GA-GNN (Genetic Algorithm-Generalized Neural Network)-Based Fault Classification System for Three-Phase Transmission System
- 53 Downloads
This research paper proposes a synergetic approach for fault classification of a three-phase transmission system. The voltage signals of all three phases at generating bus of the transmission system are acquired and processed for different operating (healthy and unhealthy) conditions. The analysis of signals is done by Clarke transform and Fourier transform. Clarke transform is used for the detection of involvement of the ground in the fault. Various types of faults such as symmetrical faults and asymmetrical faults have been considered (faults among two phases, phase A and phase B (AB), phase B and phase C (BC) and phase C and phase A (CA), faults in different phases and ground, phase A and ground (AG), phase B and ground (BG), phase A, phase B and ground (ABG), etc.). The phase angle of the measured three-phase voltage at one point of the transmission line is calculated with Fourier transform and compared in order to distinguish the type of fault. These obtained features are then utilized by trained GA-GNN for differentiating various types of faults. The different types of fault data are summarized in a normalized matrix form, and this normalized matrix is fed as input to the fuzzy logic and GA-GNN. The method is very easy in implementation and obtained results show that the method is very accurate and has practicability.
KeywordsClarke transform Fast Fourier transform Genetic algorithm Generalized neural network Fault identification
- 12.M. Rizwan, M.A. Kalam, M. Jamil, A.Q. Ansari, Wavelet-FFNN based fault location estimation of a transmission line. Electr. Eng. Res. (EER) Int. Refereed J. USA 1(3): 77–82, 2013. ISSN: 2327-7254 (print), 2327-7564Google Scholar