Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Specific Case of an Internal Control-External Control

(Bounded Domain)
YAMEOGO Pierre Claver
Dr. En Mathématiques(Ph.D)

Abstract:- We try to study the controllability of bounded  The function J is the objective function
domain 𝛀 consisting of several heated materials. For this,
we will use the functional𝐉(𝐮), which optimizes the cost. However, there are two types of controls:
The objective is to bring the excess heat inside the  Internal control,
domain and also on the border of the domain and it’s  Border control
this excess heat that is a function called control.
NB: we advise you to look at the course material on control
I. INTRODUCTION for detailed proof.

Let Ω be a bounded domain made up of heated V. INTERNAL CONTROL

materials, we give ourselves the conductivity k(x) at the
point x ∈ Ω, the heat source f in L2(Ω) , the temperature TD It is a question of bringing the excess heat inside the
imposed on ΓD by the system, the flux PN imposed on TN . domain. Let us consider the following example of control:
Let’s consider the following problem: Let 𝑓 in 𝐿2 (𝛺), 𝑦𝐷 belonging to 𝐿2 (𝛺) 𝑎𝑛𝑑 u be a
nonempty closed convex set of 𝐿²(𝛺). Let 𝑢 ∈ u,𝑦(𝑢) be the
−𝑑𝑖𝜐(𝑘(𝑥)∇𝑦(𝑥)) = 𝑓 𝑖𝑛 Ω,
solution of the following equation.
𝛾𝑦 = 𝑡𝐷 𝑜𝑛 Γ𝐷,
(𝑆)
𝜕𝑦 −Δ𝑦(𝑢) = 𝑓 + 𝑢 𝑖𝑛 Ω,
{ 𝑘𝛾 = 𝑃𝑁 𝑜𝑛 Γ𝑁. (𝑆) { (1)
𝜕𝑛 𝛾𝑦(𝑢) = 0 𝑜𝑛 Γ.

Let’s pose
II. PROBLEM
1 ∈
Let 𝑦𝐷 be a given temperature. J(𝑢) = ∫ (𝑦 − 𝑦𝐷 )𝑑𝑥 + ‖𝑢‖𝐿2 (Ω), (2)
2 Ω 2
How to act on the system (𝑆) so that y is close enough
∈
to 𝑦𝐷? = ‖𝑦 −‖2𝐿2(Ω) + ‖𝑢‖2𝐿2(Ω).
2
III. IDEA
So we get the following problem:
It is a question of bringing a surplus of heat 𝑢 so that 𝑦
𝐷𝑜𝑒𝑠 𝑢̅ 𝑒𝑥𝑖𝑠𝑡 𝑖𝑛 u
is as close as possible to𝑦𝐷. This excess heat is a function { 𝑚𝑖𝑛 (3)
called control. To leave a good choice of 𝑢, we must J(𝑢̅) = 𝑢∈𝑈J(𝑢)
optimize the cost, which reflects what we want to achieve
VI. RESOLUTION
and the means at our disposal. What amounts to considering
the functional J(𝑢) defined by: 𝑓 ∈ 𝐿2 (𝛺), 𝑢 ∈ 𝐿2 (𝛺) ⟹ 𝑓 + 𝑢 ∈ 𝐿²(𝛺),
1 ∈
J(𝑢) = ∫ 𝑔(𝑦 − 𝑦𝐷 )𝑑𝑥 + ‖𝑢‖2𝐿2(Ω); For any 𝑧 in𝐿²(𝛺),we know the problem below:
2 Ω 2
−𝛥𝑧 = 𝑔 𝑖𝑛 𝛺,
Where 𝑔 is a definite function of 𝛺 a value in ℝ. 𝐺(𝑦 − { (4)
𝛾𝑧 = 0 𝑜𝑛 𝛤.
𝑦𝐷 ) makes is possible to minimize the difference between
𝑦 𝑎𝑛𝑑 𝑦𝐷 , ∈ being very small serves not only to prove the Has a unique solution in 𝐻01 (𝛺), therefore 𝑦(𝑢) is the unique
existence and uniqueness of the solution but also to solution of problem (11).
minimize the cost.
We define the following operation:
IV. PROBLEM
𝐴 = 𝐿²(Ω) ⟶ 𝐿²(Ω)
Is there 𝑢̅ ∈u such that:
𝑔 ⟼ 𝐴(𝑔) = 𝑧.
𝑚𝑖𝑛
J(𝑢̅) = 𝑢𝜖uJ(𝑢);
Were 𝑧 is the solution of equation (4). A thus defined is
 u is in the set of admissible controls, linear and continuous.
 𝑢̅ is the optimal control
 𝑦 = 𝑦(𝑢̅) is the optimal state

