High Impedance Fault Detection in Distribution Network Using Knowledge Reasoning and ANFIS

The paper proposes a high impedance fault (HIF) detection method in distribution network using knowledge reasoning and adaptive-network-based fuzzy inference system (ANFIS). First, a number of characteristic quantities are list based on the zero-sequence current (ZSC) of HIF. Second, a subset that best reflects the fault characteristics is filtered out from alternative feature quantities based on theoretical analysis and practical testing. Then, feature quantity vectors calculated from each sample are used to train the HIF inference model based on the adaptive fuzzy neural network. The test result shows that, the third harmonic content, fifth harmonic content, total harmonic content, root mean square, peaking factor of ZSC better reflect the HIF fault information. The accuracy of the obtained fault inference model for the identification of high resistance grounding faults in distribution networks reaches 98.72%.