Fault analysis in TCSC-compensated lines using wavelets and a PNN

• Published in: Neural computing & applications Vol. 30; no. 3; pp. 891 - 904
• Main Authors: Reyes-Archundia, E., Guardado, J. L., Gutiérrez-Gnecchi, J. A., Moreno-Goytia, E. L., Guerrero-Rodriguez, N. F.
• Format: Journal Article
• Language: English
• Published: London Springer London 01.08.2018; Springer Nature B.V
• Subjects: Algorithms; Artificial Intelligence; Computational Biology/Bioinformatics; Computational Science and Engineering; Computer Science; Data Mining and Knowledge Discovery; Discrete Wavelet Transform; Image Processing and Computer Vision; Neural networks; Original Article; Probability and Statistics in Computer Science; Transmission lines; Traveling waves; Wavelet analysis; Wavelet transforms; Thyristor-controlled series capacitor; Fault localization; Traveling waves; Probabilistic neural network; Discrete wavelet transform
• ISSN: 0941-0643; 1433-3058
• DOI: 10.1007/s00521-016-2725-6

This paper describes an algorithm to detect, localize and classify fault events in overhead transmission lines compensated with a thyristor-controlled series capacitor (TCSC). During a fault event, a complex pattern of traveling wave reflections and refractions is generated at the point of fault inception. The proposed algorithm uses the discrete wavelet transform combined with a probabilistic neural network to analyze all this information and determine whether a fault condition exists in the line, the fault type and also the fault distance. In order to assess the algorithm performance, several studies were carried out under varied conditions. The obtained results demonstrate that the algorithm accuracy for calculating the fault distance is smaller than 1% of the total line length, and a 100% efficiency for determining the fault type. The algorithm is also immune to harmonic interaction due to low-frequency harmonics generated by the TCSC. A comparative advantage over previous algorithms for TCSC-compensated transmission lines is the fact that the proposed algorithm not only identifies the faulted line section but also localizes accurately the distance to the fault, using only measurements at one end of the line.