Page 523 - Decision Making Applications in Modern Power Systems

482 Decision Making Applications in Modern Power Systems

an extensive dataset is generated as exemplified in Table 18.2 18.5. To gen-
erate an appropriate dataset for training/testing, extensive simulation studies
have been performed by considering wide variation in operating conditions
as follows: fault resistance ðR f Þ from 0 to 100 Ω, fault inception angle ðφ Þ
f
from 0 to 360 degrees, fault location ðL f Þ from 1 to 197 km, source imped-
ance (Z) within 6 10%, power flow angle (δ) from 5 to 20 degrees.
In this proposed protection scheme, exclusive datasets have been
designed using an illustrative feature extraction technique, that is, WPE. To

TABLE 18.2 Dataset generated for a wide range of variation in different
parameters for wavelet packet energy and bagged decision tree-1
modules.
Parameter Common shunt faults Power swing (stable and
unstable)
Fault on line NTPC/KSTPS-to-Bhilai/ NTPC/KSTPS-to-Sipat
Khedamara
Fault clearing time (cycle) 5 2, 4, 6, 8, 10, 12, 14, 16, 18,
20
Fault location ðL f Þ 198 km in steps of 2 km 60 km in steps of 3 km
Fault inception angle ðφ f Þ 0, 90, and 270 0, 90, and 270
(degrees)
Fault resistance ðR f Þ (Ω) 0, 50, and 100 0, 50, and 100
No. of fault cases LG: 2646, LLG: 2646 LLLG:1710
LL: 2646, LLL: 0882
Total no. of cases 9020 5 8820 1 200 (no 1710
fault)

TABLE 18.3 Dataset generated for a wide range of variation in fault
parameters for wavelet packet energy and bagged decision tree-2
modules.

Parameter Stable power swing Unstable power swing
Fault on line NTPC/KSTPS-to-Sipat NTPC/KSTPS-to-Sipat
Fault clearing time (cycle) 2, 4, 6, 8 10, 12, 14, 16

Fault location ðL f Þ 60 km in steps of 3 km 60 km in steps of 3 km
Fault inception angle ðφ f Þ (degrees) 0, 90, and 270 0, 90, and 270
Fault resistance ðR f Þ (Ω) 0 0
Total no. of cases LLLG: 228 LLLG: 228