Pengaruh Perubahan Jumlah Layer Pada Prediksi Konsumsi Daya Listrik

Abstract

Electrical load forecasting is very important to meet consumer needs according to load patterns. The electricity generated if it exceeds the amount needed will cause waste, on the other hand if the generator is smaller than the demand it can result in rotating power outages which result in losses to consumers. The purpose of this study was to determine the performance metric of the mean absolute percentage error (MAPE) in forecasting the load one day ahead by using an artificial neural network (ANN) algorithm. Forecasting with this ANN uses variations in the number of layers, namely 1 layer, 2 layers, 3 layers and other constant parameters. Forecasting uses time series data in period one from 1-28 December 2017. Forecasting output finds that ANN 1 layer provides the best value with a MAPE of 5,0053%, and is able to identify existing load patterns.

Keywords

• layer, performance metrics, mape, load pattern, ann