Title :
Application of two-dimensional support vector machine in short-term Load forecasting
Author :
Yang, Jingfei ; Stenzel, Jürgen
Author_Institution :
Darmstadt Univ. of Technol., Darmstadt
Abstract :
In this paper, the short-term load forecasting for the power system with heavy impulse loads is explored. In order to eliminate the effect of random startup and shutdown of high- power motors to the load prediction, support vector machine is applied to smooth the load curve and get the essential load. Second order derivative method is employed to find outliers and replace them with a reasonable value. With the essential load, support vector machine is applied again to train the data and predict the future load. The effectiveness of the proposed method is demonstrated by its application to a practical power system.
Keywords :
load forecasting; power engineering computing; power systems; support vector machines; high- power motors; power system; second order derivative method; short-term load forecasting; two-dimensional support vector machine; Load forecasting; Metals industry; Power industry; Power system modeling; Power system security; Power systems; Predictive models; Smoothing methods; Support vector machines; Training data; impulse load; outlier; short-term load forecasting; smoothing; support vector machine;
Conference_Titel :
Power Tech, 2005 IEEE Russia
Conference_Location :
St. Petersburg
Print_ISBN :
978-5-93208-034-4
Electronic_ISBN :
978-5-93208-034-4
DOI :
10.1109/PTC.2005.4524382
