Conference proceeding
Reliability analysis of uncertain dynamic systems using importance sampling
Proceedings of ICASP 6, Sixth International Conference on Applications of Statistics and Probability in Civil Engineering, Vol.1, pp.305-312
International Conference on Applications of Statistics and Probability in Civil Engineering, 6 (Mexico City, Mexico, 06/17/1991 - 06/21/1991)
01/01/1991
Abstract
Reliability of uncertain linear systems subjected to Gaussian excitation is estimated using the importance sampling technique. Reliability is defined as the probability of system response staying within a safe domain for a duration of time. Since the input is Gaussian, the conditional responses are Gaussian. Thus, the conditional failure probability due to first crossing can be obtained easily. Similarly, for a Duffing oscillator with uncertain parameters subjected to Gaussian white noise input, the conditional mean crossing rate can be estimated easily. The success of the importance sampling technique depends on the choice importance sampling density. This paper demonstrates the use of some basic knowledge of the dynamic behavior to estimate the sampling density. Results obtained for one and two degree-of-freedom linear systems and a Duffing oscillator compared very well with Monte Carlo simulation. However, the problems encountered with importance sampling, such as the possibility of obtaining a bad estimator with a small estimated coefficient of variation, do arise in these applications.
Details
- Title: Subtitle
- Reliability analysis of uncertain dynamic systems using importance sampling
- Creators
- Yaacob IbrahimSharif Rahman
- Resource Type
- Conference proceeding
- Publication Details
- Proceedings of ICASP 6, Sixth International Conference on Applications of Statistics and Probability in Civil Engineering, Vol.1, pp.305-312
- Conference
- International Conference on Applications of Statistics and Probability in Civil Engineering, 6 (Mexico City, Mexico, 06/17/1991 - 06/21/1991)
- Language
- English
- Date published
- 01/01/1991
- Academic Unit
- Mechanical Engineering
- Record Identifier
- 9984201445802771
Metrics
1 Record Views