Comparison of two neural network optimization approaches

This paper compares two optimization methods for training Neural Networks: the typical supervised feed-forward hackpropagation algorithm and an improved Particle Swarm Optimization method. The aim is to highlight advantages and drawbacks of these techniques in order to suitably apply them to electromagnetic problems. Some numerical results and comparisons are presented analyzing a load forecasting problem. Neural Networks are trained for a particular power system load consuption signal, for future time prediction.

[1]  M.T. Hagan,et al.  Global optimization of neural network weights , 2002, Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290).

[2]  Carlos E. Pedreira,et al.  Neural networks for short-term load forecasting: a review and evaluation , 2001 .

[3]  Y. Rahmat-Samii,et al.  Particle swarm optimization in electromagnetics , 2004, IEEE Transactions on Antennas and Propagation.

[4]  T. Funabashi,et al.  One-Hour-Ahead Load Forecasting Using Neural Networks , 2002 .

[5]  W. Charytoniuk,et al.  Very short-term load forecasting using artificial neural networks , 2000 .