Application of Short Term Load Forecasting using Support Vector Machines in RapidMiner 5.0 (CROSBI ID 575251)
Prilog sa skupa u zborniku | izvorni znanstveni rad | međunarodna recenzija
Podaci o odgovornosti
Matijaš, Marin ; Lipić, Tomislav
engleski
Application of Short Term Load Forecasting using Support Vector Machines in RapidMiner 5.0
This paper presents an application of RapidMiner 5.0 in Load Forecasting using Support Vector Machines. In Power Systems, Load Forecasting is an important task since equilibrium of production and consumption in electricity grid must be maintained at all times. Short Term Load Forecasting(STLF) is forecasting of load up to 72 hours in the future. Method of Support Vector Machines (SVM) has been used for Load Forecasting for its good generalization properties which lead to low Mean Absolute Percentage Error (MAPE). Besides its importance in power system planning, differences in MAPE greater than 0, 5% have significant economic effects in liberalized electricity markets for producers and suppliers. The paper presents application of SVM on high-voltage customers’ consumption in Croatian electricity grid in RapidMiner 5.0. SVM is used as a learning algorithm and its parameters are optimized through Grid Parameter Optimization. Mentioned combination of fast Support Vector Machines with good generalization properties and Grid Parameter Optimization, that optimizes a model, is chosen because it provides a solution for STLF with lower MAPE than previously used similar day technique. First section in this paper gives an overview of the field of application. In second section a description of the algorithm is provided. Third section presents the process in RapidMiner which is followed by results and a conclusion.
short-term load forecasting; support vector machine; rapidminer
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
Podaci o prilogu
45-48.
2010.
objavljeno
Podaci o matičnoj publikaciji
Proceedings of RapidMiner Community Meeting and Conference
Podaci o skupu
RapidMiner Community Meeting and Conference - RCOMM2010
predavanje
13.09.2010-16.09.2010
Dortmund, Njemačka