Page 530 - Decision Making Applications in Modern Power Systems

A reliable decision-making algorithm Chapter | 18 489

FIGURE 18.8 Performance of WPE_BDT-2 module in case of unstable power swing: (A)
Three phase instantaneous currents of circuit-I (B) Three phase instantaneous currents of circuit-
II (C) Three phase instantaneous voltages at sending-end bus (D) Three stepped MHO relay
characteristics (E) Output of WPE_BDT-2 module during unstable PS. WPE_BDT, Wavelet
packet energy and bagged decision tree.

philosophy and the confusion matrix obtained for the tested cases has been
reported in Table 18.7 which illustrate the actual and predicted class for dif-
ferent combination of dataset. It can be observed that the proposed scheme
offers 99.8% accuracy in prediction of the fault during PS. The output of
WPE_BDT-3 module will be a two class target/output: 0 for no-fault during
the PS and 1 for fault during the PS. If the output class is “0,” the relay will
initiate a PSB function, and there is a no-fault condition. Further, if the out-
put class is “1,” the relay will initiate a trip command/signal. For instance,
phase-A1 to ground fault during PS is created at 3.018 seconds, and resultant
three-phase currents of both the circuit and voltages are depicted in
Fig. 18.9A C. The performance of three-stepped distance MHO relay with a
double blinder technique is reported in Fig. 18.9D wherein the apparent
impedance of the transmission line passes through the zone-1 for a phase to
ground fault (A1G). The output of WPE_BDT-3 module for detection of a
fault during PS has been demonstrated in Fig. 18.9E where the fault has
been detected at 3.035 seconds.

18.5.4 Performance evaluation of proposed wavelet packet energy
and bagged decision tree-4 module for classification of faults

If normal shunt fault is detected in the system in stage III or V of the pro-
posed algorithm depicted in Fig. 18.6, the next task is to classify the fault