Genetic Algorithms for Reliability Design Problems

This paper presents genetic algorithms for solving various reliability design problems, which include series systems, series-parallel systems and complex (bridge) systems. The objective is to maximize the system reliability, while maintaining feasibility with respect to three nonlinear constraints, namely, cost and weight constraints, and constraints on the products of volume and weight. In this paper, both integer reliability problems (component reliabilities are given and redundancy allocation is to be decided) and mixed-integer reliability problems (both component reliabilities and redundancy allocation are to be decided simultaneously) are studied. Numerical examples show that genetic algorithms perform well for all the reliability problems considered in this paper. In particular, as reported, some solutions obtained by genetic algorithms are better than previously best-known solutions. ______________________________________________________________________ All correspondence should be directly addressedto Prof. Dennis L. Bricker, Department of Industrial Engineering, University of Iowa, Iowa City, IA 52242, USA. [email protected]