Energy Forecasting Models Research Articles

Medium-term energy outlook for Cyprus and its policy implications

This short paper presents results from an energy forecast model for the Republic of Cyprus. The model was used during years 2010–2011 to assist national authorities in their preparation of the National Renewable Energy Action Plan and the Energy Efficiency Action Plan of Cyprus in line with the requirements of European Union Directives. Major macroeconomic and price assumptions of the model are presented, and results from model application under different assumptions with regard to the evolution of energy efficiency are outlined and discussed. On the basis of these scenarios the paper discusses policy implications for Cyprus regarding the country's compliance with EU energy and climate legislation.

Monthly Energy Consumption Forecasting Using Wavelet Analysis and Radial Basis Function Neural Network

Monthly energy forecasts help heavy consumers of electric power to prepare adequate budget to pay their electricity bills and also draw the attention of management and stakeholders to electricity consumption levels so that energy efficiency measures are put in place to reduce cost. In this paper, a wavelet transform and radial basis function neural network based energy forecast model is developed to predict monthly energy consumption. The model was developed using the monthly energy consumption of Kwame Nkrumah University of Science and Technology (KNUST), Kumasi, Ghana for a 9-year period. A mean absolute percentage error of 7.94% was achieved when the forecast model was tested over a 60-month period.Keywords: Load forecasting, artificial neural network, radial basis function, wavelet transform

A non-linear multivariable regression model for midterm energy forecasting of power systems

The objective of this paper is to describe a non-linear multivariable regression method for midterm energy forecasting of power systems in annual time base. This method performs an extensive search in order to select the appropriate transformation functions of input variables, the weighting factors and the training periods to be used, by taking into consideration the correlation analysis of the selected input variables. With this procedure the best forecasting model is formed. Results are presented that are obtained applying the described method for the Greek power system and for different categories of low voltage customers. These results are also compared to those obtained from the application of standard regression methods.

Energy forecasting models, simulations and price sensitivity: New formulation

The top-down, two-stage optimization approach provides energy analysts with a powerful tool for both policy analysis and forecasting. This approach assumes that energy is weakly separable from other inputs to the production process. An aggregate energy Divisia Price Index (DPI) links the two stages. Although the DPI is necessary for the model estimation, it can distort the outcomes of price sensitivity analysis and forecasting exercises when the model is used for forecasting and simulations without adjusting the DPI. This deprives the two-stage optimization procedure of some of its virtues. The present study attempts to explain the reason for these distortions and to show that, by reformulating the same set of estimated coefficients, the simulation module will correct the shortcomings of the conventional formulation.

Two mathematical models for energy forecasting

Mathematical models are adopted in the present work for two types of energy fore-casting. The first type is the forecasting of displacement of one energy source by the other and the second one the forecasting of the future energy demand. Several examples of applications of those models are illustrated by using historical data collected from various sources. In many cases, forecasting by these models is quite acceptable, while in some other cases, trend projection by them is not. so satisfactory. This reveals the deficiency of the forecasting models currently available. Possibles improvement of these models for handling more complicated forecasting is suggested.

Weather load model for electric demand and energy forecasting

A method of forecasting the heat sensitive portion of electrical demand and energy utilizing a summer weather load model and taking into account probability variation of weather factors is discussed in this paper. The heat sensitive portion of the load is separated from base load and historical data is used to determine the effect of weather on the system load. This method is based on regression analysis of historical load and weather information and the establishment of system load characteristics based on historical or survey information. The method has been determined primarily for forecasting demands and energy for the intermediate range of from 3 to 10 years. However, it is applicable for monthly and annual peak forecasting, but probably not applicable for short terms such as hour to hour or day to day forecast. It may also be helpful in long term forecasting with appropriate forecast of future quantities of heat sensitive load on the system. Separate weather load models are used for determining the heat sensitive portion of electric energy and demand independently. These correlation studies were made when the author was located in Birmingham, Alabama and all of the studies reported here are on systems in that general area. The author hopes to continue these weather-load correlation studies in the northeast area in the near future.