Artificial neural networks applied to faulted section location in distribution network

Abstract Recently, there has been an increase in the usage of novel devices in distribution networks for fault locating and detection, such as fault indicators (FI). They provide additional data sets that can help with faulted section location (FSL). This paper proposes an artificial neural network (ANN) method that uses the FI statuses to determine the FSL in a radial distribution network. The advantage of this approach is its high accuracy, and the method does not require the analysis and calculation of any electrical variable. This method simplifies FSL since it requires no previous comprehension of electrical variables such as voltage, current, etc. It is unaffected by fault types, making it easy to implement in a distribution network. The proposed method for FSL is tested on the IEEE 33 bus system: different fault scenarios are simulated, including single fault, multiple faults, and loss of information, all with distributed generation (DG) in the distribution system. A method based on ANN shows high accuracy and can detect faulted sections even in the case of information losses. The used method attained an accuracy of 92.1% in the DN with a single fault, 90% in the DN with multiple faults, and 94.1% in the DN with multiple faults and DGs.