Current signal based phase selection in EHV-transmission lines using wavelet transforms and neural network

This paper introduces the design and implementation of a novel phase selection technique using both wavelet transform algorithms and neural network technique to improve the accuracy and efficiency compared with traditional phase selection algorithm under a wide variety of different system and fault conditions. The technique is based on using sharp transitions of current signals generated on the faulted phase. A feature extraction method, based on wavelet transform decomposition, spectral energy extraction and fuzzy logic, is adopted for this work. The algorithm is based on neural network for the decision making part of the scheme. All the test results show that the designed algorithm is very well suited for both accurately classifying fault types and identifying the faulted phase(s) under a wide variety of different system and fault conditions in EHV-transmission lines.