The ANN of UMCP forecast based on developed ICA

This paper focuses on the relatively new application of independent component analysis (ICA) on the forecast problem of unconstrained market clearing price (UMCP) in the day-ahead spot market. The property extraction of UMCP and UMCP forecast model based improved ICA are presented in order to not only decrease the feature dimensions and the complexity of model, but also enhance the model practicability and forecast accuracy. Firstly, the whitened factor data as mixed input signals is extracted by improved fixed-point algorithm. Then artificial neural network (ANN) forecast model is built on the basis of the extracted feature samples and used to forecast UMCP. The UMCP data of America California during 1998 and 1999 is also applied to the algorithm of this paper, whose result has verified the validity of the model.

[1]  Aapo Hyvärinen,et al.  Fast and robust fixed-point algorithms for independent component analysis , 1999, IEEE Trans. Neural Networks.

[2]  Scott C. Douglas,et al.  Mean square error analyses of adaptive blind source separation algorithms , 2001, Neural Networks for Signal Processing XI: Proceedings of the 2001 IEEE Signal Processing Society Workshop (IEEE Cat. No.01TH8584).

[3]  Francesco Carlo Morabito.  Independent component analysis and neural approaches to the extraction of features from NDT/NDE , 1999, IJCNN'99. International Joint Conference on Neural Networks. Proceedings (Cat. No.99CH36339).

[4]  Ewald F. Fuchs,et al.  Influence of harmonics on power distribution system protection , 1988 .

[5]  Asoke K. Nandi,et al.  Blind Source Separation , 1999 .