论文部分内容阅读
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。
Under the conditions of cost, volume and weight constraint, the reliability optimization problem of multi-stage series-parallel systems is a non-linear, mixed optimization problem with both integer and real variables with multiple local extremums. The genetic algorithm and conjugate gradient method are combined to search for the optimal solution that other algorithms can not obtain. In the application of genetic algorithm, effective linear scaling and mixed crossover are set up to improve the convergence performance of genetic algorithm. And based on the model theory, it is numerically indicated that the problem is in accordance with the assumption of water block. Finally, numerical results show that GA is polynomial convergent on this problem.