In this study, gross domestic product, the gross national product, index of manufacturing production, number of population, number of households, oil price, fuel consumption, average ambient temperature and electricity consumption are used as inputs in neural networks to predict peak electric power demand. Three-layered feed-forward neural networks trained with a fast non-iterative training algorithm were used in this study for this long-term forecasting. Neural networks were optimized by removing redundant inputs. The results show that the peak electric power demand for 2003, 2005, 2010, 2015 and 2020 are 19,067 MW, 22,577 MW, 31,527 MW, 36,753 MW and 38,212 MW, respectively. Neural network can give the accurate estimations as indicated by the performance measures namely, root mean square error, coefficient of variation, mean bias error.
Keywords. neural network; forecasting; economic factor; electric power demand; weather climate; electricity consumption
References
N. Hubele, C. Cheng, 1989. Identification of Seasonal Short-Term Load Forecasting Models Using Statistical Decision Functions. IEEE Power Engineering Society.
S. Rahman, A. Bhatnagar, 1970. An Expert System Based Algorithm for Short-Term Load Forecasting, Part 1: Forecasting Modeling. IEEE Trans Power Appar Syst, PAS. Vol 89. No 7. pp1678-82.
J. Taweekun, A. Rajapakse, S. Chirarattananon, 2002. A Study on Use of Neural Networks and DOE-2 Program for Predicting Cooling Load and Electrical Demand in Buildings. 1st International Conference on Sustainable Energy Technologies, Portugal. ppBE20/53-BE25/53.
D.C. Park, M.A. El-Sharkawi, R.J. Marks, L.E. Atlas, M.J. Damborg, 1991. Electric Load Forecasting Using an Artificial Neural Network. IEEE Trans Power Syst. Vol 6. No 2. pp442-449.
C.C. Hsu, C.Y. Chen, 2003. Regional Load Forecasting in Taiwan-Applications of Artificial Neural Networks. Energy Conversion and Management. Vol 44. pp1941-1949.
B. Kermanshahi, H. Iwamiya, 2002. Up to Year 2020 Load Forcasting Using Neural Nets, Electrical Power and Energy System. Vol 24. pp789-799.
A. Rajapakse, K. Furuta, S. Kondo, 1999. Identification of Nonlinear Dynamic Models with Partially Connected Neural Networks Trained Using Orthogonal Least Square Estimation. Trans. IEE of Japan. Vol 119. No 3.
H.A. Braake, H.J. Can, G. Straten, H.B. Verbruggen, 1995. Two Step Approach in Training of Regulated Activation Weights Neural Networks, Internal Report R95.043, Dept. of Electrical Engineering, Delft Univ. of Technology.
J.F. Kreider, & J. Haberl, 1994. Predicting Hourly Building Energy Usage: The Great Predictor Shootout-Overview and Discussion of Results, ASHRAE Transactions. Vol 100. No 2. pp1104-1116.
How to cite this article
Taweekun, J. (2005). 'Load Forecasting in Thailand Using Neural Networks'. SET Proceedings, 4(1), pp. 620-624. https://doi.org/10.65582/set.2005.135
