Advertisement

Neural Networks for Detection of Process Change in Manufacturing Systems

  • Tep Sastri

Abstract

Two design prototypes of neural network for on-line process chnage detection in dynamic model-switching environments are presented. The process model is assumed to follow either an ARMAX or a non-linear Volterra representation. The connection strengths of the first neural network are adaptable parameter estimates of the underlying process model. The second prototype is based on the learning vector quantization procedure. Two important design considerations are discussed; i.e., what type of distance measures are suitable for process change detection and how the neural networks should be trained for on-line applications. Various change detection measures and training procedures are also discussed. Finally, on-line performance of the proposed neural networks is demonstrated via computer simulation experiments.

Keywords

Change Detection Vector Quantization Learning Vector Quantization Recursive Identification Synthetic Time Series 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. Ahalt, S. C., A. K. Krishnamurthy, P. Chen, and D.E. Melton, “Competitive learning algorithms for vector quantization,” Neural Networks, Vol. 3, pp. 277–290, 1990.CrossRefGoogle Scholar
  2. Athans, M., and D. Willner, “A practical scheme for adaptive aircraft flight control systems,” Symp. on Parameter Estimation Techniques and Applications in Aircraft Flight Testing, NASA Flt. Res. Ctr., Edwards AFB, April 1973.Google Scholar
  3. Bhat, N. and T. J. McAvoy, “Use of neural nets for dynamic modeling and control of chemical systems,” Computers Chem. Engng. Vol. 14, NO. 4/5, pp 573-583, 1990.Google Scholar
  4. Box, G.E.P., and D.A. Pierce, “Distribution of residual autocorrelations in autoregressive-integrated moving average time series models,” J. Amer. Stat. Assoc. Vol. 65, pp. 1509-1526, 1970.Google Scholar
  5. DARPA Neural Network Study Executive summary. Lincoln Laboratory, MIT Press, July 1988.Google Scholar
  6. DeSieno, D., “Adding a conscience to competitive learning,” Proc. of the Second Annual IEE Intern. Conference of Neural Networks, Vol. 1, 1988.Google Scholar
  7. Durbin, J., “The approximate distribution of partial serial correlation coefficients from residuals from regression on Fourier series,” Biometrika, Vol. 67, pp. 335-349, 1980.Google Scholar
  8. Kartikeyan, B. and A. Sarkar, “Shape description by time series,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 11, No. 9, September 1989.Google Scholar
  9. Kohonen, T., Self-Organization and Associative Memory (Second Edition). Spinger-Verlag, New York, 1988.Google Scholar
  10. Kohonen, T., “The self-organizing map,” Proc. of the IEEE, Vol. 78, No. 9, pp. 1464–1480.Google Scholar
  11. Landau, I.D., “Elimination of the real positivity condition in the design of parallel MRAS,” IEEE Trans. Auto. Control 23, 1015–1020, 1978.CrossRefGoogle Scholar
  12. Lainiotis, D.G., “Joint detection, estimation, and system identification,” Inform. Control 19, 75–92, Aug. 1971.CrossRefGoogle Scholar
  13. Lippman, R. P., “An introduction to computing with neural nets,” IEEE ASSP Magazine, pp. 4-22, April 1987.Google Scholar
  14. Ljung, L., and T. Söderström, Theory and Practice of Recursive Identification. MIT Press, Cambridge, MA. 1983.Google Scholar
  15. Narendra, K. and A. M. Anaswamy, Stable Adaptive Systems, Prentice Hall, Englewood Cliffs, N. J., 1989.Google Scholar
  16. Pandit, S. M., and S. M. Wu, and Wu, Time Series and System Analysis with Applications, Wiley, N.Y., 1983.Google Scholar
  17. Priestley, M. B., Non-linear and Non-stationary Time Series Analysis, Academic Press, N.Y. 1988.Google Scholar
  18. Rumelhart D. E., and J. L. McClelland, Parallel distributed processing: explorations in microstructure of cognition, MIT Press, 1986.Google Scholar
  19. Sarkar, A. and B. Kartikeyan, “Forecasting by Volterra type models,” J. Stat. Computat. Simulation, Vol. 28, pp. 245–260, 1987.CrossRefGoogle Scholar
  20. Sastri, T., “A state-space modeling approach to time series forecasting,” Management Science, Vol. 31, No. 11, pp. 1451-1470, 1985.Google Scholar
  21. Sastri, T., “A recursive algorithm for adaptive estimation and parameter change detection of time series models,” J. Operational Research Soc, Vol. 37, No 10, pp. 987-999, 1986.Google Scholar
  22. Sastri, T. “Sequential method of change detection and adaptive prediction of municipal water demand,” Int. J. Systems Sci., Vol. 18, No. 6, pp. 1029-1049, 1987.Google Scholar
  23. Sastri, T., J. R. English, and G. N. Wang, “Neural networks for time series model identification and change detection,” Advances in Artificial Intelligence in Economics, Finance, and Management, 1992 (forthcoming).Google Scholar
  24. Stengel, R. F., Stochastic Optimal Control Theory and Application, Wiley, N.Y. 1986.Google Scholar
  25. Willsky, A.S., “A survey of design methods for failure detection in dynamic systems,” Automatica, 12, 601-611, 1976.Google Scholar

Copyright information

© Springer-Verlag Berlin · Heidelberg 1992

Authors and Affiliations

  • Tep Sastri
    • 1
  1. 1.Industrial Engineering DepartmentTexas A&M UniversityCollege StationUSA

Personalised recommendations