NSC92-2213-E-032-029 92 08 01 93 07 31

()

93 11 2

Hydro-Thermal Generator Maintenance Scheduling Via Particle Swarm

Optimization Method

NSC 922213E032029
92 08 01 93 07 31

()


Hydro-Thermal Generator Maintenance Scheduling Via Particle Swarm
Optimization Method
NSC 92-2213-E-032-029
928193731

20

(Particle Swarm Optimization
PSO)[1][2] Eberhart Kennedy

(PSO)

1. (PSO)(Multi Agents)

2.

(Particle Swarm OptimizationPSO)

PSO

PSO
PSO

Abstract:
After electric utilities deregulation, maintenance
scheduling is likely becoming more complicated and
important. Recently, due to the rapid growth of load demand,
and the difficulties of generating system expansion, spinning
reserve of Taiwan power system now is obviously far below
the acceptable level of 20%. How to provide a reliable electric
power to the customers has become a more important issue.
Therefore, it definitely needs a feasible planning method for
maintenance scheduling. This project aims to investigate the
capability of the Particle Swarm Optimization (PSO) in
solving the maintenance scheduling problem. PSO are general
optimal algorithm and have been successfully applied to many
applications. Compared with other method, the PSO can not
only reduce the processing time but also increase the accuracy
and reliability during the solution process. Results obtained
from a sample system show that the proposed PSO method
might be a good solution method.
Keyword: Maintenance scheduling, Particle Swarm
Optimization, levelize spinning reserve.

u n ( t ) = 0
01

3.1

3.1.1

[6]

2
T
f r = R ( t ) R ave
t =1
Nt
C t = Pn ( t ) u n ( t )
n =1

min

R (t ) =

C t Lt

R ave =

(6)

en t ln

en

ln

d n

sn

4.

Nt
S n 1 u n ( t ) MP
n =1

(2)

(7)

Sn

(3)

MP

5.

Lt
T
R (t )
t =1

sn < t < sn + d n

= 1, 2,3, L , N t , t = 1, 2,3, L , T
t

(1)

t < e n or t > l n + d n

Ns

1 u s (t )
s =1

(4)

u s (t )

(8)

R ave

Ns

Pn (t )

n t

u n (t )

n t

Nt

Lt

t
R (t )

3.2

(fitness
value)(V )

01

3.1.2

1. load balance
T
u n ( t ) Pn ( t ) = Lt
t =1

3.2.1
(PSO)
()

(Global Optimal)

(Local Optimal)
(Social

(5)

2. spinning reserve
SR ( t ) rt Lt
SR (t )

rt

3. maintenance window

Behavior)

(PSO)

1.
2.(Agents)
3.
4.

3.2.4 (PSO)
1.

Initial

2. Evaluation(fitness function)
(fitness value)
3. Fine the pBest

(pBest)
4. Fine the gBest

(gBest)
5. Update the Velocity and Position(9) (10)

2.

3.2.2

1.
(Particle)
2. (Particle) (fitness
function)(fitness value)

3. (Particle)

4. (Particle)

5.

3.2.5 (PSO)[3][4]
1.
20 40 10

100 200
2.

Vid (-10,10) 20
3.
c1c2 c1 = c2 = 2
c1
= c2 = 2 0 4
4.
GA

5.
(w) Eberhart Kennedy(1995) PSO
Shi Eberhart 1998 [5]
Shi Eberhart

(Simulated AnnealingSA)
Shi Eberhart w 0.8 1.2

3.2.3(PSO)
(PSO)

(pBest)
(gBest)?
(fitness function)(fitness value)

Vid = w Vid + c1 Rand () ( pid xid )

+ c 2 Rand () ( p gd xid )

xid = xid + Vid

(9)
(10)

Vid d
w (Inertia Weight)
c1c2
Rand() 0 1
Pid
pBest
Pgd
gBest
xid

3.3 (PSO)

fr

0.790496

0.20044

46893.8

46582.2

f r

1.

3.4

A.

610MW
4 5
(Particle Swarm
OptimizationPSO)

MW
600
500

588

600

60

60

538

200
200

400

372

250

250
250

60

B.
33

72
Successive Approximation Dynamic
Programming[7]

1.
2. 4.5
3.
4.
600 [8]
5.

433

435

100 100

412
328

300
200
Load

100

6
7

10

11

12

MW
600

610
525

500

465

512

540

400

250

393

490

476
388

538

200 200 250

250

495

488

60

60

60

100 100

10

fr

610
490

476

465

585

f r

412

300
200
Load

100

6
7

11

2
3
2 3

12

f r
4

2 3

fr

1.114

0.883

1.757

1.251

0.451

0.473

0.451

0.458

0.663

0.410

1.306

0.793

147%

86.6%

289.5%

174.36%

[1]

J. Kennedy and R. C. Eberhart, Particle swarm optimization. Proc.

IEEE intl conf. on neural networks Vol.IV,pp.1942-1948.IEEE
service center,Piscataway,NJ,1995.

[2]

1.114

0.883

1.757

1.251

0.352

0.391

0.358

0.367

0.762

0.492

1.399

0.884

fr

216.5%

125.8%

390.8%

240.9%

R. C. Eberhart and J. Kennedy, A new optimizer using particle

swarm theory. Proceedings of the sixth international symposium on
micro machine and human science pp.39-43. IEEE service center,
Piscataway, NJ, Nagoya, Japan, 1995.

[3]

http://web.ics.purdue.edu/~hux/tutorials.shtml.
[4]

Y. Shi, and R. C. Eberhart, Parameter selection in particle swarm

optimization. Evolutionary Programming VII: Proc. EP 98 pp. 591-

17044

17000

600. Springer-Verlag, New York,1998

PSO

16000
15000

[5]

Y. Shi, and R. C. Eberhart, A modified particle swarm optimizer.

Proceedings of the IEEE International Conference on Evolutionary
Computation, pp. 69-73. IEEE Press, Piscataway, NJ, 1998.

MW

14000

[6]

D. K. Bhattachary and Jyoti Parikh, A System Approach to LeastCost Maintenance Scheduling for an Interconnected Power System,

13000

IEEE Transactions on Power System, Vol. 10,No. 4, November

(1995).

12000
11000

[7]

H. H. Zurn and V. H. Quintana, Generator Maintenance Scheduling

via Successive Approximation Dynamic Programming, IEEE

10000

Transactions on Power Apparatus and Systems, Vol. PAS-94, No. 2,

pp. 666-671, March/April (1975).

9000

[8]

8000
1

7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70

[9]

1.

(PSO)

2.

f r
3.

(MW)

(MW) ($/KWH)

($/MBTU)

(/)

(weeks)

250

75

10.2

1.2

11

15

200

60

11.5

1.25

11

12

3
2

100

25

14.7

1.4

11

10

60

20

16.5

1.5

10

18

5
1

10

11

12

MW 465

476

388

325

312

290

240

288

333

335

328

412

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

PSO