Advertisement

Causal Models for Distributed Fault Diagnosis of Complex Systems

  • Cosmin Danut Bocaniala
  • José Sá da Costa
Part of the Advanced Information and Knowledge Processing book series (AI&KP)

Abstract

This chapter describes a novel framework for using causal models in distributed fault diagnosis. The state-of-the-art distributed fault diagnosis methodologies lack a coherent partitioning methodology of the monitored system into a set of subsystems, such that the independence level of local diagnosis process for each subsystem is maximal and such that the communication between different subsystems, required for formulating global diagnosis, is minimal. The partitioning of the causal model is performed with regard to the d-separation property that renders each region of the partition causally independent from the rest of the model. This special property allows fault diagnosis to be performed locally, without the need of communicating with the rest of the model, as long as the border with the rest of the model is healthy, i.e., maximum independence level of local diagnosis processes. Moreover, the causal model is partitioned so that the regions of the partitions are separated by borders containing a minimal number of vertices. It follows that if communication with the neighbouring elements is needed, the computational complexity of the process is minimal.

Keywords

Fault Diagnosis Causal Model Strongly Connect Component Reachability Matrix Acyclic Digraph 
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. 1.
    Albert M, Lägle T, Wön H, Kazi A, Brighenti A, Senior C, Revuelta Seijo C, Sanz Bobi MA, Villar J (2001) Distributed architecture for monitoring and diagnosis. EU ESPRIT Project DIAMONDGoogle Scholar
  2. 2.
    Balakrishnan VK (1997) Graph theory, Schaum’s Outlines. McGraw-Hill, New YorkGoogle Scholar
  3. 3.
    Bocaniala CD, Sa da Costa J (2004) Novel framework for using causal models in distributed fault diagnosis. In: Proceedings of Workshop on Advances in Control and Diagnosis, Karlsruhe, Germany, pp. 142–147Google Scholar
  4. 4.
    Bocaniala CD, Sa da Costa J (2005) Novel methodology for partitioning complex systems for fault diagnosis purposes. In: Proceedings of the 16th IFAC World Congress, Praha, Czech RepublicGoogle Scholar
  5. 5.
    Fabre E, Benveniste A, Jard C (2002) Distributed diagnosis for large discrete events in dynamic systems. In: Proceedings of the 15th IFAC World Congress, Barcelona, SpainGoogle Scholar
  6. 6.
    Isermann R, Ballé P (1997) Trends in the application of model-based fault detection and diagnosis of technical processes. Control Engineering Practice 5(5):709–719CrossRefGoogle Scholar
  7. 7.
    Karypis G (2002) Multilevel Hypergraph Partitioning. Technical Report 02-25, Department of Computer Science and Engineering, University of Minnesota, USAGoogle Scholar
  8. 8.
    Köpen-Seliger B, Marcu T, Capobianco M, Gentil S, Albert M, Latzel S (2003) MAGIC: An integrated approach for diagnostic data management and operator support. In: Proceedings of the IFAC Symposium SAFEPROCESS’03, Washington, USA, pp. 187–192Google Scholar
  9. 9.
    Lauritzen SL, Dawid AP, Larsen BN, Leimer HG (1990) Independence properties of directed Markov fields. Networks 20:409–505MathSciNetGoogle Scholar
  10. 10.
    Letia A, Craciun F, Kope Z, Netin A (2000) Distributed diagnosis by BDI agents. In: Proceedings of the OASTED International Conference on Applied InformaticsGoogle Scholar
  11. 11.
    Lesecq S, Gentil S, Exel M, Garcia-Beltran C (2003) Diagnostic tools for a multi-agent monitoring system. In: Proceedings of IMACS IEEE CESA Multi-Conference on Computing Engineering in Systems Applications, Lille, FranceGoogle Scholar
  12. 12.
    Neal RM (2000) On Deducing Conditional Independence from d-separation in Causal Graphs with Feedback. Journal of Artificial Intelligence Research 12:87–91MathSciNetGoogle Scholar
  13. 13.
    Patton RJ (1997) Fault-tolerant control: The 1997 situation. In: Proceedings of the IFAC Symposium SAFEPROCESS’97, Hull, UK, pp. 1033–1055Google Scholar
  14. 14.
    Pearl J, Paz A (1985) Graphoids: A Graph-Based Logic for Reasoning about Relevance Relationships. Technical Report CSD-850038, Computer Science Department, Cognitive Systems Laboratory, University of California, Los Angeles, USAGoogle Scholar
  15. 15.
    Pearl J, Richter D (1996) Identifying Independencies in Causal Graphs with Feedback. In: Proceedings of the Twelfth Annual Conference on Uncertainty in Artificial Intelligence, August 1–4, Reed College, Portland, Oregon, USAGoogle Scholar
  16. 16.
    Pearl J, Verma T (1986) Formal Properties of Probabilistic Dependencies and their Graphical Representations. Technical Report CSD-860019, Computer Science Department, Cognitive Systems Laboratory, University of California, Los Angeles, USAGoogle Scholar
  17. 17.
    Spirtes P (1995) Directed Cyclic Graphical Representations of Feedback Models. In: Proceedings of the Eleventh Annual Conference on Uncertainty in Artificial Intelligence, August 18–20, Montreal, Quebec, CanadaGoogle Scholar
  18. 18.
    Tian J, Verma T, Pearl J (1998) Finding Minimal d-Separators. Technical Report CSD-980007, Computer Science Department, Cognitive Systems Laboratory, University of California, Los Angeles, USAGoogle Scholar
  19. 19.
    Viswanadham N, Sarma VVS, Singh MG (1987) Reliability of Computer and Control Systems. North-Holland Systems and Control Series, Elsevier Science Publishers B.V., Amsterdam, NetherlandszbMATHGoogle Scholar

Copyright information

© Springer-Verlag London Limited 2006

Authors and Affiliations

  • Cosmin Danut Bocaniala
    • 1
  • José Sá da Costa
    • 2
  1. 1.Computer Science and Engineering Department“Dunarea de Jos” University of GalatiGalatiRomania
  2. 2.Department of Mechanical Engineering, GCAR/IDMECTechnical University of LisbonLisbonPortugal

Personalised recommendations