• We generally understand how to position
desirable poles in the s-plane
– Although this does remains somewhat of a “black art”
as there are various arbitrary choices and rules-of-
thumb at play
• If we understand how to position poles in the z-
plane we can do direct digital design.
Alternatively, we can position poles in the s-plane
and then find out where they lie in the z-plane.
• We have already seen that poles in the s-plane and z-
plane are related by

• We’ll consider particular mappings from parts of the


s-plane. We have already seen that the j! axis
corresponds to the unit circle in the z-plane. In the
following, s = σ + jω and ω = 0.

• Fundamentally, there is a limitation on the signal


frequency that can be represented by the z-
transform. That limit is ω = ωs / 2 where ωs = 2π / T.
• That portion of the j! axis which lies in the range [-jωs/2, jωs/2] maps onto
the unit circle.
• So poles on the unit circle in the z-plane correspond to pure sinusoids and
therefore signify a marginally stable system.
• As we have already seen, poles inside the unit circle correspond to
exponentially decaying sinusoids. If all poles lie within the unit circle then
we have asymptotic stability. Poles outside the unit circle correspond to
exponentially growing sinusoids, and therefore instability.
For s = σ + jω if σ is held constant (lets say we set it to a value of σ1) and ω is
allowed to vary we get

This corresponds to vertical lines in the s-plane and circles in the z-plane
(including the unit circle).
What if we do the opposite? That is, for s = σ + jω we hold ω constant (at ω1)
if σ is allowed to vary allowed to vary we get

This corresponds to horizontal lines in the s-plane and rays emanating from
the origin in the z-plane.
Lets consider pairs of poles located at s = σ ± jω. We know that such a pole pair
corresponds to a term of the form ke σt cos(ωt + ψ). We can also define this pair of
poles in polar coordinates as (r, ±θ) as below:

In particular we would like to position the poles of a second-order system which have
the following locations:

Now translate to the z-plane:

We can then solve for the relationship between (r, ±θ) and (ζ, ωn):
The relationships
566 Chapter 8 Digitalbetween
Control z-plane pole locations and (ζ, ωn ) is somewhat complex,
geometrically:
Im(z)

1.2

7=e Ts s=-Yw +J·wVl - y2


1.0 11 - n S
T = Sampling period

0.8 1---------lr------""7I\<-

0.6 I- - - ---,l'-- -

0.4 .....:;k,

0.2

7r
wn = T
__
- LO -0.8 - 0.6 -0.4 - 0.2 o 0.2 0.8 LO Re(z)
Figure 8.4
Natural frequency (solid color) and damping loci (light color) in the z-plane; the portion below the Re(z)-axis (not
shown) is the mirror image of the upper half shown
These relationships between the locations of a pole pair at (r, ±θ) in the z-plane and
second order system parameters (ζ, ωn) allow us then to relate pole locations to “boss
parameters” such as %OS and settling time.

Example:

We have a DT system with the following closed-loop characteristic polynomial:

Get the pole locations in the z-plane in terms of (r, ±θ) then obtain the 2nd order
parameters (in this example T = 1s which is rather slow):
The examples below illustrate 4 different configurations of s-plane and corresponding z-
plane pole locations and the resulting signals produced.
The following plot from Franklin gives a similar picture:
8.2 Dynamic Analysis of Discrete Systems 567
[m( z)

/
a
/-1 / Re(z)

Figure 8.5
Time sequences associated with poi nts in the z-plane
Digital State Feedback Design
• State feedback can be applied to sampled data
systems in almost exactly the same way as for
CT systems
– The only real difference is that we place
eigenvalues in the z-plane, not the s-plane
• We proceed by example. Assume we have the
following servomotor system (again):
Zero- 0
Order
Hold
In the previous set of notes we developed the following discretized state-space
model for this system:

x1(k) represents the angle of the motor shaft (measureable by encoder count).
x2(k) represents the shaft speed (measureable by a tachometer, rate gyro, or
by rate of encoder counts).

It is important to consider whether the state variables are measureable


because otherwise full-state feedback cannot be applied.
The state feedback control law (7.2) features a constant state feedback
gain matrix K of dimension m × n and a new external reference input r(t)
necessarily having the same dimension m × 1 as the open-loop input u(t),
as well as the same physical units. Later in this chapter we will modify
Here is our usual picture of a state feedback controller:

x0

r(t) + u(t) + •
x(t) x(t) + + y(t)
B C
− +

FIGURE 7.1 Closed-loop system block diagram.


This example differs in that it has been discretized, but also in that the goal is to set the
motor’s shaft angle to zero. That makes this controller a regulator. A regulator is a
controller or compensator that works to move one or all state variables to zero. So we
can say there is no r(t), or equivalently that r(t) = 0.

In regulator design (for n=2) the input to the plant is defined as


Problem specification: Reduce settling time to 4 seconds. (Nothing else is mentioned
which means we don’t particularly care about other specifications such as %OS).

Start by looking at the open-loop system and its characteristics. We will need the
current characteristic polynomial (computed as usual except that we use |zI – A|
instead of |sI – A|).

The design process that follows goes from a unity feedback system (which is identical
to state feedback with K_1 = 1, K_2 = 0). That unity feedback system has the following
characteristic polynomial:

The eigenvalues of the unity feedback system can be obtained from the quadratic
formula then converted to polar form:
Work out the second-order parameters:

= 1.0246

Current settling time:

Since we don’t care about %OS lets just change ωn. To bring the desired settling time
down to 4 seconds we modify ωn and then get the desired pole locations:
Now we can get the desired characteristic polynomial:

We continue to design the K gain vector in the usual way. The system is not in CCF so
we use Bass-Gura and obtain K = [0.445 0.113].
The following shows the resulting improvement in system response (x(0) = [1 0]T).

You might also like