Optimal interval design for system reliability with incomplete FDS by means of improved genetic algorithms
✍ Scribed by Takao Yokota; Mitsuo Gen
- Publisher
- John Wiley and Sons
- Year
- 1998
- Tongue
- English
- Weight
- 240 KB
- Volume
- 81
- Category
- Article
- ISSN
- 8756-663X
No coin nor oath required. For personal study only.
✦ Synopsis
In previous discussions about product quality assurance, reliability has been considered a definitive value. In realistic optimal design for system reliability, however, we are required to handle interval data with specified reliability, as in the case of quality accuracy. This paper discusses an incomplete FDS system reliability optimization problem, in which nonlinear interval reliability is maximized under a nonlinear interval constraint. This is especially appropriate to the problem discussed in this paper, where a complex expression of the interval data, including multiplication and division, is given as the nonlinear constraint. A method is proposed that allows the problem to be converted into a single object nonlinear integer programming problem without interval coefficients. The optimal solution is easily derived by solving the converted problem by means of an improved genetic algorithm (GA). In the improved GA proposed in this paper, all searches are executed by arithmetic cross-over in the executable region, and the mutation is executed for selected elements within the range of the determining variables so that their values are replaced by integer values that give the best evaluation function value. Convergence to the solution is improved by the above neighborhood search. Using a numerical experiment, we examine the effect on the reliability interval of preferred selection of the constraint by a decision-maker. The effectiveness of the proposed method is demonstrated. ©1998