Forecasting electric consumption of the enterprise using artificial neural networks

About the authors:

Sameh A. Kassem, Postgraduate Student, Institute of Digital Technologies and Economics, Kazan State Power Engineering University; ali20105@mail.ru; ORCID: 0000-0002-4587-6730

Abdulla H. Ebrahim, Cand. Sci. (Phys.-Math.), Junior Researcher, Memristive Materials Laboratory, Center for Nature-Inspired Engineering, University of Tyumen, Tyumen, Russia; abdulla.ybragim@mail.ru, https://orcid.org/0000-0002-1709-9882

Abdulla M. Khasan, Postgraduate Student, Institute of Information Technologies and Intelligent Systems, Kazan Federal University; alii.hasaa@gmail.com; ORCID: 0000-0002-1988-8869

Alla G. Logacheva, Cand. Sci. (Tech.), Department of Power Supply of Industrial Enterprises, Institute of Electrical Power Engineering and Electronics, Kazan State Power Engineering University; logacheva.alla@yandex.ru; ORCID: 0000-0002-0371-7985

Abstract:

Energy consumption has increased dramatically over the past century due to many factors, including both technological, social and economic factors. Therefore, predicting energy consumption is of great importance for many parameters, including planning, management, optimization and conservation. Data-driven models for predicting energy consumption have grown significantly over the past several decades due to their improved performance, reliability, and ease of deployment. Artificial neural networks are among the most popular data-driven approaches among the many different types of models today.

This article discusses the possibility of using artificial neural networks for medium-term forecasting of the power consumption of an enterprise. The task of constructing an artificial neural network using a feedback algorithm for training a network based on the Matlab mathematical package has been implemented.

The authors have analyzed such characteristics as parameter setting, implementation complexity, learning rate, convergence of the result, forecasting accuracy, and stability. The results obtained led to the conclusion that the feedback algorithm is well suited for medium-term forecasting of power consumption.

References:

1. Voronov I. V. 2009. “Determination of parameters affecting the power consumption of an industrial enterprise using the method of expert assessments”. Herald of the Kuzbass State Technical University, no. 5, pp. 61-64. [In Russian]

2. Voronovsky G. K., Makhotilo K. V., Petrashev S. N., Sergeev S.A. 1997. Genetic Algorithms, Artificial Neural Networks, and Virtual Reality Problems. Kharkiv: OSNOVA. 107 pp. [In Russian]

3. Galushka V. V. 2013. “Formation of a training sample when using artificial neural networks in problems of finding errors in databases”. Engineering Herald of the Don, no. 2. http://www.ivdon.ru/magazine/archive/n2y2013/1597 [In Russian]

4. Golovko V. A. 2001. Neural Networks: Learning, Organization and Application. 4^th edition. Moscow: IPRZhR. 256 pp. [In Russian]

5. Kirpichnikova I. M., Solomakho K. L. 2014. “Research of methods for forecasting power consumption of a sales enterprise”. Electrotechnical Systems and Complexes, no. 3 (24), pp. 39-43. [In Russian]

6. Lisichkin V. A., Lisichkina M. V. 2007. Strategic Management: educational materials. Moscow: EAOI. 329 pp. [In Russian]

7. Manusov V. Z., Biryukov E. V. 2006. “Short-term prediction of electrical load based on a fuzzy neural network and its comparison with other methods”. Bulletin of the Tomsk Polytechnic University, vol. 309, no. 6, pp. 153-158. [In Russian]

8. Matyushkin I. 2011. Modeling and Visualization by Means of MATLAB Physics of Nanostructures. Moscow: Tekhnosfera. 188 pp. [In Russian].

9. Osipov G. V. 2014. Modeling Social Phenomena and Processes Using Mathematical Methods: a tutorial. Moscow.: NORMA: INFRA-M, 191 pp. [In Russian]

10. Fedosin S. A., Plotnikova N. P., Nemchinova E. A., Makarova N.V. 2018. “Features of teaching the construction of models for the movement of complex objects with artificial intelligence based on a neural network”. Educational Technologies and Society, vol. 21, no. 3, pp. 290-297. [In Russian]

11. Haykin S. 2008. Neural Networks: Complete Course. Translated from English by N. N. Kussul and A. Yu. Shelestova. 2^nd edition, revised. Moscow: Williams Publishing House. 1103 pp. [In Russian]

12. Chernykh I. V. 2011. Modeling of electrical devices in MATLAB, SimPowerSystems, and Simulink. Moscow: DMK Press. 288 pp. [In Russian]

13. Lean Y. U., Shouyang W. A., Kin K. L. 2007. Basic Learning Principles of Artificial Neural Networks. Boston, MA: Springer. 323 pp.

14. Rumelhart D. E., Hinton G. E., Williams R. J. 1986. “Learning internal representations by error propagation”. In: Parallel Distributed Processing: Explorations in the Microstructure of Cognition. Vol. 1: Foundations, pp. 318-62. Cambridge, MA: A Bradford Book.

15. Kathirvalavakumar T, Jeyaseeli Subavathi S. 2009. “Neighborhood based modified backpropagation algorithm using adaptive learning parameters for training feedforward neural networks”. Neurocomputing, vol. 72, pp. 3915-3921. DOI: 10.1016/j.neucom.2009.04.010