Chapter 8

67
1 Chapter 8 Chapter 8 Nonlinear Programming with Constraints

description

Chapter 8 Nonlinear Programming with Constraints. Chapter 8. Chapter 8. Chapter 8. Methods for Solving NLP Problems. Chapter 8. ; see Fig. E 8.1a. Chapter 8. Chapter 8. Chapter 8. Chapter 8. Chapter 8. Chapter 8. Chapter 8. Chapter 8. - PowerPoint PPT Presentation

Transcript of Chapter 8

Page 1: Chapter 8

1

Ch

apte

r 8

Chapter 8

Nonlinear Programming with Constraints

Page 2: Chapter 8

2

Ch

apte

r 8

Page 3: Chapter 8

3

Ch

apte

r 8

Page 4: Chapter 8

4

Ch

apte

r 8

Methods for Solving NLP Problems

Page 5: Chapter 8

5

Ch

apte

r 8

*)(**)( xhxf (a) Where 414.1/1* is called the Lagrange multiplier for the constraint h = 0

)21,21(* x ; see Fig. E 8.1a

Page 6: Chapter 8

6

Ch

apte

r 8

Page 7: Chapter 8

7

Ch

apte

r 8

Page 8: Chapter 8

8

Ch

apte

r 8

Page 9: Chapter 8

9

Ch

apte

r 8

Page 10: Chapter 8

10

Ch

apte

r 8

Page 11: Chapter 8

11

Ch

apte

r 8

Page 12: Chapter 8

12

Ch

apte

r 8

Note that there are n + m equations in the n + m unknowns x and λ

Page 13: Chapter 8

13

Ch

apte

r 8

Page 14: Chapter 8

14

Ch

apte

r 8

Page 15: Chapter 8

15

Ch

apte

r 8

Page 16: Chapter 8

16

Ch

apte

r 8

Page 17: Chapter 8

17

1 2Minimize : ( )f x xx

2 21 2Subject to : ( ) 25g x x x

By the Lagrange multiplier method.

Solution: The Lagrange function is

2 21 2 1 2( , ) ( 25)L u x x x x x u

The necessary conditions for a stationary point are

2 11

2 0L

x uxx

1 22

2 0L

x uxx

2 21 2 25

Lx x

u

2 21 2(25 ) 0u x x

Ch

apte

r 8

Page 18: Chapter 8

18

Ch

apte

r 8

Page 19: Chapter 8

19

Ch

apte

r 8

Page 20: Chapter 8

20

Ch

apte

r 8

Page 21: Chapter 8

21

Ch

apte

r 8

Penalty functions for handling equality constraints

Page 22: Chapter 8

22

Ch

apte

r 8

Page 23: Chapter 8

23

Ch

apte

r 8

for handling inequality constraints

Note g must be >0 ; r 0

Page 24: Chapter 8

24

Ch

apte

r 8

The logarithmic barrier function formulation for m constraints is

Page 25: Chapter 8

25

Ch

apte

r 8

Page 26: Chapter 8

26

Ch

apte

r 8

Page 27: Chapter 8

27

Ch

apte

r 8

Page 28: Chapter 8

28

Ch

apte

r 8

Page 29: Chapter 8

29

Ch

apte

r 8

Use xc = 2 yc = 2 for linearization

(step bounds)

Page 30: Chapter 8

30

Ch

apte

r 8

Page 31: Chapter 8

31

Ch

apte

r 8

Page 32: Chapter 8

32

Ch

apte

r 8

Page 33: Chapter 8

33

Ch

apte

r 8

Quadratic Programming (QP)

Page 34: Chapter 8

34

Ch

apte

r 8

8.3 QUADRATIC PROGRAMMING

Page 35: Chapter 8

35

Use of Quadratic Programming to Design Multivariable Controllers

(Model Predictive Control)

• Targets (set points) selected by real-time optimization software based on current operating and economic conditions

• Minimize square of deviations between predicted future outputs and specific reference trajectory to new targets using QP

• Framework handles multiple input, multiple output (MIMO) control problems with constraints on manipulated and controlled variables. Dynamics obtained from transfer function model.

