Diagnostics of analog systems using Artificial Neural Networks

eng Artykuł w języku angielskim DOI: 10.14313/PAR_226/23

wyślij Piotr Bilski Politechnika Warszawska, Wydział Elektroniki i Technik Informacyjnych

Pobierz Artykuł

Abstract

The paper presents the diagnostic applications of artificial neural networks (ANN). Aims and problems present in the contemporary diagnostics are introduced. The structure of the artificial intelligence-based system is presented and discussed in detail. Various approaches to design the on-line fault detection and location system using artificial intelligence approaches are introduced. The generic architecture of the ANN and its variations are presented. Next, their diagnostic applications, advantages and drawbacks are discussed. Application of RBF ANN-based diagnostic module to detect and identify faults of the 5th order lowpass filter is presented. Finally, usability and limitations of the ANN-based diagnostic system are provided.

Keywords

analog systems diagnostics, artificial intelligence, artificial neural networks, fault detection

Diagnostyka systemów analogowych z wykorzystaniem sztucznych sieci neuronowych

Streszczenie

W artykule przedstawiono zastosowania sztucznych sieci neuronowych w diagnostyce systemów analogowych. Opisano główne cele diagnostyki oraz problemy spotykane obecnie podczas detekcji i lokalizacji uszkodzeń. Wprowadzono ogólną strukturę systemu diagnostycznego opartego na metodach sztucznej inteligencji. Przedstawiono różne metody inteligentne, które mogą zostać zastosowane w systemie działającym w trybie on-line. Następnie omówiono ogólną architekturę sztucznej sieci neuronowej oraz jej cechy szczególnie istotne z punktu widzenia detekcji i lokalizacji uszkodzeń. Specyficzne architektury sieci wraz z ich zastosowaniami diagnostycznymi przedstawiono w szczegółach. Na przykładzie filtru dolnoprzepustowego 5. rzędu przedstawiono działanie metody diagnostycznej wykorzystującej sieć neuronową typu RBF. Omówiono możliwości i ograniczenia stosowalności sztucznych sieci neuronowych jako narzędzia diagnostycznego. 

Słowa kluczowe

diagnostyka systemów analogowych, sztuczna inteligencja, sztuczne sieci neuronowe, wykrywanie uszkodzeń

