The feedback-control-for-distributed-systems

World Academy of Science, Engineering and Technology
International Journal of Mathematical, Computational Science and Engineering Vol:1 No:12, 2007

The Feedback Control for Distributed Systems
Kamil Aida-zade, and Cemal Ardil

~
u ( ~ j , t ) = u j (t ), j = 1,..., L , t ∈ [t0 , T ]
x

International Science Index 12, 2007 waset.org/publications/4092

Abstract—We study the problem of synthesis of lumped sources
control for the objects with distributed parameters on the basis of
continuous observation of phase state at given points of object. In the
proposed approach the phase state space (phase space) is beforehand
somehow partitioned at observable points into given subsets (zones).
The synthesizing control actions therewith are taken from the class of
piecewise constant functions. The current values of control actions
are determined by the subset of phase space that contains the
aggregate of current states of object at the observable points (in these
states control actions take constant values). In the paper such
synthesized control actions are called zone control actions. A
technique to obtain optimal values of zone control actions with the
use of smooth optimization methods is given. With this aim, the
formulas of objective functional gradient in the space of zone control
actions are obtained.

The problem is to determine such control function
with
respect
to
current
values
of

ν (t ) ∈ U

~
u j (t ), j = 1,..., L that minimizes the functional:
T

l

[

]

J (v ) = α ∫ v (t )dt + ∫ u ( x, T ) − u * ( x ) dx
2

0

2

(3)

0

Actually, we have to find the mean value of functional (3)
for all possible initial and boundary conditions, i.e., the
functional to be minimized are:
T

optimization methods, lumped control synthesis.

J (v ) = α ∫ v 2 (t )dt +

I. THE FORMULATION OF

+

Keywords—Feedback control, distributed systems, smooth

0

∫( ) ∫( ) ∫ α (x )α (x )[u(x, T ;ψ
l

T

THE PROBLEM OF LUMPED CONTROL
SYNTHESIS

O illustrate the proposed approach to the study of the
problem of control synthesis in distributed systems, we
consider the problem of control of heat-exchanger process.
Let us describe briefly the process. A liquid goes through the
heat-exchanger. The temperature of the liquid at the entry
point of heat-exchanger is ψ 2 (t ) . The temperature of steam
in the jacket isν (t ) . The function
in the following problem:

ν (t )

is control parameter

(1)

u(x,0) =ψ1 (x) ∈Ψ1 (x),

(2)

is the temperature of the liquid,
Ψ1 ( x ), Ψ2 (t ) are given domains of initial and boundary
conditions.
Here

2

u(x,t)

Let us suppose that there are L sensors mounted at the heatexchanger at the points ~ j ∈ [0, l ], j = 1,..., L . The
x
sensors perform on-line monitoring and reading-in of
information about the temperature of liquid at these points into
the process control system. This information is defined by the
vector:

]

,ψ 2 , v ) − u * ( x ) dxdψ 2 dψ 1
2

1

Ψ1 x Ψ2 x 0

(4)
here u ( x, t;ψ 1 ,ψ 2 , v ) is the solution of (1), (2) under
specified
admissible
initial-boundary
functions
ψ 1 ( x ), ψ 2 (t ); the functions α 1 ( x ), α 2 (t ) are weight
functions. In the case of uniform admissible initial-boundary
conditions weight functions are determined as follows:

α 1 ( x ) = 1 / mesΨ1 ( x ),

ut (x, t) = −a1ux (x, t) − a0u(x, t) + av(t), x ∈[0, l], t ∈[0,T ]

u(0, t) =ψ2 (t) ∈Ψ2 (t).

1

α 2 (t ) = 1 / mesΨ2 (t ).

here u ( x, t;ψ 1 ,ψ 2 , v ) is the solution of (1), (2) under
specified
admissible
initial-boundary
functions
ψ 1 ( x ), ψ 2 (t ); the functions α 1 ( x ), α 2 (t ) are weight
functions. In the case of uniform admissible initial-boundary
conditions weight functions are determined as follows:

α 1 ( x ) = 1 / mesΨ1 ( x ),

α 2 (t ) = 1 / mesΨ2 (t ).

