Page 521 - Decision Making Applications in Modern Power Systems

480 Decision Making Applications in Modern Power Systems

FIGURE 18.4 Variation of SD of WPE with fault distance (A) SD values of the current of
phase-A1 during shunt fault and fault during power swing (B) SD values of voltage of phase-A1
during shunt fault and fault during power swing. SD, Standard deviation; WPE, wavelet packet
energies.

voltage respectively during PS and fault during PS with respect to varying
fault distance is shown in Fig. 18.4A and B. It is clear from Fig. 18.4 that
SD values of faulted phase current and voltage have a distinct variation with
respect to fault distance along the complete line length. Thus SD values of
all three-phase voltages and currents have been taken as input feature set to
design the input matrix for training and testing of the proposed intelligent
decision-making algorithm.

18.4.2 Bagged decision tree
In the perspective of machine learning, tree-based ensemble techniques
are commonly used for prediction and classification purposes in many
research fields such as protection of transmission lines for instance. One