Bibliografia

  1. Bilski P., Artificial intelligence methods in the diagnostics of analog systems, Oficyna Wydawnicza Politechniki Warszawskiej, Warszawa 2013.
  2. Stenbakken G.N., Souders T.M., Stewart G.W., Ambiguity groups and testability, IEEE Trans. Instr. and Meas., Vol. 38, Iss. 5, 1989, 941–947.
  3. Grzechca D., Golonek T., Rutkowski J., Analog Fault AC Dictionary Creation – The Fuzzy Set Approach, Proc. ISCAS 2006, 5744–5747.
  4. Santhosh P.M., Kumar M., Thangamani I., Mukhopadhyay D., Verma V., Rao V.V.S.S., Vaze K.K., Ghosh A.K., Neural network based diagnostic system for accident management in nuclear power plants, Proc. 2nd International Conference on Reliability, Safety and Hazard (ICRESH), Mumbai, India, 14–16 Dec. 2010, DOI: 10.1109/ICRESH.2010.5779613.
  5. Kim K., Bartlett E.B., Nuclear power plant fault diagnosis using neural networks with error estimation by series association, IEEE Trans. on Nuclear Science, 1996, Vol. 43, No. 4, 2373–2388, DOI: 10.1109/23.531786.
  6. Kulkarni M., Abou S.C., Stachowicz M., Fault Detection in Hydraulic System Using Fuzzy Logic, Proc. World Congress on Engineering and Computer Science (WCECS) 2009, Oct. 20–22, San Francisco, USA.
  7. Garza Castanon L.E., Nieto Gonzalez J.P., Garza Castanon M.A., Morales-Menendez R., Fault Diagnosis of Industrial Systems with Bayesian Networks and Neural Networks, “LNCS”, MICAI 2008: Advances in Artificial Intelligence, Oct. 27–31, Mexico, 2008, 998–1008.
  8. Marwala T., Mahola U., Nelwamondo F.V., Hidden Markov Models and Gaussian Mixture Models for Bearing Fault Detection Using Fractals, Proc. International Joint Conference on Neural Networks, July 16–21, 2006, Vancouver, Canada, 3237–3242.
  9. Engelbrecht A.P., Computational Intelligence: An Introduction, John Wiley & Sons, 2007.
  10. Khomfoi S., Tolbert L.M., Fault diagnosis system for a multilevel inverter using a neural network, Proc. 31st Annual Conference of IEEE Industrial Electronics Society, Raleigh, NC, USA, 6–10 Nov., 2005.
  11. Sahin S., Becerikli Y., Yazici S., Neural Network Implementation in Hardware Using FPGAs, “LNCS”, 4234, ICONIP 2006, Part III, 2006, 1105–1112.
  12. Maki Y. Loparo K.A., A Neural-Network Approach to Fault Detection and Diagnosis in Industrial Processes, “IEEE Trans. Control Systems Techn.”, Vol. 5, No. 6, Nov 1997, 529–541.
  13. Ahmed R.M., El Sayed M.A., Gadsden S.A., Habibi S.R., Fault Detection of an Engine Using a Neural Network Trained by the Smooth Variable Structure Filter, Proc. IEEE Int. Conf. Control Applications (CCA), Denver, CO, USA. September 28–30, 2011, 1190–1196.
  14. Bernieri A., Betta G., Pietrosanto A., Sanson C., A Neural Network Approach to Instrument Fault Detection and Isolation, “IEEE Trans. Instr. Meas.”, Vol. 44, No. 3, 1995, 747–750.
  15. Yang M.-S., Yang J.-H., On Parameter Estimation of Control Chart Patterns Using RBF Neural Network, 4th IEEE Conference on Industrial Electronics and Applications, ICIEA, 25–27 May 2009, Xi’an, China, 1498–1502.
  16. Cui L., Wang C., Yang B., Application of RBF Neural Network Improved by PSO Algorithm in Fault Diagnosis, “Journal of Theoretical and Applied Information Technology”, Vol. 46, No. 1, 2012, 268–273.
  17. Połok B., Bilski P., Optimization of the neural RBF classifier for the diagnostics of electronic circuit, 15th IMEKO TC10 Workshop, 6–7 July, 2017, Budapest, Hungary, 121–126.
  18. Drewnik M., Pasternak-Winiarski Z., SVM Kernel Configuration and Optimization for the Handwritten Digit Recognition, Proc. CISIM 2017, 16–18 June 2017, Białystok, Poland, 87–98.
  19. B. Long M. Li, H. Wang, S. Tian, Diagnostics of Analog Circuits Based on LS-SVM Using Time-Domain Features, “Circuits, Systems, and Signal Processing”, Dec. 2013, Vol. 32, No. 6, 2683–2706.
  20. Kurek J., Osowski S., Support vector machine for fault diagnosis of the broken rotor bars of squirrel-cage induction motor, “Neural Computing and Applications”, Vol. 19, Iss. 4, 2010, 557–564, DOI: 10.1007/s00521-009-0316-5
  21. Deák K., Kocsis I., Vámosi A., Keviczki Z., Failure Diagnostics with SVM in Machine Maintenance Engineering, “Annals of the Oradea University”, No. 1, 2014, 19–24.
  22. Bilski P., Automated selection of kernel parameters in diagnostics of analog systems, “Przegląd Elektrotechniczny”, No. 5, 2011, 9-13.
  23. Dong M., Cheang T., Chan S., On-Line Fast Motor Fault Diagnostics Based on Fuzzy Neural Networks, “Tsinghua Science and Technology”, Vol. 14, No. 2, 2009, 225–233.
  24. Calado J.M.F., Sa da Costa J.M.G., A Hierarchical Fuzzy Neural Network Approach for Multiple Fault Diagnosis, UKACC International Conference on CONTROL ‘98, 1–4 September 1998, 1498–1503.
  25. Bilski P., Ambiguity groups detection in analog systems diagnostics using Self-Organizing Maps, Proc. IMEKO TC10 Workshop, June 27–28 2016, Milan, Italy, 294–299.
  26. Bilski P., Unsupervised learning-based hierarchical diagnostics of analog circuits, 15th IMEKO TC10 Workshop, 6-7 July, 2017, Budapest, Hungary, 99–104.
  27. Mirea L., Ron J. Patto, Recurrent Wavelet Neural Networks Applied to Fault Diagnosis, 16th Mediterranean Conference on Control and Automation Congress Centre, Ajaccio, France June 25–27, 2008, 1774–1779.
  28. Xuhong W., Yigang H., Diagonal Recurrent Neural Network Based On-line Stator Winding Turn Fault Detection for Induction Motors, Proc. Eighth International Conference on Electrical Machines and Systems, 27–29 Sept. 2005, Nanjing, China, 2266–2269.
  29. Xu X., Chen R., Recurrent Neural Network Based On-line Fault Diagnosis Approach for Power Electronic Devices, Third International Conference on Natural Computation (ICNC 2007).
  30. Patan K., Obuchowicz A., Korbicz J., Cascade network of dynamic neurons in fault detection systems, European Control Conference (ECC), 1999, 4232–4237.
  31. Huang Y.-C., Yani H.-T., Huang C.-L., A New Intelligent Hierarchical Fault Diagnosis System, IEEE Transactions on Power Systems, Vol. 12, No. 1, 1997, 349–356.
  32. Dai H., He J., A Novel Fault Diagnosis Method for Rolling Element Bearings Using Kernel Independent Component Analysis and Genetic Algorithm Optimized RBF Network, “Research Journal of Applied Sciences, Engineering and Technology”, Vol. 6, No. 5, 2013, 895–899.
  33. LeCun Y., Denker J., Solla S., Optimal brain damage, Advances in NIPS2, Morgan Kaufman, San Mateo, 1900, 598–605.
  34. Bilski P., Mazurek P., Wagner J., Application of k Nearest Neighbors Approach to the Fall Detection of Elderly People Using Depth-Based Sensors, Proc. IDAACS 2015 Conference, 24–26 Sept. 2015, Warsaw, Poland, 733–739.
  35. Bilski A., Wojciechowski J., Automatic parametric fault detection in complex analog systems based on a method of minimum node selection, “Int. J. Appl. Math. Comput. Sci.”, Vol. 26, No. 3, 2016, 655–668, DOI: 10.1515/amcs-2016-0045.
  36. Korbicz J., Kościelny J.M., Kowalczuk Z., Cholewa W., Fault Diagnosis. Models. Artificial Intelligence. Applications, Springer Verlag, 2004.
  37. Bartyś M., Chosen Issues of Fault Isolation. Theory, Practice and Applications, PWN, Warszawa 2014.