Ch

apte

r 8

Page 36: Chapter 8

36

Successive Quadratic Programming

• Considered by some to be the best general nonlinear programming algorithm

• Repetitively approximates nonlinear objective function with quadratic function and nonlinear constraints with linear constraints

• Uses line search rather than QP step for each iteration• Inequality constrained Quadratic Programming (IQP)

keeps all inequality constraints• Equality constrained Quadratic Programming (EQP) only

keeps equality constraints by utilizing and active set strategy

• SQP is an Infeasible Path method

Ch

apte

r 8

Page 37: Chapter 8

37

Ch

apte

r 8

Page 38: Chapter 8

38

Ch

apte

r 8

solve for ,x

Page 39: Chapter 8

39

Ch

apte

r 8

Generalized Reduced Gradient (GRG)

Page 40: Chapter 8

40

Ch

apte

r 8

Page 41: Chapter 8

41

Ch

apte

r 8

Page 42: Chapter 8

42

Ch

apte

r 8

Page 43: Chapter 8

43

Ch

apte

r 8

Page 44: Chapter 8

44

Ch

apte

r 8

Page 45: Chapter 8

45

Ch

apte

r 8

Page 46: Chapter 8

46

Ch

apte

r 8

Page 47: Chapter 8

47

Ch

apte

r 8

Page 48: Chapter 8

48

Ch

apte

r 8

Page 49: Chapter 8

49

Ch

apte

r 8

Page 50: Chapter 8

50

Ch

apte

r 8

Page 51: Chapter 8

51

• sequential simplex• conjugate gradient• Newton’s method• Quasi-Newton

Ch

apte

r 8

Page 52: Chapter 8

52

Ch

apte

r 8

Page 53: Chapter 8

53

Ch

apte

r 8

Page 54: Chapter 8

54

Ch

apte

r 8

Page 55: Chapter 8

55

Ch

apte

r 8

Page 56: Chapter 8

56

Ch

apte

r 8

Page 57: Chapter 8

57

Ch

apte

r 8

Page 58: Chapter 8

58

Ch

apte

r 8

Page 59: Chapter 8

59

Ch

apte

r 8

Page 60: Chapter 8

60

Ch

apte

r 8

Page 61: Chapter 8

61

Ch

apte

r 8

Page 62: Chapter 8

62

Ch

apte

r 1

we have to supplement blast furnace gas with fuel oil, but we want to minimize the purchase of fuel oil.

Ch

apte

r 8

Page 63: Chapter 8

63

Ch

apte

r 1

define: X1 = amount of fuel oil used in generator 1X2 = amount of fuel oil used in generator 2X3 = amount of BFG used in generator 1X4 = amount of BFG used in generator 2P1 = mw output of generator 1P2 = mw output of generator 2

range of operation of generator 1 and generator 218 ≤ P1 ≤ 3014 ≤ P2 ≤ 25

Fuel effects in the generators are additive (can operateon either BFG or fuel oil)

Ch

apte

r 8

Page 64: Chapter 8

64

Ch

apte

r 1

10 units of BFG are available (on the average):

1 unit BFG = Btu equivalent of 1 ton/hr. fuel oil.

We need 50 mw power at all times.

5021 PP

Experimental data needed?

Ch

apte

r 8

Page 65: Chapter 8

65

Ch

apte

r 1

Mathematical Statement

1 2

fuel oil to gen 1 fuel oil to gen 2

Min f x x

a. operating ranges 18 ≤ P1 ≤ 30& requirements 14 ≤ P2 ≤ 25

P1 + P2 = 50

b. availability of x3 + x4

blast furnace gas

c. operating P11 (x1) fuel oilcharacteristics P12 (x3) BFG

P21 (x2) fuel oil P22 (x4)

BFG

gen 1

gen 2

2221212111 PPPPPP

Ch

apte

r 8

Page 66: Chapter 8

66

Ch

apte

r 1

BFG addmay

wesince sufficientnot isequation this

00145.15186.4609.1 21111 XXP

22212

422

221

12111

312

BFG )(

oil fuel )(

1generator in reqt.)(

PPP

XP

XP

PPP

XBFGP

fcn of burners, heat transfer characteristics (convex functions)

Ch

apte

r 8

Page 67: Chapter 8

67

Ch

apte

r 1

Solution

NLP 4 ineq. const.piece-wise LP 6 eq. const.

1

2

3.05

30

20

optf

P

P

No fuel oil is used in generator 1.In generator 2, fuel oil provides 58%of the power (rest is BFG).

heat transfer characteristics may change, or BFG mayvary w.r.t. time (on-line solution)

Ch

apte

r 8