IJISRT22DEC1510 www.ijisrt.com 2081

Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
We have
We get the following optimality system:
𝐴(𝑓 + 𝑢) = 𝑦(𝑢).
𝑢̅ ∈ 𝑢, 𝑝̅ ∈ 𝐻01 (Ω), 𝑦(𝑢̅) ∈ 𝐿2(Ω)
By replacing 𝑦(𝑢)by its value in(2), we get: −∆𝑦(𝑢) = 𝑓 + 𝑢 𝑖𝑛 Ω,
(𝑆𝑂1) {
1 ∈ 𝛾𝑦(𝑢) = 0 𝑜𝑛 Γ.
J(𝑢) = ‖𝑦(𝑢) − 𝑦𝐷 ‖2𝐿2(Ω) + ‖𝑢‖2𝐿2(Ω), −∆𝑃̅ = 𝑦̅ − 𝑦𝐷 𝑖𝑛 Ω,
2 2 (𝑆𝑂2) {
𝛾𝑝̅ = 0 𝑜𝑛 Γ,
1 ∈ {(𝑆𝑂3) < 𝑃̅ + 𝜖𝑢̅, 𝜐 − 𝑢̅ > 𝐿2(Ω) ≥ 0.
= ‖Α(𝑓 + 𝑢) − 𝑦𝐷 ‖2𝐿2(Ω) + ‖𝑢‖2𝐿2(Ω),
2 2
VII. PARTICULAR CASE
1 1
= {‖ ‖Α(𝑢)‖2𝐿2(Ω) + ∈ ‖𝑢‖2𝐿2(Ω) ‖} + 〈𝐴(𝑓) − 𝑦𝐷 , 𝐴(𝑢)〉, 1
2 2 If 𝑢 = 𝐿²(𝛺)then 𝑢̅ = 𝑝̅ ,
𝜖
J1 (u) + 〈A(f) − yD , A(u)〉. 1
If𝑢 = 𝐿22 (𝛺) 𝑡ℎ𝑒𝑛 𝑢̅ = 𝜖
(𝑝̅ )−. 𝑵𝑩:
Where:
In the case of shareholder internal Control, we have
1 1
J1 (𝑢) = {‖ ‖Α(𝑢)‖2𝐿2(Ω) + ∈ ‖𝑢‖2𝐿2(Ω) ‖}, −∆𝑦(𝑢) = 𝑓 + 𝜓𝑢 𝑖𝑛 Ω,
2 2 {
𝛾𝑦(𝑢) = 0 𝑜𝑛 Γ.
And 〈𝐴(𝑓) − 𝑦𝐷 , 𝐴(𝑢)〉 is a constant.
Where 𝑢 ∈ ℝ, 𝜓 ∈ 𝐿∞ (Ω)
Minimizing the functional J amounts to minimizing the
functional J1 . We put: VIII. BORDER CONTROL

