Equivalent circuit of a 6–10 kV cable network for locating single-phase earth faults using an artificial neural network

This is the authors' abstract. We don't add key points for this paper.

Location of single-phase earth faults in 6–10 kV cable networks is a challenging task, especially in the presence of complex transient processes and various neutral grounding regimes. Traditional methods based on the analysis of zero-sequence currents and voltages often require complex and time-consuming calculations and do not always provide high fault location accuracy. The use of artificial neural networks improves the efficiency of single-phase earth fault location through automatic transient analysis and adaptation to changing network conditions and parameters. However, training and testing the neural network requires an accurate equivalent circuit for the 6–10 kV cable network. To study transient processes during earth faults, T-shaped and L-shaped equivalent circuits of a 6–10 kV cable line have been used. Modeling of ground faults has been performed in MATLAB/Simulink. Also, an artificial neural network has been developed to determine the location of damage based on transient parameters. A T-shaped equivalent circuit for a 6–10 kV cable network is proposed, taking into account the portion of the network beyond the earth fault location. The trained neural network has demonstrated high fault location accuracy under various neutral grounding conditions and transient resistances. The use of a T-shaped equivalent circuit for a 6–10 kV cable network in training and testing an artificial neural network as the basis of the method of locating single-phase earth faults allows increasing its selectivity and adaptability. It is especially important for networks with a voltage of 6–10 kV operating with different neutral grounding modes for different types of earth faults.