Electrical Power Usage Prediction using A Multi Input Single Output Heuristic Network

Abstract

The electrical power usage forecasting is the basis for energy investment planning and plays an important role in developing institutions and agencies. The combination of computational mathematical concepts and computer technology has widely used for forecasting electric power usage while those methods proved very powerful to predict the electric power usage in the future. There are two main roots in logic and reasoning in the philosophy of science and mathematics which are the basis of all computational activities. One of them is a heuristic approach that has widely applied in various studies in the areas of predictive problems, selection, and search problems. In this study, the prediction of electric power usage for each category carried out by applying the concept of a heuristic network. Time series data modeling is done using a weighted network. The values of each network weighting are obtained using a heuristic approach. The purpose of this study is to simultaneously predict two categories of electricity use by implementing a heuristic network. The results of the study show that the MISO (Multi Input Single Output) Heuristic Network can be stated to be significant enough to carry out the activity of predicting two categories of time series data simultaneously. Furthermore, the results of this study obtained concluded that the parameter that has the most dominant influence on training results is the number of model orders.