System of automatic control of power usage based on a neural network

Authors

  • O. Zhyvoglyad O. Zhyvoglyad National technical university of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”

Keywords:

Control system, neuron network, NARX, Smart Grid, Levenberg-Marquardt optimization, Bayesian regulation

Abstract

Abstract — in this article is considered a problem of prediction amount generated energy from renewable sources with the help of a neural network. Our system can be part of some systems for tasks of managing distribution of energy. First our mission is to analyze information. We sorted all incoming data form renewable sources and divide them on four types each for the season (summer, autumn, winter, spring). The second one is that we predict future values using recurrent neural network called “Nonlinear autoregressive model process with exogenous input” (NARX) that trained on previous data from the sources and weather conditions that already been sorted by the season for more accurate values. In this paper, we will closely analyze this prediction model and give some modelling results. After we have predicted values we can analyze and compare it with estimate consumption level, then we can control distribution of electric energy over the grid by unplugging some not important systems. If we can`t equalize levels of regeneration and consumption we convert our system to general source of electric energy to buy some time to renewable source recharge the battery. The system mostly independent but it`s can be integrated in a big or small Smart Grid. Such systems often include renewable sources of energy and they need maintaining distribution. In addition, our system can save money just because the main priority of it is efficient use of electric energy. We all know that energy from renewable sources is cheaper than from general electric grid and system try to use it on 100%.

References

Eugen Diaconescu, The use of NARX Neural Networks to predict Chaotic Time Series, WSEAS TRANSACTIONS on COMPUTER RESEARCH, Issue 3, Volume 3, March 2008.

M. T. Hagan, O. D. Jesus, R. Schultz, Training Recurrent Networks for Filtering and Control, in (editors) L.R. Medsker, L.C. Jain, Recurrent Neural Networks – Design and Applications, CRC Press, 2001.

D.P. Mandic, J.A. Chambers, Recurrent Neural Networks for Prediction, JohnWiley&Sons, 2001.

Zina Boussaada, Octavian Curea, Ahmed Remaci, Haritza Camblong, Najiba Mrabet Bellaa, A Nonlinear Autoregressive Exogenous (NARX) Neural Network Model for the Prediction of theDaily Direct Solar Radiation, Energies 2018, 11, 620

H.B. Demuth, M. Beale, Users’ Guide for the Neural Network Toolbox for Matlab, The Mathworks, Natica, MA, 1998

Downloads

Published

2019-12-23

How to Cite

Zhyvoglyad, O. (2019). System of automatic control of power usage based on a neural network. Lighting Engineering & Power Engineering, 2(52), 68–71. Retrieved from https://lepe.kname.edu.ua/index.php/lepe/article/view/406

Issue

Vol. 2 No. 52 (2018): Lighting engineering and power engineering

Section

Power Engineering

License

The authors that are published in this journal agree with the following terms:

  • The authors reserve the right of authorship of his work and pass to the journal the right of first publication of this work is licensed under a Creative Commons Attribution License, which allows others to freely distribute published work with reference to authors of original works and works first published in this journal.
  • The authors have the right to enter into a separate additional agreement for non-exclusive distribution of work in the form in which it was published in the magazine (for example, to place work in electronic storage agencies or publish as part of the monograph), providing the reference to the first publication in this journal.
  • Journal policy allows and encourages the placement by the authors on the Internet (eg, in storage facilities or personal websites) the manuscript of the works before the submission of the manuscript to the editor as well as during its editorial processing, as it contributes to productive scientific discussion and has a positive impact on efficiency and dynamics citing published work (see. The Effect of Open Access).