摘要
一、问题的提出 复杂随机服务系统的优化或规划设计问题,因系统和运行环境的复杂性及随机性,一般都采用计算机模拟技术。如果把N个随机服务系统组成一个复杂的排队网络系统,用计算机模拟也能得到比较好的解决。可是,当每个子系统的状态都比较多(设i子系统的状态数为S_i),因为模拟技术不能直接求最优解这一固有缺陷,要确定系统的最优运行状态就必须把所有各子系统不同状态之间的匹配情况都进行模拟(这便需要对整个系统模拟S_1 S_2…S_N次),然后再比较各次模拟结果,从中选出最优解。如果系统的状态数较大或一次模拟花费的机时较长,则S_1 S_2…
Although computer simulation can describe the complexity and randomness of an operating system well, it can not solve the optimal solution of this system directly. The optimal soiution is usually obtained by comparing a lot of simulation results, so it requires long compiler time. On the other hand, although mathematical programming technique can not describe the comptexity and randomness well, it can solve the optimal solution directly. Depend on the charateristics of a queuing network system, this paper puts forward a new method combining the two techniques to solve the optimal design and plan of this system. From the view of theory and application, thes method is worth being studied and extended.
出处
《系统工程理论与实践》
EI
CSCD
北大核心
1992年第6期38-45,共8页
Systems Engineering-Theory & Practice