Abstract
The accuracy of short-term electricity load forecasting is of great interest since it allows avoiding unexpected blackouts and lowering operating costs. In this paper, we aim to implement the artificial neural networks to model and forecast the half-hourly electric load demand in Tunisia over the period 2000–2008. To improve the quality of forecasts, the proposed artificial neural network model uses not only past electric load values as inputs, but also climatic and calendar variables. To determine the optimal structure of the neural network model, this paper employs the pattern search algorithm. Moreover, the neural network model is equipped with the Levenberg–Marquardt learning algorithm. Our findings confirm the performance of this algorithm to the view of evaluation indicators since the mean absolute percentage error values range between 1.1 and 3.4%. The analysis also shows the superiority of the Levenberg–Marquardt algorithm compared to the resilient back propagation algorithm and the conjugate gradient algorithm. In the light of the current research, we stress the aptness of the proposed artificial neural network model in forecasting short-term electricity demand.
Similar content being viewed by others
Notes
While other meteorological and climatic factors, such as snow, fog, humidity, and wind speed, might affect electricity demand, we do not introduce them as inputs for two reasons. First, data on such climatic variables do not exist in Tunisia. Second, and most important, several authors, such as [31,32,33] point out that temperature is the most important climatic determinant of electricity demand since it has a direct impact on it.
To conduct the study, we used the software Matlab R2013b. Particularly, two toolboxes have been used, namely the pattern search optimization toolbox and the neural network toolbox.
As mentioned earlier, there is a big similarity regarding the electric load during Tuesday, Wednesday and Thursday. Consequently, we focus only on Thursday in the rest of the paper.
Appendix A presents the formulas of the different measures employed in the analysis.
It is the fastest algorithm for the problem of model identification and function approximation. The memory space required for this algorithm is relatively small compared to other algorithms proposed by Matlab.
It is an iterative algorithm in a finite number of iterations. Its advantage in terms of computing time, due to a clever initialization (preconditioning), allows obtaining in only few steps close estimates. .
More details on the structure of the ANN using the pattern search optimization algorithm are displayed in Appendix B.
The correlation coefficient would be equal to one if the predicted values are equal the observed values. In this case, all the data points would fall on the fitted regression line.
References
Zhou, C., Chen, X.: China’s energy consumption prediction considering error correction based on decompose–ensemble method. Energy Syst. (2018). https://doi.org/10.1007/s12667-018-0300-1
Bunn, D., Farmer, E.: Economic and operational context of electric load prediction. In: Bunn, D., Farmer, E. (eds.) Comparative Models for Electrical Load Forecasting, pp. 3–11. Wiley, Hoboken (1985)
Ghalehkhondabi, I., Ardjmand, E., Weckman, G.R., Young II, W.A.: An overview of energy demand forecasting methods published in 2005–2015. Energy Syst. 8, 411–447 (2017)
Hong, W.C.: Modeling for energy demand forecasting. In: Hong, W.C. (ed.) Intelligent Energy Demand Forecasting, pp. 21–40. Springer, New York (2013)
Chow, T.W.S., Leung, C.T.: Neural networks based short term load forecasting using weather compensation. IEEE Trans. Power Syst. 11, 1736–1742 (1996)
Taylor, J.W., Buizza, R.: Using weather ensemble predictions in electricity demand forecasting. Int. J. Forecast. 19, 57–70 (2003)
Oliveira, M.O., Marzec, D.P., Bordin, G., Bretas, A.S., Bernaedon, D.: Climate change effect on very short-term electric load forecasting. Proceedings of the 2011 IEEE Trondheim PowerTech, pp. 944–950 (2011)
Hussain, A., Rahman, M., Memon, J.A.: Forecasting electricity consumption in Pakistan: the way forward. Energy Policy 90, 73–80 (2016)
