一個複雜的工程系統包含很多元件，我們分析的情況就不能忽略這些元件有互相的影響，而在一個現實的工程系統中，連鎖失效的反應是非常常見的，我們發現現有的方法（敏感度分析、FMEA...等）都存在著一些缺陷，造成運用在實際的工程問題上的困難。

本研究是在探討結合不同的資料與分析方法來建立一個較完整的評估連鎖失效系統的方法。當我們只擁有一些關於元件與系統可靠與否的對應資料，我們利用B-reliability importance measure來評估元件的重要性，且結合元件破壞的順序與互相影響情形來建立一個較完整的評估連鎖失效系統的方法Cascading Importance Measure，並將此方法運用在模型上且藉由可靠度測試來驗證此方法。

Cascading Importance Measure方法評估的是元件可靠度對系統可靠度的重要性，但是我們是藉由調整設計變數來調整元件可靠度，而調整的設計變數並不只影響自己的元件。所以我們藉由一些關於變數與元件之間的敏感度分析去建立一套更完整的評估方法 CIM+ 來達到評估變數與系統可靠度之間的關係，且藉由測試結果來驗證我們的評估結果。

最後我們透過一個有連鎖失效情況的電子系統，利用我們的上述的兩種方法分別分析這個電子系統，並且藉由測試結果來驗證我們的評估結果。

Our research aims to build a more integrated analysis to measure the importance in a cascading failure system by combining different existing data. We find out the influence between components can not be ignored in a complex system. The cascading failure condition is very common in a engineering system, but the common methods that analyze influence from components to system (ex:sensitivity, FMEA,B-reliability importance measure...) are not suit enough to deal with a black-box cascading failure system.

In most design cases, we have limited data about the success or failure condition of some components and system from current design. In order to build a more integrated analysis, we use the B-reliability importance measure combine with the failure dependency matrix and sequence of failure that we called Cascading Importance Measure (CIM). Based on the limited existing data, the CIM analysis is hard to completely correct in the testing result. Because the CIM is measuring the importance of component reliability to system reliability, and adjusting design variables to improve the components' reliability. But the change of the design variable is not only change the reliability of it's own component,

To build a more completed analysis, we use the information of sensitivity analysis from variables to each component's performance.
Under the data of sensitivity analysis from variables to each component's performance, we can build a new method to enhance CIM (CIM+).
And measure the importance from design variables to system reliability.

Finally, we use our two method to a electrical system which is a typical cascading failure system. And then verify the result of two methods by testing result of improving system reliability.

[1] W. Meeker and L. Escobar, Statistical Methods for Reliability Data. Wiley-Interscience, 1998.
[2] I. Gijbels, Censored data," Wiley Interdisciplinary Reviews: Computational Statistics, vol. 2,
no. 2, pp. 178{188, 2010.
[3] L. Ljung, Signal Analysis and Prediction. Birkhuser Boston, 1998.
[4] S. M. Rinaldi, J. P. Peerenboom, and T. K. Kelly, Identifying, understanding, and ana-lyzing
critical infrastructure interdependencies," IEEE Control Systems Magazine, pp. 11{25, 2001.
[5] Normal distribution from wikipedia." http://en.wikipedia.org/wiki/Gaussian_
distribution, 2014.
[6] Exponential distribution from wikipedia." http://en.wikipedia.org/wiki/Exponential_
distribution, 2014.
[7] Log-normal distribution from wikipedia." http://en.wikipedia.org/wiki/Lognormal_
distribution, 2014.
[8] A. Haldar and S. Mahadevan, Probability, reliability, and statistical methods in engineering design.
John Wiley, 2000.
[9] Failure mode and e ects analysis from wikipedia." http://en.wikipedia.org/wiki/Failure_
mode_and_effects_analysis, 2014.
[10] Sensitivity analysis from wikipedia." http://en.wikipedia.org/wiki/Sensitivity_
analysis, 2014.
[11] Z.W.Birnbaun, On the importance of di erent components in a multicomponent system," Tech-
nical Report, vol. 54, no. 38, pp. 1{23, 1968.
[12] H. Dui, X. Zhao, Shenggui, and S. Sun, The integrated importance measure of multi-state
coherent system for maintenance process," IEEE TRANSACTIONS ON RELIABILITY, vol. 61,
no. 2, pp. 266{273, 2012.
[13] W. Kuo and X. Zhu, Importance Measure in Reliability, Risk, and Optimization. John Wiley and
sons, 2012.
[14] I. Lee, Reliability-Based Dedign Optimizaion and Robust Design Optimizayion Using Univariate
Dimension Reduction Method. ProQuest LLC, 2008.
[15] D.-Y. Wu, Composites of surrogate models with data uncertainty," Master's thesis, National
Cheng Kung University, 2013.
[16] P.-Y. Lin, Optimal sampling augmentation and resorce allocation for design with inadequate
uncertainty," Master's thesis, National Cheng Kung University, 2011.
[17] Y.-C. Han, A baysesian based updated scheme for belt pulley system design with censored life
data," Master's thesis, National Cheng Kung University, 2013.
[18] R. O. Hamburger, D. A. Foutch, and C. A. Cornell, Translating research to practice: Fema/sac
performance-based design procedures," Earthquake Spectra, vol. 19, no. 2, pp. 255{267, 2003.
[19] J. Jin and S. El-Tawil, Evaluation of fema-350 seismic provisions for steel panel zones," Structural
Engineering, vol. 131, no. 2, pp. 250{258, 2005.
[20] S. T. Meier, Improving design sensitivity through intervention-sensitive measures," Evaluation,
vol. 25, no. 3, pp. 321{334, 2004.
[21] N. H. Kim, H. Wang, and N. V. Queipo, Adaptive reduction of design variables using global sen-
sitivity in reliability-based optimization," International Journal of Reliability and Safety, vol. 1,
no. 1/2, pp. 103{119, 2006.
[22] H. Liu and W. Chen, Probabilistic sensitivity analysis methods for design under uncertainty," in
AIAA/ISSMO Multidisciptinary Analysis and Optimization, no. 10 in 4589, (Albany, New York),
pp. 1{16, 2004.
[23] I. Dobsona, B. A. Carreras, V. E. Lynch, and D. E. Newman, Complex systems analysis of series
of blackouts: Cascading failure, critical points, and self-organization," Chaos, vol. 17, no. 026103,
pp. 1{13, 2007.
[24] D. P. Nedic, I. Dobson, D. S. Lirschen, B. A. Carreras, and V. E. Lynch, Criticality in a cascading
failure blackout model," International Journal of Electrical Power and Energy Systems, vol. 28,
pp. 627{633, 2006.
[25] H. Ren and I. Dobson, Using transmission line outage data to estimate cascading failure propa-
gation in an electric power system," IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS,
vol. 55, no. 9, pp. 927{931, 2008.
[26] M. Zhongwei, Z. Shuangxi, and L. Zongxiang, Cascading failure analysis of bulk power system
using small-world network model," in Probabilistic Methods Applied to Power Systems, no. 8,
(Iowa State, University), pp. 635{640, 2004.
[27] A. Phadke and J. Thorp, Expose hidden failures to prevent cascading outages [in power sys-
tems]," Computer Applications in Power, IEEE, vol. 9, no. 3, pp. 20{24, 1996.
[28] Boost converter from wikipedia." http://en.wikipedia.org/wiki/Boost_converter, 2014.