» Download Article

Authors
Zheng Niu; Yuguang Niu; Yurong Li; Jizhen Liu

Digital Object Identifier (DOI)
10.3182/20060829-4-CN-2909.00234

Page Numbers:
1402-1406

Index Terms
fault isolation,dynamic time warping,comparability measure,robustness

Abstract
The precise mathematical model of industrial process is difficult to be obtained, so a fault isolation method independent of process mathematical model is proposed. This method is based on dynamic time warping (DTW) technology and isolates process fault by process data analysis . First, the fault pattern storehouse is established through history data analysis and process knowledge. Then the detection sample is carried on pattern matching with each fault sample in fault pattern storehouse by DTW. Finally, the matching results are quantitatively evaluated to isolate fault through comparability measure. The simulation result of three-tank-system indicates that this method has good robustness to the process time -varying characteristic, thus has high fault isolation precision.

References

[1] Bo Xu, Hailong Tang and Xingshan Li (2004). DTW
    based quantitative fault diagnosis of gas path
    component in turbofan. Journal of Beijing
    University of Aeronautics and Astronautics,
    30(6), 524-528.
[2] Chiang L. H., Russell E. L. and Braatz R. D. (2001).
    Fault detection and diagnosis in industrial
    systems. Springer, London.
[3] Donghua Zhou and Yinzhong Ye (2000). In: Modern
    fault diagnosis and fault tolerant control.
    Tsinghua University Press, Beijing.
[4] Gertler J. J. (1998). Fault detection and diagnosis in
    engineering systems. Marcel Dekker, Inc., New
    York.
[5] Isermann R. (1995). Model based fault detection and
    diagnosis methods. Process of the American
    Control Conference, New Jersey, pp. 1605-1609.
[6] Isermann R. and Freyermuth B. (1991). Process fault
    diagnosis based on process model knowledge-Part
     I: Principles for fault diagnosis with
    parameter estimation. Journal of Dynamic
    Systems, Measurement and Control, 113(4),
    620-626.
[7] Keogh E. J. and Pazzani M. J. (2000). Scaling up
    dynamic time warping for data mining
    application. Proceedings of the 6th International
    Conference on Knowledge Discovery and Data
    Mining, Boston, pp. 285-289.
[8] Scinivasan A. and Batur C. (1994). Fault detection
    and isolation in an unsupervised learning
    environment. Pattern Recognition Letters, 15(3),
    235-242.
[9] Xiaobin Huang, Yuguang Niu, Jizhen Liu and
    Yuanzhi Sun (2003). Fault diagnosis for sensors
    based on fuzzy dynamical model. Proceedings of
    the CSEE, 23(3), 183-187.
[10] Yong Chen and Jun Liang (2002). Monitoring and
     fault diagnosis based on PCA for multivariable
     control system. Journal of Engineering Design,
     9(5), 257-260.
[11] Zhang J., Martin E. and Morris A. J. (1995). Fault
     detection and classification through multivariate
     statistical techniques. Process of the American
     Control Conference, New Jersey, pp. 751-755.
[12] Zheng Niu, Jizhen Liu and Yuguang Niu (2005).
     Fault detection under varying load conditions
     based on dynamic multi-principal component
     models. Chinese Journal of Power Engineering,
     25(4), 554-558.