Enhanced Power System Security Assessment Through Intelligent Decision Trees - Complex System Modelling and Control Through Intelligent Soft Computations - page 350

Table 3 Incremental
estimation of k

k P SEO

L

min MW ð Þ P SEO

L

max MW ð Þ Gap (MW)

5 12,500 12,700 200

8 11,627 12,500 873

12 12,000 12,700 700

15 11,627 12,700 1,073

20 11,627 12,650 1,023

25 11,627 12,700 1,073

is performed to characterize the load state space and find the boundary limits of

max , which is found to be {11,627, 12,700} MW
as shown in Table 2 . The margin sensitivities are also computed along every k stress
directions, which are used to estimate the change in boundary limits due to the
in

total SEO load P SEO
L

P SEO
L

;

min

uence of component combination change. Table 2 shows the estimated boundary
limits for all the remaining combinations. The

fl

final boundary limits are estimated as

11,446 and 12,700 MW.
Table 3 shows the process of estimating k for LHS in an incremental fashion.
Beyond k = 15, the boundary region is identi

ed fairly consistently. The Expec-
tation-Maximization algorithm based clustering method, when applied to historical
record of stress directions, optimally grouped the stress directions into 21 clusters.
This information is useful to quickly zero in on the ideal value for k.
Figure 11 shows the boundary characterization from a simulation performed for
24,000 operating conditions with randomly selected combinations of discrete

Fig. 11 Boundary characterization in total SEO load state space