摘要
在造价、体积和重量约束条件下，多级串并联系统的可靠性优化问题是一个具有多局部极值的、非线性的、同时具有整数和实数变量的混合优化问题。把遗传算法和共轭梯度法结合起来，对该问题搜索到了其它算法未能得到的最优解。在遗传算法的应用中，设定了有效的线性定标技术和混合交叉操作，改善了遗传算法的收敛性能。并基于模式理论从数值上表明该问题是符合积水块假设的。最后还从数值上表明遗传算法对该问题是多项式收敛的。
A reliability optimization problem of general serialparallel system, with the constraints of cost, volume and weight, is a nonlinear optimization problem with a large number of local extreme values. Furthermore, both continuous variables such as reliability and discrete variables such as redundancy are involved in the problem. So it is very difficult for traditional optimization techniques to search the global or nearly global optimal solution of such a problem. Genetic algorithm (GA) is a robust evolutionbased stochastic search algorithm having been widely applied to a variety of fields such as function optimization and machine learning. Here, GA is adopted to solve the system reliability optimization problem, and obtains a more optimal solution than other algorithms. In the application of GA, an effective linear scaling technique and a hybrid crossover operator are proposed, which improve the convergence performance of GA. The results of numeric calculation manifest that the problem is GAadaptive, and that the algorithm can converge nonexponentially.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
1998年第7期54-57,共4页
Journal of Tsinghua University(Science and Technology)
关键词
遗传算法
系统可靠性优化
可靠度分配
genetic algorithm (GA)
optimization of system reliability
reliability allocation
redundancy allocation
