Electricity Forecasting For The Small Scale Power System Using Artificial Neural Network

The study is about to forecast the electricity demand values of UTP. The
electricity profile of GDC (UTP) has been analyzed based on the historical data
gathered. Using the analyzed data, forecast models have been developed prior to do
forecasting. The models are being developed using Artificial Neural Network method.
There are four models have been developed based on the conditions in UTP. Modell
is developed to forecast for one week ahead for Semester OFF. Model2 is developed
to forecast for one week ahead for Semester ON. Furthermore, Model 3 and 4 are
developed to forecast for 30 days ahead for Semester OFF and ON respectively.
Upon developed the robust models, all the models have been simulated using five (5)
different hidden neurons. As to obtained accurate forecasting result, the models have
been simulated for twenty simulations for each of the hidden layer. From that, the
error between forecasted and actual load have been obtained. From the result of the
error calculation, the best forecast model is being chosen. Upon completing the
project, the conclusion has been made based on the result from the forecasting as well
as the values of MAPE.