II. TWO DEGREE-OF-FREEDOM (2 DOF) CONTROL
Consider a single-input-single-output (SISO), linear timeinvariant (LTI) plant defined by a rational transfer function (1).
( ) ( ) ( ) ⁄ ( ) ( ) ⁄ (1)
The transfer function is strictly proper, and is both controllable
and observable. The nominal plant polynomials ( ) and ( )
are coprime with ( ) ( ).
The basic controller design objectives are as follows:
1. To ensure an acceptable response to some standard
inputs (model matching problem). To achieve this,
it is obviously necessary that the poles of the
compensated system must be in the in left half
978-1-4673-4569-9/13/$31.00 ©2013 IEEE 493
plane (LHP). This is known as stabilization
problem. It would, however, be better if the poles
could be placed at desired locations (poleplacement problem).
2. To ensure that the system remains stable not only
for the nominal plant but also in the face of plant
uncertainties (robustness problem).
3. To reject external disturbances/noise (disturbance
rejection/ noise attenuation problem).
To achieve these, the most commonly used controller is the
conventional error-driven controller (also known as the servo
controller) which is as shown in Fig. 1.
This controller can achieve pole-placement in the sense that
all the closed-loop poles given by the characteristic equation
(2) can be arbitrarily placed by suitably choosing the controller
polynomials ( ) and ( ).
( ) ( ) ( ) ( ) ( ) (2)
The corresponding loop and input-output transfer functions
then become fixed, and are given by, respectively, (3) and (4).
( )
( ) ( )
( ) ( )
(3)
( )
( )
( ) ( )
( )
(4)
It is to be noted that pole-placement causes some additional
zeros to appear in the input-output transfer function which may
be undesirable from response point of view. More important,
however, is the fact that the additional poles and zeros
appearing in the loop transfer function ( ) may render the
closed-loop system unacceptable from robustness point of
view, especially when the plant has RHP poles and zeros.
Further, a desire to achieve a fast transient response may
compromise the loop performance of a system or cause plant
input saturation. In view of these, it would be useful to have a
design procedure that maintains a desired response
performance while varying the loop performance. Such a
design is possible with a 2 DOF compensator [10] as shown in
Fig. 2.
Equation (5) gives a desired closed-loop transfer function
with α an arbitrary scalar and
̂
( )
( )
( )
( )
( )
̂ ( )
(5)
Now determine the controller polynomials (6) and (7)
( ) (6)