DOI QR코드

DOI QR Code

Principal Component Analysis Based Method for a Fault Diagnosis Model DAMADICS Process

주성분 분석을 이용한 DAMADICS 공정의 이상진단 모델 개발

  • Received : 2016.02.22
  • Accepted : 2016.07.05
  • Published : 2016.08.31

Abstract

In order to guarantee the process safety and prevent accidents, the deviations from normal operating conditions should be monitored and their root causes have to be identified as soon as possible. The statistical theories-based method among various fault diagnosis methods has been gaining popularity, due to simplicity and quickness. However, according to fault magnitudes, the scalar value generated by statistical methods can be changed and this point can lead to produce wrong information. To solve this difficulty, this work employs PCA (Principal Component Analysis) based method with qualitative information. In the case study of our previous study, the number of assumed faults is much smaller than that of process variables. In the case study of this study, the number of predefined faults is 19, while that of process variables is 6. It means that a fault diagnosis becomes more difficult and it is really hard to isolate a single fault with a small number of variables. The PCA model is constructed under normal operation data in order to get a loading vector and the data set of assumed faulty conditions is applied with PCA model. The significant changes on PC (Principal Components) axes are monitored with CUSUM (Cumulative Sum Control Chart) and recorded to make the information, which can be used to identify the types of fault.

Keywords

fault diagnosis;principal component analysis;cumulative sum control chart

References

  1. C. J. Lee, "Optimal Multi-floor Plant Layout Based on the Mathematical Programming and Particle Swarm Optimization", Industrial Health , Vol. 53, No. 6, pp. 491-497, 2015. https://doi.org/10.2486/indhealth.2014-0234
  2. C. J. Lee, G. Lee and J. M. Lee, "A Fault Magnitude Based Strategy for Effective Fault Classification", Chemical Engineering Research and Design, Vol. 91, No. 3, pp. 530-541, 2012.
  3. Tarun Chopra, Jayashri Vajpai, "Classification of Faults in DAMADICS Benchmark Process Control System using Self Organizing Maps" International Journal of Soft Computing and Engineering, Vol. 1, Issue. 3, 2011.
  4. Sa da Costa, J. and R. Louro, "Modelling and Simulation of an Industrial Actuator Valve for Fault Diagnosis Benchmark," Proceedings of the Fourth International Symposium on Mathematical Modelling, pp. 1212-1221, 2003.
  5. M Bartys, R Patton, M Syfert, S de las Heras, J Quevedo, "Introduction to the DAMADICS Actuator FDI Benchmark Study," Control Engineering Practice, Vol. 14, No. 6, pp. 577-596. 2006. https://doi.org/10.1016/j.conengprac.2005.06.015
  6. D. H. Hwang, H. W. Cho, C. H. Han and J. H. Kim ,"On-Line Monitoring Methods using Multivariate Statistical Method," Chemical Industry and Technology, Vol. 15, No. 3, pp. 247-255, 1997.
  7. H. D. Lee, M. H. Lee, H. W. Cho, C. H. Han and K. S. Chang, "Online Quality Monitoring using Multivariate Statistical Methods in Continuous-Stirred MMA-VA Copolymerization Process", Korean Chemical Engineering Research, Vol. 35, No. 5, pp. 605-612, 1997.
  8. S. J. Hong and C. H. Han, "Data-Driven Software Sensor Design for Monitoring, Diagnosis and Control," Chemical Industry and Technology, Vol. 17, No. 2, pp. 172-181, 1999.
  9. S. J. Hong, C. K. Heo and C. H. Han, "Local Composition Soft Sensor in a Distillation Column using PLS," Korean Chemical Engineering Research, Vol. 37, No. 3, pp. 445-452, 1999.
  10. Y. H. Lee, C. H. Han and J. K. Lee, "Real-time Monitoring for a Batch PVC Polymerization Process Based on Multivariate Data Compression Methods," Korean Chemical Engineering Research, Vol. 37, No. 2, pp. 319-329, 1999.
  11. K. H. Yoon, Y. H. Lee, and C. H. Han, "Adaptive Block-Wise RPLS Considering Similarity of Blocks," Korean Chemical Engineering Research, Vol. 41, No. 5, pp. 592-597, 2003.
  12. D. M. Yoon, Y. H. Lee, C. H. Han, H. S. Ah and S. H. Chang, "Fault Detection and Diagnosis in Film Processing Plants," Korean Chemical Engineering Research, Vol. 41, No. 5, pp. 585-591, 2003.
  13. S. Lee, S. Yeom and K. S. Lee, "Methods for Performance Monitoring and Diagnosis of Multivariable Model-Based Control Systems," Korean Chemical Engineering Research, Vol. 21 No. 3, pp. 575-581, 2004. https://doi.org/10.1007/BF02705490
  14. C. J. Lee, S. O. Song and I. S. Yoon, "The Monitoring of Chemical Process using the Support Vector Machine," Korean Chemical Engineering Research, Vol. 42, No. 5, pp. 538-544, 2004.
  15. J. Y. Park and C. J. Lee, "Principal Component Analysis Based Method for Effective Fault Diagnosis," Journal of the Korean Society of Safety, Vol. 29, No. 4, pp. 73-77, 2014. https://doi.org/10.14346/JKOSOS.2014.29.4.073
  16. C. J. Lee, G. B. Lee and J. M. Lee, "A Fault Magnitudebased Strategy for Effective Fault Diagnosis and Isolation," Journal of Chemical Engineering of Japan, Vol. 48, No. 1, pp. 44-51, 2015. https://doi.org/10.1252/jcej.14we116
  17. K. J. Lee, B. W. Lee, D. H. Choi and T. O. Kim "A Study on Fault Detection Monitoring and Diagnosis System of CNG Stations Based on Principal Component Analysis(PCA)", Journal of the Korean Institute of Gas, Vol. 18, No. 3, pp. 53-59, 2014. https://doi.org/10.7842/kigas.2014.18.3.53
  18. E. S. page, "Cumlative Sum Charts", Technometrics, Vol.3 No.1, pp.562-593, 1961.

Acknowledgement

Supported by : 부경대학교