SISTEM PERAMALAN KONSUMSI DAYA LISTRIK DENGAN ARTIFICIAL NEURAL NETWORK BACKPROPAGATION

Abstract

Artificial neural network is a multi-layer network implemented using computational program to solve complex problems through the process of calculating the Backpropagation algorithms. One problem that can be solved is a forecasting system based on time series data, time series data for the input data is using data of power consumption in kWh units on the LWBP and on the WBP, occupancy in 2011.The network has been trained to produce accurate forecasting MSE on the LWBP 7.48738 e-12, MSE on the WBP 1.11035 e-10, and MAPE on the LWBP 1.6141 e-9%, MAPE on the WBP 2.50 E-8%, for a network that is not trained to predict the power consumption to test the validity of the LWBP R = 0.837 and R = 0.835 the WBP. In the training and testing system optimized with 8 neurons in the hidden layer. Keywords : Artificial Neural Networks, Backpropagation algorithms, data time series, forecasting system