文档介绍：Proceedings of the 1999 IEEE Canadian Conference on Electrical puter Engineering
Shaw Conference Center, Edmonton, Alberta, Canada May 9-12 1999
Wavelet work Based Short Term Load
Forecasting of Electric Power mercial Load
Anant Oonsivilai and El-Hawary,.
******@is2 . -******@
Department of Electrical & Computer Engineering
DalTech(TUNS) ,Dalhousie University
Halifax, Nova Scotia, Canada B3 J 2x4
Abstract- In an electric power system, the system load term load forecasting [2-1,6-10] have provided
consists of domestic, commercial, industrial and encouraging results.
municipal load sectors. This paper presents an ANNs have emerged as a popular expert system learning
approach for predicting electric power system tool for modeling, classification, and approximating
commercial load using a wavelet work. continuous functions. In NN models, many kind of basis
Morlet and Mexican hat wavelets are used to generate functions have been used such as splines, polynomial,
the transfer functions of hidden layer nodes of the radial basis, linear, hard limit, log-sigmoid, and tangent
sigmoid transfer functions. Recently, works,
work. A wavelet work is trained for
or wavelet works, which consist of
a particular power system load. Results show that
wavelets as the transfer function in the hidden layer, have