非相同元件并联的S-P网络可靠性优化研究

Research on Reliability Optimization of S-P Network with Non-identical Components in Parallel

研究非相同元件并联的单目标-单约束S-P网络系统可靠性优化问题。选择常用的群体智能算法,包括模拟退火算法、蚁群算法、遗传算法、粒子群优化算法对模型求解。通过模拟仿真发现:几种算法给出的最优解情况不尽相同,蚁群算法求解精度高、收敛率100%,但执行时间长、解的编码复杂、解空间搜索复杂;粒子群优化算法收敛性较好,最优解的收敛率比较高;遗传算法搜索到最优解数量较少,收敛率比较低;模拟退火算法也能收敛到最优解,但收敛率较低,优点是容易实现。选择算法时,要依据问题的规模、时间、收敛率与精度进行选择。

The research is about non- identical components in parallel S -P network system reliability optimi- zation problems. To choice some intelligence algorithm commonly used, including simulated annealing algorithm, ant colony algorithm, genetic algorithm, particle swarm optimization algorithm to solve the model. Through the simulation found that the optimal solution of several algorithms are not the same, ant the algorithm has high precision and convergence rate of 100%, but the execution time is long, the encoding solution is complex, to search the solution space is complex. Particle swarm optimization algorithm has better convergence, the convergence rate is relatively high. To search the optimal solution genetic algorithm is less in quantity, the convergence rate is relatively low. Simulated annealing algorithm can converge to the optimal solution, the convergence rate is low, but it is easy to realize.