OPTIMAL-DESIGN OF SYSTEM RELIABILITY BY AN IMPROVED GENETIC ALGORITHM

When genetic algorithms are applied to large-scale optimization proble ms taking only integer-valued solutions, too often the chromosomes gen erated fail to satisfy the constraint conditions. In such cases, the s election operator is unable to produce high-quality chromosomes, worse ning the search efficiency. As a consequence, many times good solution s cannot be obtained. In this paper, in order to cope with such a prob lem, a new algorithm with improved chromosome selection mechanism is p roposed, taking into account the evaluation function including some in formation about chromosomes not contained in the feasible region of th e nonlinear integer programming problem under consideration. The propo sed method considers how far each unfeasible chromosome is from the fe asible region. The improved algorithm was applied to a large-scale pro blem of optimal design of system reliability formulated as a nonlinear integer programming problem and numerical experiments confirmed the e ffectiveness of the method.