Soft computing techniques for fault detection in power distribution systems: A review

Abstract

Power distribution systems play an important role in modern society. Increasing size and capacity of power systems have rendered them more complex which in turn has led to reduced reliability of such systems. Power distribution systems are always prone to faults. Faults in power systems are generally due to short circuits, lightning etc. Fast and proper restorations of outages are crucial to improve system reliability. For quick and adequate recovery actions such as the determination of the propriety of carrying out forced line charging and the necessity of network switching, and an efficient patrolling, understanding the cause of a fault in an electric power system in the system operation is essential. Moreover, unknown faults may add to unnecessary costs if effective restorations and identifications can't be done quickly. So, identification of a fault on a transmission line needs to be correct and rapid. However, recognition of the causes of distribution faults accurately generally lack in expert personnel. Also the knowledge about the nature of these faults is not easily transferable from person to person. So, effective means of fault identification needs to be encouraged. In this paper, some of the unconventional approaches for condition monitoring of power systems comprising of wavelet transform, along with the application of soft computing techniques like artificial neural networks, fuzzy logic, support vector machines, genetic algorithm and hybrid combinations based on these have been studied.