The point of maximum likelihood in a failure domain yields the highest value of the probability density function in the failure domain. The maximum-likelihood point thus represents the worst combination of random variables that contribute in the failure event. In this work Genetic Algorithms (GAs) with an adaptive penalty scheme have been proposed as a tool for the determination of the maximum likelihood point. The utilization of only numerical values in the GAs operation makes the algorithms applicable to cases of non-linear and implicit single and multiple limit state function(s). The algorithmic simplicity readily extends its application to higher dimensional problems. When combined with Monte Carlo Simulation, the proposed methodology will reduce the computational complexity and at the same time will enhance the possibility in rare-event analysis under limited computational resources. Since, there is no approximation done in the procedure, the solution obtained is considered accurate. Consequently, GAs can be used as a tool for increasing the computational efficiency in the element and system reliability analyses.
[1]
Kalyanmoy Deb,et al.
Optimization for Engineering Design: Algorithms and Examples
,
2004
.
[2]
C. Bucher,et al.
Importance sampling for randomly excited dynamical systems
,
2003
.
[3]
Alfred M. Freudenthal,et al.
SAFETY AND THE PROBABILITY OF STRUCTURAL FAILURE
,
1956
.
[4]
Helio J. C. Barbosa,et al.
A new adaptive penalty scheme for genetic algorithms
,
2003,
Inf. Sci..
[5]
Mitsuo Gen,et al.
Genetic algorithms and engineering design
,
1997
.
[6]
Ronald L. Wasserstein,et al.
Monte Carlo: Concepts, Algorithms, and Applications
,
1997
.
[7]
David E. Goldberg,et al.
Genetic Algorithms in Search Optimization and Machine Learning
,
1988
.