Lightning Fault Classification for Transmission Line Using Support Vector Machine

• Main Authors: Asman S.H., Aziz N.F.A., Kadir M.Z.A.A., Amirulddin U.A.U., Roslan N., Elsanabary A.
• Other Authors: 57194493395
• Format: Conference Paper
• Published: Institute of Electrical and Electronics Engineers Inc. 2024
• Subjects: accuracy; computational time; k-Nearest Neighbor (k-NN); lightning fault; Support Vector Machine (SVM); Electric lines; Electric power transmission; Learning algorithms; MATLAB; Motion compensation; Nearest neighbor search; Transmissions; Accuracy; Computational time; Failure trees; Fault classification; K-near neighbor; Lightning faults; Lightning strikes; Support vector machine; Support vectors machine; Transmission-line; Support vector machines

Transmission lines are susceptible to a variety of phenomena that can cause system faults. The most prevalent cause of faults in the power system is lightning strikes, while other causes may include insulator failure, tree or crane encroachment. In this study, two machine learning algorithms, Support Vector Machine (SVM) and k-Nearest Neighbor (kNN), were used and compared to classify faults due to lightning strikes, insulator failure, tree and crane encroachment. The input variables for the models were based on the root mean square (RMS) current duration, voltage dip, and energy wavelet measured at the sending end of a line. The proposed method was implemented in the MATLAB/SIMULINK programming platform. The classification performance of the developed algorithms was evaluated using confusion matrix. Overall, SVM algorithm performed better than k-NN in terms of classification accuracy, achieving a value of 97.10% compared to k-NN's 70.60%. Moreover, SVM also outperformed k-NN in terms of computational time, with time taken by SVM is 3.63 s compared to 10.06 s by k-NN. � 2023 IEEE.