to the solution of a bilinear optimal control problem with state constrains by the...

34
To the Solution of a To the Solution of a Bilinear Optimal Control Bilinear Optimal Control Problem with State Problem with State Constrains Constrains by the Doubled-Variations by the Doubled-Variations Method Method E.A. Rovenskaya E.A. Rovenskaya Lomonosov Moscow State University, Lomonosov Moscow State University, Moscow, Russia Moscow, Russia and International Institute for Applied and International Institute for Applied Systems Analysis, Systems Analysis, Laxenburg, Austria Laxenburg, Austria

Post on 19-Dec-2015

218 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

To the Solution of a Bilinear To the Solution of a Bilinear Optimal Control Problem with Optimal Control Problem with

State Constrains State Constrains by the Doubled-Variations by the Doubled-Variations

MethodMethodE.A. RovenskayaE.A. Rovenskaya

Lomonosov Moscow State Lomonosov Moscow State University, University,

Moscow, RussiaMoscow, Russia

and International Institute for and International Institute for Applied Systems Analysis, Applied Systems Analysis,

Laxenburg, AustriaLaxenburg, Austria

Page 2: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Problem Formulation Problem Formulation

Assumptions:

(A1)

(A2) and are commuting matrices;

(A3) is a continuous function increasing in x for all t

and measurable in t

(A4) for all is a continuously-differentiable function,

(A5)

)0(

0)(),(

)0(),()()()(

max))(,()]([

0

0

t

txtc

xxtutBxtAxtx

dttxtfuJT

nRtx )(1],[)( Ruutu

00 x

A Bnn RRTxtfxt ],0[:),(),(

nRTxcp ],0[:)(:00),0( 0 xc

0J0U

optimal utility value

the set of optimal controls

Page 3: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Equivalent formulation Equivalent formulation

t

tBpAt dssutpxetx0

0)( )()(,)(

0)0(),()()()( xxtutBxtAxtx

],0[ Tt

T

dttxtfJ0

))(,(

TT

tBuAt dttptdtxetfJ00

0)( ))(,(),(

:)( pp is a continuous increasing function

Page 4: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

0),()( 0)( xetctv tBpAt

0)(),( txtc

)()(

)()()(

),()()()(

*2

*1

0)(

21

tcBt

tcAtct

xettuttv tBpAt

0),0()0( 00 xcvvAssumption (A5)

Equivalent formulation Equivalent formulation

0),0( 0 xc

Page 5: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

max))(,()]([0

T

dttptuJ

0)0(,),()()()( 00)(

21 vvxettuttv tBpAt

t

dssutp0

)()(

]),0[( Tt

],[)( uutu

0)( tv

),()(

),()(

utvtv

utptp

0J

0U

optimal utility value

optimal controls set

Equivalent formulation Equivalent formulation

Page 6: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Discrete approximation Discrete approximation

:],0[ T ,iti ),...,1,0( mi Ttm

},{)( uuutu i),[ 1 ii ttt

mm uuuuuu },{),...,,( 110

),( utp 0)(),( 0 uputp

1

0

)(),(i

jji uuputp

),[ 10 ttt

),[ 1 ii ttt

))(),...,(),((),( 110 upupuputp m

Uniform grid on

control approximation)1,...,1,0( mi

:)(tu

approximation of :)(tp

)(tu

)1,...,1,0( mi

Property:

*

*

),(),(

),(),(

uutputp

uutputp

*2),(),( uutputp

Page 7: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

],0[ t),(

2*

*

*),(),(*

)(

)(

)(1)(),( utpautpautpa e

tE

tE

tEetuaeauutv

)(2*

*

*)()(*

)(

)(

)(1),( upaupa

i

uipa ii etE

tE

tEeuaeauutv

),( utvApproximation of :)(tv

),[ 1 ii ttt)1,...,1,0( mi

0),0( vuv

)(uI

,))(()),(()(1

00

m

iii

t uphdtutpeuI

1i

i

t

t

ti dteh

),...,1,0( mi

Аппроксимация показателя качества :)(uJ

где

Discrete approximation Discrete approximation

Page 8: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

),( utv i

max))(()(1

0

m

iii uphuI

mm uuuuu },0{),...,,( *110

0I

0U

Optimal utility value

Optimal controls set

),( ProblemProblem

Page 9: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

kkkk IUuu 0000 ,)( ),( kk

Let

.0,0 kk

,0,0 kk ,00 JIk 00 Uuk ].,0[2 TL

be an optimal control in problem,

Then

(i)

(ii)

so that

weakly in

LemmaLemma 1 1

),( Problem Problem

Page 10: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Solution to Solution to ),(

)(2*

*

*)()(*

)(

)(

)(1)( upaupa

i

uipa ii etE

tE

tEeuaeautv

*0)(

0)(

0

,)(

0,)(),(

uuеслиet

uеслиetutv

i

uipa

i

uipa

Problem Problem

),[ 1 ii ttt )1,...,1,0( mi

1

1

)(

)(

i

i

i

i

t

t

i

t

t

i

dtt

dtt

0)(

)(

)()(

0)(

)()(

2*

*

*

*

2*

**

tE

tE

tE

aut

tE

tEaut

Условия (У2) и (У3)

*

0)(

0)(

001

,

0,),(),(

0

0

uuеслиe

uеслиeutvutv

iupai

iupai

ii

i

i

Рассмотрим 00 Uu

Page 11: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Решение ),(

0u ),( 0utv i

задачи

- допустимый элемент, то ),...,1,0( mi

Определение.

mj tt - критическая точка, если замена 0ju

*u приводит к

),( 01 utv j или ,),( )(0

0

upajj

jeutv

Так как

на

mj tt - регулярная точка в противном случае.

