Energy consumption forecasting

Competition on retail electricity market is growing and, thus, imposing new requirements on all the market players. To keep large consumers from leaving for wholesale electricity market and to protect them energy sales organizations need to introduce new management mechanisms that would preserve and enhance competitive condition of the company. As a dealer between residential consumer and power plant, energy sales enterprise must conduct permanent research on changes in regional need for energy. The task is to make the most accurate forecast of energy consumption for further purchase of necessary energy on wholesale market. Inaccurate forecast at deficit or excess of the purchased energy forces energy sales enterprise to buy up or to sell energy surplus at inexpedient price. The article considers energy consumption forecasting on the basis of data obtained from one of energy sales enterprises. Problem solution is based on development of model and forecast method. Model development involves principal component analysis. The authors analyze whether “ambient temperature” factor must be used, and reveal main regularities in energy consumption changes. The research shows that forecast plotting should involve data division by days of the week. The most effective model is model of forecasting for nearest three hours by previous hours. At that it is possible to obtain forecast with margin of error of3 % max.