ADVANCED CONTROL SYSTEM DESIGN
11. Discrete State Space Design for
Digital Control Systems
Theory Contents:
11.1 Introduction
11.2 Discrete State Space Equations from Transfer Functions
11.3 Selection of Sample Interval and Closed Loop Eigenvalues
11.4 Integral Control and Reduced Order Observers
11.4.1 Integral Control
11.4.2 Reduced Order Observers
11.5 Examples
Objective:
Learn how to apply all the continuous time techniques to the design of digital
control systems.
1
ADVANCED CONTROL SYSTEM DESIGN
11.1. Introduction
The Discrete State Space Equation defines the response of the system at
evenly spaced time steps, keeping u constant between two steps (i.e. put
through a zero order hold).
Therefore the matrices E and F correspond to the zero order hold equivalent
of the continuous system.
If we want to obtain the input-output relationship, i.e. the z-domain transfer
function, we just have to take z-transforms of the following equations:
2
ADVANCED CONTROL SYSTEM DESIGN
11.1. Introduction
3
ADVANCED CONTROL SYSTEM DESIGN
11.1. Introduction
This is the z-domain transfer function corresponding to the system:
We can see that the characteristic polynomial of matrix E is the denominator
of the z-domain transfer function. Therefore:
The Eigenvalues of matrix E are the poles of the z-domain transfer function.
This is a very nice property because:
- Feedback design for discrete systems is identical to feedback design for
continuous systems. I.e. the closed loop poles will be the eigenvalues of:
- All the theory developed for continuous systems is applicable to discrete
systems. 4
ADVANCED CONTROL SYSTEM DESIGN
11.1. Introduction
Techniques identical in continuous and discrete state space design:
- Transformation of state variables.
- Control canonical form, observer canonical form, modal form.
- Controllability and Observability.
- Feedback design.
- Observer design.
Therefore, we ALREADY know how to deal with discrete state space systems.
Actually, there are some little things to know:
- How to obtain the discrete state space equations from transfer functions.
- Selection of the sample interval.
- How to implement the term “1/s” that appears in integral control and
reduced order observer design.
5
ADVANCED CONTROL SYSTEM DESIGN
11.2. Discrete State Space Equations
from Transfer Functions
6
ADVANCED CONTROL SYSTEM DESIGN
11.3. Selection of the Sample Interval
and Closed Loop Eigenvalues
The sampling rate
Should be chosen to be between 5 to 10 times the CLOSED LOOP NATURAL
FREQUENCY:
If we have an observer, the closed loop poles of the observer will be 10 times
faster than the closed loop natural frequency of the control loop (wo).
Therefore, the OBSERVER SAMPLING INTERVAL (TO) will be:
If observer and controller have different sampling rates, the eigenvalues of the
observer can be chosen to be exactly the same as those of the plant.
The observer will still be much faster than the main loop, since T0=T/10
7
ADVANCED CONTROL SYSTEM DESIGN
11.3. Selection of the Sample Interval
and Closed Loop Eigenvalues
8
ADVANCED CONTROL SYSTEM DESIGN
11.3. Selection of the Sample Interval
and Closed Loop Eigenvalues
9
ADVANCED CONTROL SYSTEM DESIGN
11.4. Integral Control and Reduced
Order Observers
In continuous time design there are two cases in which we need to implement
the term “1/s”. One of them is integral control and the second case is
reduced order observer design.
The implementation of the “1/s” term is slightly different in the discrete time
domain, as we will see.
CASE 1: INTEGRAL CONTROL
r(k)
1/(z-1)
10
ADVANCED CONTROL SYSTEM DESIGN
11.4.1 Integral Control
INTEGRAL CONTROL
11
ADVANCED CONTROL SYSTEM DESIGN
11.4.2 Reduced Order Observer
CASE 2: REDUCED ORDER OBSERVER
12
ADVANCED CONTROL SYSTEM DESIGN
11.4.2 Reduced Order Observer
REDUCED ORDER OBSERVER
This difference equation, together with eq.
define the reduced order observer
13
ADVANCED CONTROL SYSTEM DESIGN
11.4.2 Reduced Order Observer
14
