ISSN 2071-8594

Russian academy of sciences

Editor-in-Chief

Gennady Osipov

R.F. Gibadullin, D.V. Lekomtsev, M.Y.Perukhin Neural Network Data Processing for Analysis of the Industrial Networks Parameters

Abstract.

We used artificial neural networks and diagnostic network information to assess the condition of PROFINET (Process Field Network). An artificial neural network determines whether the network works fine or not. An important part of this work is data preprocessing. An essential part of the work is data preprocessing. It is done using quantization, data aligning, reducing the number of inputs and other preprocessing techniques to create a new version of the dataset to improve the accuracy. The obtained data makes possible to do a number of experiments and to find out what approach of data preprocessing shows the best results. The results were evaluated on two datasets. The first dataset contains diagnostic data of a well-functioning network, and the second one consists of data in which network problems were detected. The highest accuracy obtained in this work is 98.91% of recognizing problems in the network and the accuracy of 87.70% when the network is working fine. The work also opens up opportunities to improve accuracy in the future.

Keywords:

PROFINET, Industrial Ethernet, Network Diagnostics, Artificial Neural Networks, Machine Learning, Data Preprocessing.

PP. 80-87.

DOI 10.14357/20718594200108

References

1. What is the difference between Ethernet and Industrial Ethernet. Available at: http://www.innovasic.com/news/industrial-ethernet/whatis-the-difference-between-ethernet-and-industrial-ethernet/ (accessed April 19, 2019).
2. The Industrial Ethernet Book. Available at: http://www.iebmedia.com/index.php?id=12993&parentid=74&showdetail=true (accessed April 19, 2019).
3. Sestito, G., A. Turcato, A. Dias, M. Rocha, M. Silva, P. Ferrari, and D. Brandao. 2018. A Method for Anomalies Detection in Real-Time Ethernet Data Traffic Applied to PROFINET [Transactions on Industrial Informatics]. 14(4):2171-2180. doi: 10.1109/TII.2017.2772082
4. Shuan, L., T. Fei, S. King, G. Xiaoning, and L. Mein. 2016. Network Equipment Failure Prediction with Big Data Analytics [International Journal of Advances in Soft Computing & Its Applications]. 8(3).
5. Mossin, A., D. Brandao, G. Sestito, and R. Torres. 2016. Automatic Diagnosis for Profibus Networks [Journal of Control, Automation and Electrical Systems]. 27(6):658-669.
6. Wang, Z., M. Zhang, D. Wang, C. Song, M. Liu, J. Li, L. Lou, and Z. Liu. 2017. Failure prediction using machine learning and time series in optical network [Opt. Express]. 25(16):18553-18565.
7. Sestito, G., E. Mossin, D. Brandão, and A. Dias. 2014. Artificial neural networks and signal clipping for Profibus DP diagnostics [12th IEEE International Conference on Industrial Informatics (INDIN)]. Brazil. 242-247. doi: 10.1109/INDIN.2014.6945515
8. PROFINET Diagnosis – PROFINET-INspektor® NT – Indu-Sol. Available at: https://www.indusol.com/en/products/profinet/diagnostics/profinetinspektorr-nt/ (accessed April 19, 2019).
9. Chollet, F. (2015) keras, GitHub. Available at: https://github.com/fchollet/keras (accessed April 19, 2019).
10. Bergstra, J., O. Breuleux, F. Bastien, P. Lamblin, R. Pascanu, G. Desjardins, J. Turian, D. Warde-Farley, and Y. Bengio. 2010. Theano: a CPU and GPU Math Expression Compiler in Python [Proceedings of the Python for Scientific Computing Conference (SciPy)], Austin. 3-10.