Neural Networks in Forecasting Electrical Energy Consumption

Thispaperpresentsan artificial neuralnetwork(ANN)approach to electric energyconsumption(EEC)forecasting. In order providethe forecastedenergyconsumption,the ANNinterpolates betweenthe EECandits determinantsin a training data set. In this study, twoANNmodelsare presented and implementedon real EECdata. Thefirst modelis a univariate modelbasedon past consumptionvalues. Thesecondmodelis a multivariate model based on EECtime series and a weather dependent variable, namely,degreedays (DD).Forecastingperformance measuressuch as meansquare errors (MSE),meanabsolute deviations (MAD),meanpercentage square errors (MPSE) meanabsolute percentageerrors (MAPE)are presentedfor both models. Moreover, autoregressive and ARIMAmodeling was applied (Saab, Badrand Nasr2001). In this paper, two models were built to forecast electrical energy consumption(EEC)in Lebanonusing artificial neural networks(ANN).Thefirst modelis univariate and fully connected modelbased on past EEC values. The second model is a mu...

• Type: Conference
• Year: 2001
• Where: FLAIRS
• Authors: George E. Nasr, E. A. Badr, M. R. Younes
• Keywords: Artificial Intelligence, EECtime Series, FLAIRS 2001, Modelbasedon Past Consumptionvalues, Order Providethe Forecastedenergyconsumption