A hybrid artificial neural network and wavelet packet transform approach for fault location in hybrid transmission lines

Abstract

This paper presents a single-ended traveling-wave-based fault location (F.L.) method in a hybrid transmission line (HTL) with an overhead section combined with a cable section. For this, the software has been developed in a MATLAB programming environment. Wavelet packet transform is used to extract transient information of the aerial mode current and voltage signals. The normalized current and voltage wavelet entropy (features) are fed to the feature selection part of the software. Regarding the HTL construction, the optimal features are obtained using the support vector machine and particle swarm optimization. A three-layer artificial neural network is trained to identify the faulty section and half using the optimal features of post fault signals. The square of the aerial mode voltage wavelet coefficients is applied to locate the fault using Bewley's diagram. The proposed approach is applied for F.L. in HTL. Transient simulations are obtained through EMTP-RV software for various fault scenarios, including fault types, resistances, inception angles, and locations. The post fault signals are fed to the developed software. The results illustrate the high accuracy of the proposed method in comparison to previous works.