Page 79 - Mathematical Models and Algorithms for Power System Optimization

Optimization of Annual Generator Maintenance Scheduling 69

Unit 1 Unit 2 Unit 3 Unit N

K feasible maintenance
start times

N generator units
Fig. 3.9
Search strategy of the fuzzy dynamic programming.

The recursive algorithm of the FDP is:

μ
μ
μ
μ
μ
μ
μ
μ D Þ ¼ max min e ð i, tÞ, e ð i, tÞ, e ð i, tÞ, e ð i, tÞ, e ð i, tÞ, e ð i, tÞ, e i 1, tÞ
e i, tð
ð
c2
c1
g2
c4
D
c3
g1
J T
(3.20)
The optimal decision is:
μ
μ Þ ¼ max e l, kÞg (3.21)
f
ð
e optð
D D
K
μ
where e i, tð Þ—highest fuzzy membership function value when reaching state (i,t), J—j 2 J,
D
the set of maintenance start time acceptable for unit i, T—t 2 T, the set of maintenance start
time for unit i, K—k 2 K, the set of maintenance start time acceptable for the last unit i, l—last
unit l scheduled for maintenance.
3.6.2 Main Calculation Procedure
The solution procedure based on the proposed approach is shown in Fig. 3.10. At Step 1, the
GMS data file is read in. At Step 2, the general information, such as the study interval and time
intervals of each date, etc. are formed. At Step 3, the priority is determined by the ES or
dispatcher in the input data. At Step 4, the equivalent load curve is formed. At Step 5, the
reserve margin is constructed. At Step 6, based on the ES, dynamic programming is applied to