Prediksi Beban Listrik Harian Pada Sektor Industri Berbasis SVM Dengan Kernel Polinomial

Abstract

The industrial sector need an information system of da ily electrical load forecasting, to control the electrical load, backup electrical en ergy and operational arrangements of the industrial activities. The electric load prediction information system must be accurate by a small error value for that go al is reached. The objective research prod uce an information systems for accurate electrical load da ilyforecasting by using three variables training data. They are times series of the pa st electric load da tas, the data of production capacity and da ytypes data. Based Support Vector Machine (SVM) using a polinomial k ernel f unction, theinformation system of daily electricity load prediction on the industrial sector is capable of producing 3.34% MAPE value by SVM training da ta of 11 months an d the system work s by Kernel-order polinomial 2