Neural Network Computing for the Electric Power Industry

G. Load Forecasting

APPLICATION OF ARTIFICIAL NEURAL NETWORKS TO LOAD FORECASTING

Alain J. Germond      Nicolas Macabrey      ^* Thomas Baumann

Laboratoire de Réseaux d’Energie Electrique

Ecole Polytechnique Fédérale de Lausanne

CH-1015 Lausanne (Switzerland)

Abstract

After a brief review of neural network models and references to the application of supervised learning for short-term load forecasting, this paper analyses the application of Kohonen’s self-organizing feature map to short-term forecasting of peak electrical load. The Kohonen self-organizing feature map is used for the classification of electrical loads. The network not only learns similarities of load patterns in an unsupervised manner, but it uses the information stored in the ...