Утверждение.

.00 jumj tt Пусть - критическая точка. Тогда

Лемма 2.

.*0 uu j 2 mj ttПусть - регулярная точка. Тогда

Page 12: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Решение ),(

),...,( ** uuU }),(:2,...,1{ UtvmiL i

задачи

Рассмотрим

L

.min* iiL

:0K )},0{],,0[(),( * muutKutv

Лемма 4.

Пусть

L не пусто.

Тогда ).3,...,0(*0 miuu j

Лемма 3.

пусто.

Тогда (i)

(ii)

),2,...,0( **0 iiuu j

]).,[(2),( 20

* mitttKutv

и

Пусть

Пусть

Page 13: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Решение исходной задачи

tесли

tEtaE

tE

tE

u

tеслиu

tu ,

)(1)(

)(

)(1

,

)(

*

2*

*

*

*

*

0

taueEtE*

0* )(

Теорема.

Управление

оптимально в исходной задаче.

где -единственный корень уравнения

Page 14: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Решение исходной задачи

tеслиeE

E

tE

tET

tеслиeTtT

twtau

tau

,)(

)(

)(

)(

,)(

),()(*

*

*

*

0

)(0

0

*

*

tеслиeE

EEC

tеслиeCtC

twtau

t

,)(

)(

,)(

),()(*

**00

00

*

tеслиtE

tеслиeEtE

tau

),(

,)(

*

00

*

Оптимальные траектории

t

dssE

tw

)(

),(*

Page 15: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Моделирование

ConstEtE ** )(

tеслиeT

tеслиeTtT

thau

tau

,

,)(

))((0

00

*

*

tеслиeeh

hTC

tеслиCtC

thauau ,11

,)(

)(00

00

**

tеслиE

tеслиeEtE

tau

,

,)(

*

00

*

0

*

*ln

1

E

E

au

*

1E

h

0

0

)(ln dttYeJ t

Page 16: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Калибровка модели

0.4a

год2050

006.0*u

годГтCE /0.13*

USDтрилT .6.60

USDтрилC .3.00 годГтC /7.0

год2000

годГтСE /97.60 103.0 год

Начальный момент времени

Конечный момент времени

Page 17: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Моделирование

год2026

)(tY)(tC

Время переключения

)(tE

Page 18: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 19: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 20: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 21: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 22: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 23: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 24: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 25: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 26: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 27: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 28: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 29: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 30: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,
Page 31: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Введение

Измененияклимата

Влияние на окружающую среду

и жизнь человека

Эмиссии и концентрации

парниковых газов

Экономическийрост

Повышениетемпературы

• повышение уровня Мирового океана• дальнейшее потепление• лесные пожары• …

• сельское хозяйство• запасы пищи и пресной воды• здоровье людей• …

Page 32: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Модель: экономическая часть

)(tC

)(tY валовый выпуск продукции

капитализация «производственных» технологий

капитализация «очищающих» технологий

)(tT

)(tY

0

))(( dttYfeJ t

)())(( * tYtuu

Показатель качества max

развитие технологий)(* tYu

)()( tYtu

развитие «производственных» технологий

развитие «очищающих» технологий

)()( taTtY

),()()()( tTtTtautT

),()())(()( * tCtTtuuatC

0)0( TT

0)0( CC

Page 33: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

)(0 * tE

)(tE)(

)(

)(

)()( 0 tC

tT

tC

tYtE

)()( * tEtE

Модель: экологическая часть

промышленные выбросы парниковых газов

(У1)

(У2) монотонно убывает на

(У3)

)(* tEt

],0[ t

)(|)(|02*

** tE

autE

для всех

функция ],0[ t

],0[ tдля всех

(У4) ,*00 EE ),0(**0 EE

0

00 C

TE где

Page 34: To the Solution of a Bilinear Optimal Control Problem with State Constrains by the Doubled-Variations Method E.A. Rovenskaya Lomonosov Moscow State University,

Постановка задачи

)()( taTtY

)()())(()( * tCtTtuuatC )()()()( tTtTtautT

max))((0

dttYfeJ t

)()(

)( * tEtC

tT

0)0( TT 0)0( CC

],0[ t

Uu )(

U

],0[)( *utu

Множество допустимых управлений

-- множество таких измеримых функций

:)(u ]),0[( t

монотонно возрастает на

)(yfy функция

],0[ t

Дополнительное условие:

Будем считать, что множество допустимых элементов не пусто

непрерывна,