Taylor, J.W.: Short-term electricity demand forecasting using double seasonal exponential smoothing. J. Oper. Res. Soc. 54, 799–805 (2003)
Mbamalu, G.A.N., El-Hawary, M.E.: Load forecasting via suboptimal seasonal autoregressive models and iteratively reweighted least squares estimation. IEEE Trans. Power Syst. 8, 343–348 (1993)
Clements, A.E., Hurn, A.S., Li, Z.: Forecasting day-ahead electricity load using a multiple equation time series approach. Eur. J. Oper. Res. 251, 522–530 (2016)
Weron, F.: Electricity price forecasting: a review of the state-of-the-art with a look into the future. Int. J. Forecast. 30, 1030–1081 (2014)
Hamzacebi, C.: Forecasting of Turkey’s net electricity energy consumption on sectoral bases. Energy Policy 35, 2009–2016 (2007)
Dutta, G., Jha, P., Laha, A.K., Mohan, N.: Artificial neural network models for forecasting stock price index in the Bombay Stock Exchange. J. Emerg. Mark. Finance 5, 283–295 (2006)
Guresen, E., Kayakutlu, G., Daim, T.U.: Using artificial neural Network models in stock market index prediction. Expert Syst. Appl. 38, 10389–10397 (2011)
Nag, A.K., Mitra, A.: Forecasting daily foreign exchange rates using genetically optimized neural networks. J. Forecast. 21, 501–511 (2002)
Panda, C., Narasimhan, V.: Forecasting exchange rate better with artificial neural network. J. Policy Model. 29, 227–236 (2007)
Sehgal, N., Pandey, K.K.: Artificial intelligence methods for oil price forecasting: a review and evaluation. Energy Syst. 6, 479–506 (2015)
Karimi, H., Dastranj, J.: Artificial neural network-based genetic algorithm to predict natural gas consumption. Energy Syst. 5, 571–581 (2014)
Hsu, C.C., Chen, C.Y.: Regional load forecasting in Taiwan-Applications of artificial neural networks. Energy Convers. Manage. 44, 1941–1949 (2003)
Mandal, P., Senjyu, T., Urasaki, N., Funabashi, T.: A neural network based several hour-ahead electric load forecasting using similar days approach. Int. J. Electr. Power Energy Syst. 28, 367–373 (2006)
Kandananond, K.: Forecasting Electricity Demand in Thailand with an Artificial Neural Network Approach. Energies 4, 1246–1257 (2011)
Gürbüz, F., Öztürk, C., Pardalos, P.: Prediction of electricity energy consumption of Turkey via artificial bee colony: a case study. Energy Syst 4, 289–300 (2013)
Panapakidis, I.P.: Application of hybrid computational intelligence models in short-term bus load forecasting. Expert Syst. Appl. 54, 105–120 (2016)
Gonzalez-Romera, E., Jaramillo-Moran, M.A., Carmona-Fernandez, D.: Monthly electric energy demand forecasting based on trend extraction. IEEE Trans. Power Syst. 21, 1946–1953 (2006)
Bakirtzis, A., Petridis, V., Klartzis, S., Alexiadis, M., Maissis, A.: A neural network short-term load forecasting model for the Greek power system. IEEE Trans. Power Syst. 11, 858–863 (1996)
Park, D.C., El-Sharkawi, M.A., Marks, R.J., Atlas, L.E., Damborg, M.J.: Electric load forecasting using an artificial neural network. IEEE Trans. Power Syst. 6, 442–449 (1991)
Darbellay, G.A., Slama, M.: Forecasting the short-term demand for electricity—do neural networks stand a better chance? Int. J. Forecast. 16, 71–83 (2000)
Khotanzad, A., Afkhami-Rohani, R., Lu, T.L., Abaye, A., Davis, M., Maratukulam, D.J.: ANNSTLF—a neural-network-based electric load forecasting system. IEEE Trans. Neural Netw. 8, 835–846 (1997)
Hippert, H.S., Bunn, D.W., Souza, R.C.: Large neural networks for electricity load forecasting: are they overfitted? Int. J. Forecast. 21, 425–434 (2005)
Jain, A., Satish, B.: Clustering based short term load forecasting using artificial neural network. IEEE/PES Power Syst. Conf. Expos. 10, 1109 (2009). https://doi.org/10.1109/PSCE.2009.4840241
Al-Zayer, J., Al-Ibrahim, A.A.: Modeling the impact of temperature on electricity consumption in the Eastern Province of Saudi Arabia. J. Forecast. 15, 97–106 (1996)
Liu, J.M., Chen, R., Liu, L.M., Harris, J.L.: A semi-parametric time series approach in modeling hourly electricity loads. J. Forecast. 25, 537–559 (2006)
Madić, J.M., Radovanović, M.R.: Optimal selection of ANN training and architectural parameters using taguchi method: a case study. FME Trans. 39, 79–86 (2011)