The condition (2) has the following meaning. During
operation of heat-exchanger, current control must not
"remember" exact initial-boundary conditions; conversely, the
control must be tuned at their "mean" values and must depend
only on the states at the observable points:

~
~
v(t ) = v(t ; u1 (t ),..., u L (t )) .

Kamil Aida-zade is with the Institute of Cybernetics NAS of Azerbaijan
Republic, Baku, Azerbaijan.
Cemal Ardil is with the National Academy of Aviation, Baku, Azerbaijan.

336

(5)


ν = (ν {,...,ν i ...i ...,ν { )
1...1
m ... m
{

This formulation resembles the classical one, but here we
consider a more general case when v (t ) can not change its
value at any time instance, instead, v(t ) remains constant till
the values of all observable states remain in some definite
zone. We suppose here that the set of all admissible values of
phase variable is partitioned into zones, i.e.:

1

L

(10)

L

L

L

which directly determines the course of heat-exchanger
process and, consequently, the value of objective functional
(11).
II. SOLVING PROBLEM OF ZONE CONTROL SYNTHESIS

u ≤ u ( x, t ;ψ 1 ,ψ 2 , v ) ≤ u , ∀(ψ 1 ,ψ 2 , v(t )) ,

(6)

here u, u are usually known values, determined by
operational consideration. Then it holds for observable states
as well:

~
u ≤ u j (t ) ∈ u ,

j = 1,..., L.

ν (q+1) (αq ) = P (ν (q) − αq gradJ(ν q )), q = 0, 1, ...,
U

(7)

Let us partition the space of observable states into intervals

)
)
)
) )
ˆ
ˆ
ˆ
[u i −1 , u i ) by the values: u = u 0 < u1 < ... < u m = u .

These values determine the zones of state values at the
observable points. So we have:

For numerical solving zone control synthesis problem, we
can use the methods of finite-dimensional smooth
optimization, in particular, iterative method of gradient
projection type:

)
) )
)
)
)
~
~
~
ui1−1 ≤ u1 (t) < ui1 , ui2 −1 ≤ u2 (t) < ui2 ,...,uiL −1 ≤ uL (t) < uiL .(8)

(11)

Let us present the obtained formulas of functional gradient
in the space of optimized parameters.
One of important elements in calculating gradient is the
~
time interval when phase state u (t ) belongs to the one or
another phase parallelepiped. Let us denote by
Π i1 ,...,iL (ψ 1 ,ψ 2 , v ) ∈ [0, T ] the time period during which (8)

i1 = 1, m, ..., iL = 1, m,
i.e.
is = 1,..., m, s = 1,..., L (the dependence of Π i1 ,...,i L on

holds,

L - dimensional parallelepipeds in
~
L - dimensional phase state space u j (t ) j = 1,...,L . The
The sets (8) constitute

L

total number of these parallelepipeds is m .
It is clear that feedback with the object (i.e. sensing) is
required to determine whether the state at the observable
points falls into the ( i1 ,..., i L )-th zone. The aggregate of all
zones covers all possible values of observable states. It is also
clear that control functions (5) change its values only at those
time instants when the set of states at the observable points
passes from one phase parallelepiped (8) to another one.
The problem consists in determination of constant value of
v(t ) for each zone while the states of all observable points
remain in this zone, i.e. the following holds:

~
~
v(t) = v(u1 (t),...,uL (t )) = vi1 ,...,iL = const, t ∈[0, T ]

(9)

ψ 1 ,ψ 2 , v

is evident).
In classical formulation the adjoint system with respect to
Ρ( x, t ) = Ρ( x, t;ψ 1 ,ψ 2 , v ) under fixed initial-boundary
conditions and fixed control is as follows:

Ρt' = −a1 Px' + a 0 P

(12)
∗

Ρ( x, T ) = 2(u( x, T ) − u ( x ))
Ρ( l , t ) = 0 .

