Prediction own energy consumption nuclear power plants using data mining methods
While in operation, nuclear power plant consumes a significant amount of electricity, the so’called consumption for own needs. The existing practice in the Russian Federation is that nuclear power plants must order in advance from the grid operator required volume of electricity so that any deviation of the actual energy consumption of the predicted value results in some penalties. It is quite difficult to predict required value of energy consumption in advance, so prediction of energy consumption with high accuracy using available operational information is very important.
This article discusses the use of various methods of data analysis and data mining to predict the energy consumption for own needs of nuclear power plant using actual operational data. Among these methods are smoothing by medians, exponential smoothing and support vector regression. First of all the simple method of prediction currently used at the NPP was considered. Then the comparison of proposed methods is performed, both among themselves and with the currently used one at the plant. The special emphasis is placed on support vector regression. As a final result, the method of forecasting of nuclear power plant energy consumption for own needs using support vector regression with significantly higher accuracy is proposed.
- Alfares H.K., Nazeeruddin M. Electric Load Forecasting: literature survey and classification оf methods. International Journal of Systems Science. 2002, v. 33, no. 1, pp. 23-34.
- Lukashin Yu.P. Adaptive Methods of Short-term Time Series Prediction – Moscow, Finansy i Statistika Publ., 2003, 416 p. (in Russian).
- Smola A.J., Schцlkopf B.A. Tutorial on Support Vector Regression. Statistics and computing. 2004, v. 14, no.3, pp. 199–222.
- Haykin S. Neural Networks: A Comprehensive Foundation. N.Y., Prentice Hall, 1999.
- Barsegyan A.A., Kupriyanov M.S., Stepanenko V.V., Holod I.I. Methods and Models of Data Analysis: OLAP and Data Mining. Sankt-Petersburg. BHV-Peterburg Publ., 2004, 336 p. (in Russian).
- Anderson T.W. The Statistical Analysis of Time Series. John Wiley & Sons, 1971.