PREDIKSI THD TEGANGAN SISTEM TENAGA LISTRIK MENGGUNAKAN SUPPORT VECTOR MACHINE DENGAN FUNGSI KERNEL GAUSSIAN RBF

Abstract

The voltage wave distortion that occurs in the electric power system will adversely affect the operation of the power system and provide an adverse effect on the loads using the voltage. The voltage wave distortion is measured by the amount of harmonic distortion (Total Harmonics Distortion of Voltage, THD V). The level of THD V in the electric power system should always be monitored in order to anticipate the adverse effects. One method of monitoring the harmonic level is by predicting THD V. This research produces a prediction system of voltage harmonics in power system using Support Vector Machine with Gaussian Kernel RBF
function. SVM is an intelligent machine system that has been proven superior when applied as a prediction method. This study uses three schemes in testing SVM system, they are the use of one variable as SVM training data, two variables as SVM training data, and three variables as SVM training data. The best result obtained in this research is when the prediction system of THD V using SVM is given train data with one variable that is past THD V variable, which yield MAPE 3,25%.
Keywords : Forcasting, Harmonics, THD V, SVM, Gaussian RBF