Skip to main content
×
Home
    • Aa
    • Aa

Reasoning about system-level failure behavior from large sets of function-based simulations

  • David C. Jensen (a1), Oladapo Bello (a1), Christopher Hoyle (a2) and Irem Y. Tumer (a2)
Abstract
Abstract

This paper presents the use of data clustering methods applied to the analysis results of a design-stage, functional failure reasoning tool. A system simulation using qualitative descriptions of component behaviors and a functional reasoning tool are used to identify the functional impact of a large set of potential single and multiple fault scenarios. The impact of each scenario is collected as the set of categorical function “health” states for each component-level function in the system. This data represents the space of potential system states. The clustering and statistical tools presented in this paper are used to identify patterns in this system state space. These patterns reflect the underlying emergent failure behavior of the system. Specifically, two data analysis tools are presented and compared. First, a modified k-means clustering algorithm is used with a distance metric of functional effect similarity. Second, a statistical approach known as latent class analysis is used to find an underlying probability model of potential system failure states. These tools are used to reason about how the system responds to complex fault scenarios and assists in identifying potential design changes for fault mitigation. As computational power increases, the ability to reason with large sets of data becomes as critical as the analysis methods used to collect that data. The goal of this work is to provide complex system designers with a means of using early design simulation data to identify and mitigate potential emergent failure behavior.

Copyright
Corresponding author
Reprint requests to: David C. Jensen, Department of Mechanical Engineering, University of Arkansas, 204 Mechanical Engineering Building, Fayetteville, AR 72701, USA. E-mail: dcjensen@uark.edu
Linked references
Hide All

This list contains references from the content that can be linked to their source. For a full set of references and notes please see the PDF or HTML where available.

E. Coatanéa , S. Nonsiri , T. Ritola , I.Y. Tumer , & D. Jensen (2011). A framework for building dimensionless behavioral models to aid in function-based failure propagation analysis. Journal of Mechanical Design 133(12), 121001.

V. Estivill-Castro (2002). Why so many clustering algorithms: a position paper. ACM SIGKDD Explorations Newsletter 4(1), 6575.

T. Kurtoglu , & I.Y. Tumer (2008). A graph-based fault identification and propagation framework for functional design of complex systems. Journal of Mechanical Design 130(5), 18.

T. Kurtoglu , I.Y. Tumer , & D. Jensen (2010). A functional failure reasoning methodology for evaluation of conceptual system architectures. Research in Engineering Design 21(4), 209.

D.A. Linzer , & J. Lewis (2011 a). poLCA: an R package for polytomous variable latent class analysis. Journal of Statistical Software 42(10), 129.

S. Lloyd (1982). Least squares quantization in PCM. IEEE Transactions on Information Theory 28(2), 129137.

S. Sierla , I.Y. Tumer , N. Papakonstantinou , K. Koskinen , & D. Jensen (2012). Early integration of safety to the mechatronic system design process by the functional failure identification and propagation framework. Mechatronics 22(2), 137151.

I.Y. Tumer , & C. Smidts (2010). Integrated design and analysis of software-driven hardware systems. IEEE Transactions on Computers 60(8), 10721084.

J. Vermunt , & J. Magidson (2004). Latent class analysis. In The Sage Encyclopedia of Social Science Research Methods ( M.S. Lewis-Beck , A.E. Bryman , & T.F. Liao , Eds.), Vol. 1. Thousand Oaks, CA: Sage.

Recommend this journal

Email your librarian or administrator to recommend adding this journal to your organisation's collection.

AI EDAM
  • ISSN: 0890-0604
  • EISSN: 1469-1760
  • URL: /core/journals/ai-edam
Please enter your name
Please enter a valid email address
Who would you like to send this to? *
×

Keywords:

Metrics

Full text views

Total number of HTML views: 1
Total number of PDF views: 9 *
Loading metrics...

Abstract views

Total abstract views: 122 *
Loading metrics...

* Views captured on Cambridge Core between September 2016 - 28th May 2017. This data will be updated every 24 hours.