A forecasting model based on time series analysis applied to electrical energy consumption

One of the main objectives of the European Union (EU) is the reduction of energy consumption and the elimination of energy wastage. These two issues are extremely important, especially in large energy demanding areas, such as transportation, manufacturing, etc.. Electricity consumption prediction is a basic tool for energy management system. Precise prediction of transportation companies helps the energy providers to make right decision for proper distribution of electricity. In this paper, the authors present a Time Series Analysis Model and its application to the electricity consumption of public transportation in Sofia (Bulgaria) in 2011, 2012 and 2013. This technique is based on the dataset analysis and is able to arise the trend slope, the periodic pattern and the random component as a function of time. The innovation of the presented model is in the multiple seasonality and in its ability in following the monthly oscillations. The dataset analysed will show a strongly periodic pattern that will be reconstructed with three different seasonal coefficients. The adoption of statistical tests for linearity and stationarity will show that the series under study is nonlinear and stationary. Comparison between models with two and three seasonalities will be performed in terms of error analysis. A validation on the January 2013 dataset for the triple seasonality model will show interesting results in terms of very low mean error and standard deviation. In addition, a proper interpretation of the model coefficients will open the way to the implementation of improved energy management strategies.