Single line to ground fault distance estimation in medium voltage network based on measurement from secondary side of distribution substations

Abstract

Faults may occur in every power system network in the world. It can be caused by natural phenomenon and also human errors. Faults may cause damage to the system which resulting in system failure. Earth fault is the most common fault that occurrs in power system network. Faults cannot be avoided but the risk of damage caused by faults can be reduced. In order to reduce the risk, locations of fault must first be identified. Many methods can be used to identify fault locations in a network such as impedance-based method and wavelet method. The working principle of these methods are detecting fault locations using an algorithm which was created based on characteristics of the power network during fault events. The purpose of this project is to develop an algorithm regarding single line to ground fault distance estimation in medium voltage network based on measurement from secondary side of distribution substations. This project was done by modelling a medium voltage network using ATP-EMTP software. Based on the network model, measurement of voltage sag was taken when single line to ground fault occurs on the network. The measurement was taken at the secondary side of distribution transformer. All of the measurement data was then collected in order to train in artificial neural network (ANN) in MATLAB. Based on the fault location algorithm developed, it is shown that it was able to estimate the fault location distance accurately. The average error of the detection was less than 0.5 km and this proves that the algorithm works perfectly in detecting fault locations in power system network. Thus, a fault detection algorithm can be developed by using the voltage sag values collected.