Desain Arsitektur Teknik Cerdas Hibrid untuk Identifikasi Dan Klasifikasi Gangguan Di Sistem Distribusi

Yoldi Yuwandira, Azriyenni Azhari Zakri


This paper presented a method of adaptive neuro fuzzy inference system (ANFIS) for classified short circuit faults in power distribution line at PT. Chevron Pacific Indonesia (CPI) in Bangko which has 4 branch line which are divided into BKO1, BKO2, BKO3 dan BKO4 with the length of each line are 30 Km, 25 Km, 25 Km dan 30 Km. Fault classification can improve the reliability of the power system. It needs quicken repair process and also reduces the possibility of equipment damage. The power distribution line is modeled using ETAP software. Short circuit types that simulated are phase to ground (F-T), phase to phase (F-F), phase to phase to ground (F-F-T), and three-phase (F-F-F). The ANFIS design is constructed using Matlab GUI 2016a version toolbox. Input data for this ANFIS design are short circuit value which is fed by three phases of voltage and current. The result of the simulations and data testing obtained Root Mean Square Error (RMSE) value for each line in Bangko PT. CPI with each membership function of GaussMF and GBellMF are 0.00046435% and 0.00029686% for BKO1, 0.000021651% and 0.00048023% for BKO2, 0.00021065% and 0.00029155% for BKO3, and then 0.0001479% and 0.00025739% for BKO4 respectively.
Keywords: ANFIS, fault classifier, membership function, power distribution, short circuit faults

Full Text:



  • There are currently no refbacks.