Improved Estimation of Electricity Demand Function by Integration of Artificial Neural Network, Principal Component Analysis, Data Envelopment Analysis and Data Preprocessing Methods

Due to various seasonal and monthly changes in electricity consumption and difficulties in modeling it with conventional methods, an integrated algorithm is proposed. This study presents an integrated Artificial Neural Network (ANN), Principal Component Analysis (PCA), Data Envelopment Analysis (DEA) and Data Preprocessing methods to estimate and predict electricity demand for seasonal and monthly changes in electricity consumption. Preprocessing and post processing techniques in data mining field is used in present study. We study the impact of data preprocessing and postprocessing on artificial neural network (ANN) performance and 680 ANN-MLP is constructed for this. DEA is utilized to compare constructed ANNs models. Average, min, max and standard deviation of mean absolute percentage error (MAPE) of each constructed ANN is used as DEA inputs. Another unique feature of this study is the utilization of principle component analysis (PCA) to define input variables versus trial and process method. Also, Preferred Time series model is selected from linear (ARIMA) and nonlinear model. For this, after selecting preferred ARIMA model, Mcleod-Li test is applied to determine nonlinearity condition. When, nonlinearity condition is satisfied, preferred nonlinear model is selected and compare with preferred ARIMA model and finally one of this is selected as time series model. Finally, a new algorithm is developed for time series estimation, in that for each case, ANN or conventional time series model is selected for estimation and prediction.

Keywords:

Neural Networks , Time Series Analysis , Electricity Consumption Forecasting , DEA , PCA.16). Models in these and other studies split to