Regression Analysis of Time Series for Forecasting the Electricity Consumption of Small Consumers in Case of an Hourly Pricing System

Short-term electricity load forecasting is required for many functions, for example managing and controlling power systems, planning loads for power stations etc. In an open electricity market, load forecasting is necessary for electricity sellers, who have to buy necessary amounts of electricity from power exchange. Therefore, forecasting accuracy affects the seller’s economic results. Also, accurate load forecasting could help retail electricity sellers to offer consumers real-time variable electricity tariff packages. A lot of different methods have been used for load forecasting and many novel methods have been proposed and designed. The aim of this article is to analyze dependence of load of a small group of residences in Estonia on temperature and to show how it is possible to forecast electricity consumption one day ahead with a simple regression analysis of time series method. In addition, it gives an overview of the accuracy of regression analysis of time series method, and shows how to correct the day-ahead forecast error when forecasting two hours ahead. The results of this study demonstrate that regardless of a very large stochastic component, a relatively accurate load forecasting is possible when using the regression analysis of time series method. This article also shows how electricity sellers could use regression analysis of time series method to forecast load in order to offer consumers a real-time electricity pricing system. In addition, the regression analysis of time series method based forecasting model proposed in this article can also be used by electricity sellers to make optimal purchases from power exchange and increase their rate of profit. Key-Words: regression analysis, real-time pricing, time series forecasting, short-term load forecasting, load forecasting based on temperature.