Two-Stage Neural Networks Based Classifier System for Fault Diagnosis
This chapter gives a description of a two-stage classifier system for fault diagnosis of industrial processes. The first-stage classifier is used for fault detection and the second one is used for fault isolation and identification. The first stage classifier operates as primary fault detection unit, and it is used to distinguish between normal operating state and abnormal operating states. In order to reduce the number of false alarms, a penalizing factor is introduced in the training error cost function. The second-stage classifier is used to differentiate between different detectable faults. In order to increase the reliability of fault identification, the probabilities of classification performed by this classifier are averaged within the fault duration time. The performance of the proposed approach is validated by application to a valve actuator fault diagnosis problem.
KeywordsFault Detection Fault Diagnosis Fault Identification Fault Isolation Faulty State
Unable to display preview. Download preview PDF.
- 1.Angeli C and Chatzinikolaou A (2004) On-Line Fault Detection Techniques for Technical Systems: A Survey. International Journal of Computer Science & Applications 1(1):12–30Google Scholar
- 4.Calado JMF, Korbicz J, Patan K, Patton RJ and Sa da Costa JMG (2001) Soft Computing Approaches to Fault Diagnosis for Dynamic Systems. European Journal of Control 7:248–286Google Scholar
- 7.DeMers D and Cottrell GW (1993) Non-linear dimensionality reduction. In: Giles CL, Hanson SJ and Cowan JD (eds) Advances in Neural Information Processing Systems 5, Morgan Kaufmann, San Mateo, CA, pp. 580–587Google Scholar
- 8.Duda RO, Hart PE and Stork DG (2000) Pattern Classification. John Wiley & Sons, New York, 2nd EditionGoogle Scholar
- 9.EC FP5 Research Training Network DAMADICS Development and Application of Methods for Actuator Diagnosis in Industrial Control Systems, http://www.eng.hull.ac.uk/research/control/damadics1.htmGoogle Scholar
- 14.Jollife IT (1986) Principal Component Analysis. Springer Series in Statistics. Springer-Verlag, New YorkGoogle Scholar
- 15.Lipnickas A, Korbicz J (2004) Adaptive Selection of Neural Networks for a Committee Decision. International Scientific Journal of Computing 3(2):23–30.Google Scholar
- 16.Lipnickas A, Sa da Costa J and Bocaniala C (2004) FDI based on two stage classifiers for fault diagnosis of valve actuators. Application to DAMADICS benchmark. In: Proceeding of the 11th International Conference EPEPEMC’2004, 2–4 September, Riga, Latvia, vol. 3, pp. 147–153Google Scholar
- 19.Patton RJ, Lopez-Toribio CJ, Uppal FJ (1999) Artificial intelligence approaches to fault diagnosis for dynamic systems. International Journal of applied mathematics and computer science 9(3):471–518Google Scholar
- 20.Raudys Š (2001) Statistical and Neural Classifiers: An Integrated Approach to Design. Advances in Pattern Recognition. Springer BerlinGoogle Scholar
- 21.Teeuwsen SP, Erlich I and El-Sharkawi MA (2002) Feature reduction for neural network based small-signal stability assessment. In: Proceedings of the 14th PSCC, Sevilla, 24–28 June, session 14Google Scholar
- 27.Wang LX (1992) Fuzzy systems are universal approximators. In: Proceedings of IEEE Fuzzy Systems Conference, pp. 1163–1170Google Scholar