𝑎: 𝐿²(Ω)x 𝐿²(Ω) ⟶ ℝ This is to bring the excess heat to the Boundary of the
domain. Consider
(𝑢, 𝜐) ⟶< 𝐴(𝑢), 𝐴(𝜈) > 𝐿2 (Ω)+< 𝑢, 𝜈 >
The following boundary control example
𝑙: 𝐿2 (Ω) ⟶ ℝ
Let 𝑓 in 𝐿2 (), 𝑦𝐷 belong to 𝐿2 (), 𝑔 a function of
(𝑢, 𝜐) ⟶< 𝐴(𝑢) − 𝑦𝐷 , 𝐴(𝜈) > 𝐿2 (Ω). 𝐿2 (𝑇)
andu nonempty closed convex of 𝐿2 (𝑇). 𝐿𝑒𝑡 𝑢 𝜖 u ,
y(𝑢) be a solution of the following equation:
 a is bilinear, continous, coercive and symmetric,
−∆y(𝑢) = 𝑓𝑖𝑛Ω,
l is continous linear, (𝑆1) { (6)
𝛾y(𝑢) = 𝑔 + 𝑢 𝑜𝑛Γ.
u a nonempty closed convex set of L²(Ω),
We pose:
By stampacchia’s theorem, there exists a unique u
̅ in
usolution. 1 ∈
J(𝑢) = ∫ (y(𝑢) − y𝐷 )𝑑𝑥 + ‖𝑢‖𝐿2 (T), (7)
2 Ω 2
The solution u
̅ is characterized by:
2 ∈
= ‖y(𝑢) − y𝐷 ‖ + ‖𝑢‖2𝐿2(T).
𝑢̅ ∈ u 𝐿2 (Ω) 2
{ (5)
𝑎(𝑢̅, 𝜐 − 𝑢̅) ≥ 𝑙(𝜈 − 𝑢̅) ∀ 𝜈 ∈ u
IX. RESOLUTION
Having obtained the existence and uniqueness of the
solution, we are interested in giving the characteristics of u
̅.  Simplification
To do this, we introduce the optimality system. By By setting:
interpreting, the reaction y(𝑢) = y1 (𝑢) + 𝑧,

a(𝑢̅, 𝜐 − 𝑢̅) ≥ 𝚤(𝜐 − 𝑢̅) ∀ 𝜐 ∈ 𝑢, we obtain: Wherey1 (𝑢) is the solution of the problem:
−∆y(𝑢) = 0 𝑖𝑛Ω,
< 𝑦̅ − 𝑦𝐷 , 𝐴(𝜐 − 𝑢̅) > 𝐿2 (𝛺)+ ∈ < 𝑢̅, 𝜐 − 𝑢̅ > 𝐿2 (𝛺) ≥ 0. (𝑆2) { (8)
𝛾y(𝑢) = 𝑔 + 𝑢 𝑜𝑛Γ.
The Conjoint state 𝑝̅ ∈ 𝐻01 (𝛺) is the solution of the
following problem: And 𝑧 belongs to 𝐻01 (Ω) solution de :
−∆z = 𝑓𝑖𝑛Ω,
(𝑆1) { (9)
−∆𝑝̅ = 𝑦̅ − 𝑦𝐷 𝑖𝑛 Ω 𝛾z = 0 𝑜𝑛Γ.
{
𝛾𝑝̅ = 0 𝑜𝑛 Γ,
System 𝑆1 becomes
We therefore have −∆y(𝑢) = 0 𝑖𝑛Ω,
(𝑆1) { (10)
𝛾y(𝑢) = 𝑔 + 𝑢 𝑜𝑛Γ.
< 𝑝̅ + 𝜖𝑢̅, 𝜐 − 𝑢̅ > 𝐿²(Ω) ≥ 0.

