Short time load forecasting based on simulated annealing and genetic algorithm improved SVM

A support vector machines method (SVM) is presented for the hourly load forecasting of the coming days. In this approach, improved SVM based on simulated annealing algorithm and genetic algorithm. The new approach is used for the next day load forecast. These two methods are combined to find the improved parameters for Support Vector Machine. It proves that the combined method is useful in improve the SVM method. The load forecast results of this method are compared with the current neural network load forecast program (a conventional time series package). Both programs are utilized to predict the hourly load of one day ahead. Based on simulation results, the improved SVM approach provides a better performance than the neural network and the regular SVM algorithm.

[1]  T. Chow,et al.  Nonlinear autoregressive integrated neural network model for short-term load forecasting , 1996 .

[2]  Robert E. Smith,et al.  A genetic algorithm based approach to thermal unit commitment of electric power systems , 1995 .

[3]  Bernhard Sendhoff,et al.  Variable encoding of modular neural networks for time series prediction , 1999, Proceedings of the 1999 Congress on Evolutionary Computation-CEC99 (Cat. No. 99TH8406).

[4]  Y. Shimakura,et al.  Short-term load forecasting using an artificial neural network , 1993, [1993] Proceedings of the Second International Forum on Applications of Neural Networks to Power Systems.

[5]  S. J. Kiartzis,et al.  A neural network short term load forecasting model for the Greek power system , 1996 .

[6]  T. Hesterberg,et al.  A regression-based approach to short-term system load forecasting , 1989, Conference Papers Power Industry Computer Application Conference.

[7]  Moon-Hee Park,et al.  Short-term Load Forecasting Using Artificial Neural Network , 1992 .

[8]  Sandra Paterlini,et al.  Evolutionary approaches for statistical modelling , 2002, Proceedings of the 2002 Congress on Evolutionary Computation. CEC'02 (Cat. No.02TH8600).

[9]  Saifur Rahman,et al.  A real-time short-term load forecasting system using functional link network , 1997 .

[10]  A. Conejo,et al.  A parallel repair genetic algorithm to solve the unit commitment problem , 2003, 2003 IEEE Power Engineering Society General Meeting (IEEE Cat. No.03CH37491).

[11]  C. Lucas,et al.  A new genetic algorithm with Lamarckian individual learning for generation scheduling , 2003 .

[12]  G. Purushothama,et al.  Simulated Annealing with Local Search: A Hybrid Algorithm for Unit Commitment , 2002, IEEE Power Engineering Review.

[13]  Francisco D. Galiana,et al.  Unit commitment by simulated annealing , 1990 .

[14]  Y. W. Wong,et al.  Short-term hydrothermal scheduling part. I. Simulated annealing approach , 1994 .

[15]  S. J. Kiartzis,et al.  Short-term load forecasting in an autonomous power system using artificial neural networks , 1997 .