Control Engineering
Lecture 13
MIMO systems
Decentralized control
Consider a two-input, two-output plant
having the transfer function
Decentralized control
Let us say that k12 and k21 depend on the
operating point (a common situation, in practice).
Operating point 1 (k12 = k21 = 0)
Clearly, there is no interaction at this operating
point.
Thus, we can safely design two SISO controllers.
To be specific, say we aim for the following
complementary sensitivities:
Decentralized control
The corresponding controller transfer
functions are C1(s) and C2(s), where
The two independent loops perform as
predicted.
Decentralized control
Operating point 2 (k12 = k21 = 0.1)
With the previously designed controller we
apply a unit step in the reference for output 1
at t = 1 and a unit step in the reference for
output 2 at t = 10. The closed-loop response
is shown on the next slide. These results
would probably be considered very
acceptable, even though the effects of
coupling are now evident in the response.
Effects of weak interaction in control loop
with SISO design
Decentralized control
Operating point 3 (k12 = -1, k21 = 0.5)
With the same controllers and for the same
test as used at operating point 2, we obtain
the results on the next slide.
We see that a change in the reference in
one loop now affects the output in the other
loop significantly.
Effects of strong interaction in control
loops with SISO design
Decentralized control
Operating point 4 (k12 = -2, k21 = -1)
Now a simulation with the same reference
signals indicates that the whole system
becomes unstable. We see that the original
SISO design has become unacceptable at
this final operating point.
Pairing of Inputs and Outputs
If one is to use a decentralized architecture,
then one needs to pair the inputs and
outputs. In the case of an m m plant
transfer function, there are m! possible
pairings. However, physical insight can
often be used to suggest sensible pairings.
Relative Gain Array
One method that can be used to suggest pairings
is a quantity known as the Relative Gain Array
(RGA). For a system with matrix transfer function
Go(s), the RGA is defined as a matrix with the ijth
element
where [Go(0)]ij and [Go-1(0)]ij denote the ijth
element of the plant d.c. gain and the jith element
of the inverse of the d.c. gain matrix respectively.
Relative Gain Array
Note that [Go(0)]ij corresponds to the d.c. gain
from the ith input, ui, to the jth output, yj, while the
rest of the inputs, ul for l {1, 2, , i-1, i+1, , m}
are kept constant. Also [Go-1]ij is the reciprocal of
the d.c. gain from the ith input, ui, to the jth output,
yj, while the rest of the outputs, yl for l {1, 2, ,
j-1, j+1, , m} are kept constant. Thus, the
parameter ij provides an indication of how
sensible it is to pair the ith input with the jth output.
Relative Gain Array
One usually aims to pick pairings such that
the diagonal entries of are large. One
also tries to avoid pairings that result in
negative diagonal entries in .
For a 2 x 2 system RGA is
Relative Gain Array
Consider again the system
The RGA is then
Relative Gain Array
For 1 > k12 > 0, 1 > k21 > 0, the RGA
suggests the pairing (u1, y1), (u2, y2). We
recall from our earlier study of this example
that this pairing worked very well for k12 = k21
= 0.1 and quite acceptably for
k12 = -1,
k21 = 0.5. In the latter case, the RGA is
Relative Gain Array
However, for k12 = -2, k21 = -1 we found that
the centralized controller based on the
pairing (u1, y1), (u2, y2) was actually unstable.
The corresponding RGA in this case is
which indicates that we probably should
have changed to the pairing (u1, y2), (u2, y1).
Example
Quadruple-tank apparatus
Example
Example
This system has an approximate transfer
function
The RGA for this system is
Example
For 1 < 1 + 2 < 2 the system is of minimum
phase. If we take, for example, 1 = 0.7 and
2 = 0.6, then the RGA is
This suggests that we can pair (u1, y1) and
(u2, y2).
Example
Because the system is of minimum phase,
the design of a decentralized controller is
relatively easy in this case. For example,
the following decentralized controller gives
the results shown on the next slide
Decentralized control of a minimum-phase fourtank system
Example
For 0 < 1 + 2 < 1, the system is
nonminimum phase. If we take, for
example 1 = 0.43 and 2 = 0.34, then the
system has a NMP zero at s = 0.0229, and
the relative gain array becomes
Example
This suggests that (y1, y2) should be
commuted for the purposes of decentralized
control. This is physically reasonable, given
the flow patterns produced in this case.
This leads to a new RGA of
Example
Note, however, that control will still be much
harder than in the minimum-phase case.
For example, the following decentralized
controllers give the results shown on the
next slide.
Decentralized control of a nonminimumphase four-tank system
