Prediction of Short-Term Electricity Consumption by Artificial Neural Networks Using Temperature Variables

Today, energy consumption is one of the most important indicators of countries' development levels. Energy, the most important inputof social and economic development, is a necessary requirement for the increase of living standard and sustainable development.Electricity is one of the most preferred and consumed energy types because of easy use, easy transportation and clean energy. Electricityconsumption varies depending on various social and economic variables such as population, economic growth and gross domesticproduct as well as on climatic variables such as temperature, precipitation and humidity. The electricity used for heating and coolingneeds is bigger in electricity consumption. Weather conditions cause increase and decrease in electricity consumption. Temperature isthe meteorological variable with the highest effect. As the comfort temperature gets away from the accepted temperature range, theelectricity consumption also increases. The balance of electricity production and consumption is also important in terms of atmosphericdisasters. In study, the relation of Turkish electricity consumption with temperature was examined between January 2012 and November2016. In monthly and seasonal time periods, it was researched how the consumption was changed due to the temperature and how muchit was changed and it was aimed to make a more consistent consumption estimation by adding as a temperature input to the consumptionestimation model. In the study, short-term electricity consumption estimations were made by using Artificial Neural Network (ANN)method and data groups modeled by Levenberg-Marquardt backpropagation algorithm on MATLAB programing language. Thetemperature data is produced with a weighted average by consumption amount. The temperatures of 12 provinces with the largest shareof consumption in Turkey are weighted according to their consumption rates. In the model weighting, the percentage of effect 1 daybefore is more than 1 week before and 1 week before is more than 1 year before. For this reason, deviations in the recent history moreinfluence the model. Model results show that, error rates are considered to be reasonable. It is planned to establish the model on aregional basis for future work. When estimating regional electricity demand, a model can be developed by using different meteorologicalvariables. It is predicted that rainfall data will increase the performance of estimates of the inclusion of temperature data in the Bosphorusregion, which has a high population density and a high level of residential consumption.

PDF

___

E. Almeshaiei ve H. Soltan, A methodology for Electric Power Load Forecasting, Alexandria Engineering Journal, no. 50, p. 137–144, 2011.
H. Toros, R. Ayaz, A. Ajder, A. Durusu ve H. Yıldırım, Online Load Shifting Of Electricity ProductionConsumption For Reducing Environmental Hazard, European Journal of Science and Technology (EJOSAT), cilt 1, no. 2, pp. 39-42, 2013.
A. S. Khwaja, X. Zhang,, A. Anpalagan and B. Venkatesh, Boosted neural networks for improved shortterm electric load forecasting, Electric Power Systems Research, p. 431–437, 2017.
J. Moral-Carcedo ve J. PerezüGarcia, Temperature effects on firms’ electricity demand: An analysis of sectorial differences in Spain, Applied Energy, no. 142, p. 407– 425, 2015.
J. C. Lam, «Climatic and economic influences on residential electricity consumption,» Energy Conversion and Management, cilt 39, no. 7, pp. 623-629, 1998.
G. Aneiros, J. Vilar ve P. Raña, Short-term forecast of daily curves of electricity demand and price, Electrical Power and Energy Systems, no. 80, pp. 96-108, 2016.
L. Friedrich ve A. Afshari, Short-term forecasting of the Abu Dhabi electricity load using,» Energy Procedia, no. 75, pp. 3014-3026, 2015.
A. Azadeh, S. F. Ghaderi ve S. Sohrabkhani, Annual Electricity Consumption Forecasting By Neural Network İn High Energy Consuming İndustrial Sectors of Iran, Energy Conversion and Management, no. 49, p. 2272– 2278, 2008.
S. Yıldız, Türkiye Elektrik Piyasası Kısa Dönemli Referans Fiyat Tahmini,, İTÜ, Enerji Enstitüsü, İstanbul, 2015.
J. Boylan, « Toward a More Precise Definition of Forecastabilitiy,» Business Forecasting, SAS Institute Inc., Cary, North Carolina, USA.
EPDK, Elektrik Piyasası 2015 Yılı Piyasa Gelişim Raporu, T.C. Enerji Piyasası Düzenleme Kurumu (EPDK), Ankara, Türkiye., 2016.