1 Vanderbilt University/MACS Lab Bridge Workshop – 03/01 TRANSCEND: Diagnosis of Continuous Dynamic Systems from Transients Gautam Biswas MACS Laboratory Dept. of EECS Vanderbilt University http://www.vuse.vanderbilt.edu/~biswas Collaborators: Pieter Mosterman, Eric Manders, Joel Barnett, Sriram Narasimhan, Philippus Feenstra, Liguo Yu This project supported by: HP Labs (Agilent Labs), USA, PNC, Japan, the DARPA Software-enabled Control program (# F33615-99-C-3611), and the NASA-IS program. Vanderbilt University/MACS Lab Bridge Workshop – 03/01 Model-Based Diagnosis of Dynamic Systems • Values: Qualitative vs. Quantitative 7 qualitative models do not require numerical parameter values but diagnosis is less precise 7 computational complexity of qualitative methods may be less • Temporal Behavior: Discrete vs. Continuous 7 discrete methods may be easier to design but are less precise, coarser 7 spurious results FDI Models: examples GDE State Estimation Parameter Estimation Quantitative Sampath, et al. Lunze, et al. Fault Signatures & TCGs (Mosterman and Biswas) Qualitative Discrete Continuous
17
Embed
Continuous Dynamic Systems from Transients - …dx01/bridge_material/biswas.pdf · discrete methods may be easier to design but are less precise, coarser ... Struss and Dressler,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
• Dynamic, Continuous System• Abrupt Faults (not incipient or intermittent)
In theory, we are looking to characterize step response to abrupt change in a parameter value – called a transient.� Modeling Transients in Qualitative
Framework (Signatures) � Tracking dynamic effects of faults by
Progressive Monitoring
• Adequate Modeling Schemes(parameter to measurement relations)
•Work with Real Data���� Noisy, therefore, statistical techniques for change detection
• Parallels from control theory and systems dynamics -- Loop Analysis (e.g., Mason’s Gain rule) in control theory and Frequency Response Analysis
• Parallels from AI MBD approaches: reasoning in two steps: hypothesis generation and hypothesis refinement
• Differences from traditional FDI approaches – Qualitative Reasoning Framework + Local Analysis + Local to Global Propagation (i.e., consistency checking) over time
where Rk(t) is the remainder term based on y(k+1)(t).
Signal transient due to a fault at t0 can be expressed as discontinuous magnitude change, y(t0), plus first and higher order derivative changes, y'(t0), y''(t0), ….., y(k)(t0).
Discriminatory Power of Qualitative Fault Signatures
1. Abrupt change – direction of abrupt change + direction of change immediately following abrupt change, (+,+), (+,-), (-,+), and (-,-)
2. No abrupt change – first direction of change of the signal only, (0,+), and (0,-)
Problem: further +,- changes provide no discriminatory evidence because qualitative information contains no time constant information.
Ways to handle this problem:• measurement selection – end up needing
many more measurements than log4[k], where k – number of fault hypotheses.
• estimate fault parameters values; true fault is the one whose parameter is consistent across multiple measurements. Use gradient descent methods for parameter estimation.
Conclusions• Developed a comprehensive methodology
for diagnosis based on transient analysis� FDI-based models to capture continuous
dynamics� Derived topological model (TCG) for localized
analysis of faults – dynamics captured as fault signatures
� Progressive Monitoring to match signatures to observations
• Dealing with real signals but processing them qualitatively – required signal to symbol transformation techniques (based on time-frequency analysis)
• Limitations of qualitative reasoning led to combined approach. Reduced set of fault hypotheses spawned fault observers for fault parameter estimation
• P.J. Mosterman and G. Biswas, “Monitoring, Prediction, and fault Isolation in Dynamic Physical Systems, Proc. AAAI-97, pp. 100-105, Providence, RI, July 1997.
• P.J. Mosterman and G. Biswas, “Diagnosis of Continuous Valued Systems in Transient Operating regions,” IEEE Trans. On Systems, Man, and Cybernetics, vol. 29A, no. 6, pp 554-565, Nov. 1999.
• E.J. Manders, et al., “A combined Qualitative-Quantit-• ative for efficient Fault Isolation in Complex Dynamic
Systems, Safeprocess, pp. 1074-1079, 2000.
• E.J. Manders, P.J. Mosterman, and G. Biswas, “Signal to Symbol Transformation for Robust Diagnosis in TRANSCEND,” Tenth Intl. Workshop on Principles of Diagnosis, Loch Awe, Scotland, pp. 155-165, June 1999.
• S. McIlraith, G. Biswas, D. Clancy, and V. Gupta, “Towards Diagnosing Hybrid Systems,” Tenth Intl. Workshop on Principles of Diagnosis, Loch Awe, Scotland, pp. 193-203, June 1999.
• Narasimhan, et al., Safeprocess 2000• Narasimhan and Biswas, DX-01.