IJISRT22DEC1510 www.ijisrt.com 2082

Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
We pose : X. CONCLUSION
1 ∈
J(𝑢) = ‖y(𝑢) − y𝐷1 ‖ + ‖𝑢‖2𝐿2(Ω). In short, to solve the problem we must optimize the
2 2
cost according to the goal we are looking for and taking into
Where account the means at our disposal. But the case here requires
y𝐷1 = y𝑑 -z. a good mastery of the solutions of optimal control and some
demonstrations to prove the existence, uniqueness and
 Transposition formula stability of solutions. Also, we give some extensions to our
We associate to (S2 ) the following transposition equation: investigation.
−∆y(𝑢) = 0 𝑖𝑛Ω, REFERENCES
(𝑆1) { (11)
𝛾y(𝑢) = 𝑔 𝑜𝑛Γ.
[1.] S. Agmon ; A. Donglis, L. Nirenberg ; estimates near
Let 𝑓 be a function of 𝐿2 (𝛺)( ) admitting a unique the boundary for solutions elliptic partial differential
solution, we get from ( ) equations satisfying general boundary conditions.
Comm. Pure Appl. Math, 12(1959); 623 −
𝜕𝑧 727. 𝐼𝐼. 𝑖𝑑. 17 (1964),35-92.
∫ (Δy1(𝑢)) 𝑧 𝑑𝑥 = 0 ⇒ ∫ 𝑓𝑧 𝑑𝑥 = ∫ (𝑔 𝑑𝜎. [2.] H. Antorienwiz; Newton’s method and boundary
Ω Ω Ω 𝜕𝑛
value problems. 5. of computer and system sciences,
Considering the following problem: 2(1968); 177-202.
[3.] Boundary value problems for non linear ordinary
𝑓𝑖𝑛𝑑 y 𝑖𝑛 𝐿2 (Ω)𝑠𝑢𝑐ℎ 𝑡ℎ𝑎𝑡 differential equation pacific journal mat; 17(1966);
(𝑃𝑉) { 𝜕𝑧 191-197.
∀ 𝑓 𝐿2(Ω) , ∫ 𝑓y𝑑𝑥 = − ∫ 𝑔 𝑑𝑥. [4.] Bailey et L. Champine et P. Woltman ; Non linear
Ω Ω 𝜕𝑛
two point boundary value problems. Accad. Press,
𝑙 𝐿2(Ω) ⟶ ℝ 1968.
[5.] A. V. Balakrishman; on a new computing technique
𝜕𝑧 in optimal control and its application to minimal time
𝑔 ⟼ −∫ 𝑔 𝑑𝜎. flight profile optimization. Proc. Not. Accad. Sc
Γ 𝜕𝑛
January 1968.
𝑙is linear and continuous application on 𝐿²(𝛺), by riesz’s [6.] E. J. Beltrami methods of non linear analysis and
theorem, there exists a unique 𝑦 ∈ 𝐿²(𝛺) such that optimization. Accad. Press 1969.
[7.] F. E. Browder, non linear elliptic boundary value
< 𝑓, 𝑔 > 𝐿2 (Ω) = 𝑙(𝑓); ‖𝑦 ‖𝐿²(Ω) = ‖𝑙‖ (ℒ((𝐿2 ((Ω), ℝ)). problems. Bull. Amer Math. Soc, 69(1963).
[8.] A. Friendman, Free boundary problems for parabolic
We define the operator equations.Molting of solids; J. Maths, 9(1959); 499-
518.
A𝐿2 (Γ) ⟶ 𝐿2 (Ω) [9.] J. L. Lions; some non linear evolution equations. Bull.
Soc. Math. France, 93(1965), 43-96.
𝑔 ⟼ A(𝑔) = y1 [10.] J. L. Lions and E. Magenos, Idem, Vol. 3, Dunod,
Replacing 1969.
[11.] O. A. Olinik; on the existence, uniqueness, stability
A(𝑓 + 𝑢) = 𝑦(𝑢) and approximation of solution of Prandtl’s system for
the non boundary layer. Rend. Acc. Naz…..,
In J(𝑢), 𝑊𝑒 have: 41(1966), 32-40.
[12.] V.I. CHESALIN, A problem with nonlocal boundary
J =J1 ; conditions for certain abstract hyperbolic equations,
Diff. Uravn., 15 (1979), No. 11, 2104-2106,
With [13.] V.I. CHESALIN, A problem with nonlocal boundary
1 conditions for abstract hyperbolic equations, Vestn.
J1 (𝑢) = {‖A(𝑢)‖2𝐿2(Ω) + 𝜖 ‖𝑢‖𝐿(Ω)}+ < A,A -y𝐷1 Beloruss. Gos. Univ. Ser. 1 Fiz. Mat. Inform., (1998),
2 No. 2, 57-60,
> 𝐿2 (Ω)

We pose

𝑎∶ 𝐿2 (Ω)x 𝐿2 (Ω) ⟶ ℝ

(𝑢, 𝜐) ⟼ < A(𝑢), A(𝜐) > 𝐿2 (Ω)+ < 𝑢, 𝜐 > 𝐿2 (Γ)

𝑙 ∶ 𝐿2 (Ω) ⟶ ℝ

𝜐 ⟼ −< A(𝑢) − y𝐷1 , A(𝜐) > 𝐿2 (Ω).

IJISRT22DEC1510 www.ijisrt.com 2083