Transmission line fault detection and classification using cross-correlation and k-nearest neighbor

Article type: Research Article

Authors: Dasgupta, Aritra^{a; *} | Debnath, Sudipta^b | Das, Arabinda^b

Affiliations: [a] Department of Electronics and Instrumentation Engineering, Netaji Subhash Engineering College, Kolkata, India | [b] Department of Electrical Engineering, Jadavpur University, Kolkata, India

Correspondence: [*] Corresponding author: Aritra Dasgupta, Department of Electronics and Instrumentation Engineering, Netaji Subhash Engineering College, Kolkata 700152, India. E-mail:[email protected]

Abstract: A method for detecting and classifying transmission line faults using cross-correlation and k-Nearest Neighbor (k$-NN) has been presented in this article. A unique analogy between the cross-correlogram obtained from the sound phase and a faulty phase in an electric power system, defined here as the fault correlogram, and a normal electrocardiogram (ECG) of human heart has been validated in the proposed work. The proposed method uses synthetic fault data within half cycle of pre-fault and half cycle of post-fault to detect and classify the different faults under varying fault parameters. EMTP/ATP software has been used as the platform to carry out simulation of the power system network followed by signal processing in MATLAB.

Keywords: Fault, cross-correlation, k-NN, electrocardiogram (ECG)

DOI: 10.3233/KES-150320

Journal: International Journal of Knowledge-based and Intelligent Engineering Systems, vol. 19, no. 3, pp. 183-189, 2015