(13)
(14)

The following theorem holds.

Theorem 1. The components of functional gradient in the
problem (1), (2), (3) in the space of piecewise constant control
actions (9) for arbitrary control ν ∈ U under appropriate
normalization are defined by the formula:

while (8) holds.
l

dJ
= a0 ∫ ∫ ∫
dvi1 ,...,iL
ψ 1ψ 2 0

The number of different values attained by steam
temperature equals to the number of phase parallelepipeds

m L . The total number of
L
parameters to be optimized is also m . They determine the
defined by inequalities (8), i.e.,

value of steam temperature for all possible values of liquid
temperature at the observable points.
So, the considered problem of heat-exchanger process
control with the use of feedback at the class of piecewise
constant functions
dimensional vector:

consists

in

optimization

of

mL -

∫ P( x, t )dtdxdψ

2

dψ 1

(15)

Π i1 ,..., i L

here P( x, t ) is the solution of adjoint problem (12) – (14),
that corresponds to current zone control.
The formula (15) allows us to compute the functional
gradient at current value of control actions and to use it further
in iterative procedures of smooth optimization, for example,
(11).
Let us emphasize important advantage of the proposed in
the paper synthesized zone control as compared to synthesized

337


classical control of

ν (u( x, t ))

type. It is caused by technical

complexity of on-line acquirement of information about
current states of objects at the observable points and
construction of control actions on the basis of this
information. The construction of zone control is performed
during the time when the states of object lie in the definite
zones of phase space. This argues for robust features of zone
control.
Also it is of practical interest to apply the proposed
approach to constructing control actions from the results of
observation of object current state values only at certain points
of object. In many cases, this corresponds to actual capacities
of measurement systems at objects.
And, finally, it is clear, that one can easy extend the study
of illustrative problem of heat-exchanger control to a number
of other problems of control of distributed objects, governed
by functional equations of another type.


III. NUMERICAL EXPERIMENTS
The proposed method of the considered problem solving
was applied to some test problems. When executing numerical
experiments, poor convergence to the solution of the problem
(1-3) has been revealed. Probably, this fact is linked with the
possibility of abrupt variation of ν (t ) when trajectory transits
from one phase parallelepiped to another if the values of
piece-wise constant control actions assigned to these
parallelepipeds are essentially different. To overcome this, the
calculations were performed in two stages. At the first stage
we used a control function the value of which for a given
fixed trajectory is equal to a weighted linear combination of
all values of control actions for all phase parallelepipeds. The
weighted coefficients for the control ν i1 ... iL were inversely
proportional to the distance from the current trajectory to the
i1 ... iL -th phase parallelepipeds, i.e.:
m

m

i1 =1

i L =1

ν sm (t ) = ∑ ...∑ λi ... i ν i ... i
λi ...i = e
1

L

−

1

ρ i2 ... iL
1
2σ 2

, here

L

1

ρ i ...i
1

L

L

,

(16)

is the distance from trajectory

u(x,t) to the i1,…, iL –th phase parallelepiped. Control in the
form (16) we will call "smoothed control".
Numerical experiments were carried out for problems with
different terminal function u*(x).
REFERENCES
[1]
[2]
[3]
[4]

Aida-zade K.R. The problem of control in the mean with respect to
regional control actions. Transactions of the National Academy of
Sciences of Azerbaijan, №4, 2003.
Systems and Control Encyclopedia. Ed. M.G.Singh. V.1-8. Pergamon
Press, 1987.
The Control Handbook. Ed. W.S. Levine. CDC. Press. IEEE Press,
1996.
W. Harmon Ray. Advanced Process Control. McGraw-Hill, 1981.

338

The feedback-control-for-distributed-systems

More Related Content

What's hot (20)

Viewers also liked (8)

Similar to The feedback-control-for-distributed-systems (20)

More from Cemal Ardil (20)

Recently uploaded (20)

The feedback-control-for-distributed-systems