Beale, M.H., Hagan, M.T., Demuth, H.B.: Neural Network Toolbox User’s Guide. The Mathworks Inc, Massachusetts (2015)
Amjady, N., Keynia, F.: A new neural network approach to short term load forecasting of electrical power systems. Energies 4, 488–503 (2011)
Tanoto, Y., Ongsakul, W., Marpaung, C.O.P.: Levenberg–Marquardt recurrent networks for long term electricity peak load forecasting. Telkomnika 9, 257–266 (2011)
Rodrigues, F., Cardeira, C., Calado, J.M.F.: The daily and hourly energy consumption and load forecasting using artificial neural network method: a case study using a set of 93 households in Portugal. Energy Procedia 62, 220–229 (2014)
Acknowledgments
The authors are grateful to the Editor-in-Chief, Professor Q.P. Zheng, and two anonymous referees for their constructive comments on earlier versions of the manuscript. They also acknowledge the Tunisian Company of Electricity and Gas for providing data used in this research.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Appendix A: Forecast performance measures
Measure | Formula |
---|---|
Mean absolute error (MAE) | \( MAE = \frac{1}{n}\mathop \sum \limits_{t = 1}^{n} \left| {y_{t} - \hat{y}_{t} } \right| \) |
Mean squared error (MSE) | \( MSE = \frac{1}{n}\mathop \sum \limits_{t = 1}^{n} (y_{t} - \hat{y}_{t} )^{2} \) |
Root mean squared error (RMSE) | \( RMSE = \sqrt {\frac{1}{n}\mathop \sum \limits_{t = 1}^{n} (y_{t} - \hat{y}_{t} )^{2} } \) |
Mean percentage error (MPE) | \( MPE = \frac{1}{n}\mathop \sum \limits_{t = 1}^{n} \frac{{y_{t} - \hat{y}_{t} }}{{y_{t} }}{\text{x}}100 \) |
Mean absolute percentage error (MAPE) | \( MAPE = \frac{1}{n}\mathop \sum \limits_{t = 1}^{n} \frac{{\left| {y_{t} - \hat{y}_{t} } \right|}}{{y_{t} }}{\text{x}}100 \) |
Correlation coefficient (R) | \( R = \frac{{\mathop \sum \nolimits_{t = 1}^{n} (y_{t} - \bar{y}_{t} )(\hat{y}_{t} - \hat{y})}}{{\sqrt {\mathop \sum \nolimits_{t = 1}^{n} (y_{t} - \bar{y}_{t} )^{2} \times \mathop \sum \nolimits_{t = 1}^{n} (\hat{y}_{t} - \hat{y}_{t} )^{2} } }} \) |
Appendix B: The Structure of the ANN using the pattern search optimization algorithm
2.1 B1: The Levenberg–Marquardt algorithm (MLP Lm)
The optimal structure of the neural network is shown in the following Matlab graph:
See Fig. 5.
The network is composed of 58 input neurons and 48 output neurons representing the half-hour loads and three hidden layers. The first layer comprises 18 neurons; the second layer comprises 35 neurons while the third layer 12 neurons. Each hidden layer is provided with a tangent sigmoid transfer function (tansig) and a linear function (purlin) for the output layer.
2.2 B2: The resilient back-propagation algorithm (MLP rp)
The optimal structure of the neural network is shown in the following Matlab graph:
See Fig. 6.
As for the Levenberg–Marquardt algorithm, the network is composed of 58 input neurons and 48 output neurons. The first layer comprises 21 neurons; the second layer comprises 20 neurons while the third layer 20 neurons. Each hidden layer is provided with a tangent sigmoid transfer function (tansig) and a linear function (purlin) for the output layer.
2.3 B3: The conjugate gradient algorithm (MLP cgb)
The optimal structure of the neural network is shown in the following Matlab graph:
See Fig. 7.
As for the previous algorithms, the network is composed of 58 input neurons and 48 output neurons. The first layer comprises 28 neurons; the second layer comprises 5 neurons while the third layer 18 neurons. Each hidden layer is provided with a tangent sigmoid transfer function (tansig) and a linear function (purlin) for the output layer.
The different parameters of the three algorithms can be summarized in the following table:
See Table 5.
Appendix C: Plots of observed and predicted values for training, validation, testing, and overall datasets
Rights and permissions
About this article
Cite this article
Houimli, R., Zmami, M. & Ben-Salha, O. Short-term electric load forecasting in Tunisia using artificial neural networks. Energy Syst 11, 357–375 (2020). https://doi.org/10.1007/s12667-019-00324-4
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12667-019-00324-4