随机动态装卸混合问题的分区求解策略

Region partitioning policy for stochastic dynamic pick-up and delivery problem

针对需求服从一般分布的随机动态装卸混合问题,提出一种求解该问题的分区求解策略,分析了需求稀少和需求密集情况下该策略的渐近性.仿真比较了需求服从一般分布情形下分区求解策略、随机队列中位策略、多车场随机队列中位策略和堆栈策略的求解效果,以及需求服从一般分布和需求服从均匀分布情形下分区求解策略的求解效果.结果表明,对于需求服从一般分布的随机动态装卸混合问题,分区求解策略是一种有效的求解策略.

Region partitioning policy is proposed to solve stochastic dynamic pick-up and delivery problem with general demand distributions. The asymptotic properties of this policy in light and heavy traffic are analyzed respectively. Simulation is made to compare the solution effect of region partitioning policy, stochastic queue median policy, m stochastic queue median policy and stacker crane policy under general demand distributions assumptions. Another simulation is made to compare the solution effect of region partitioning policy between general and uniformly demand distributions. Results indicate that region partitioning policy is an effective policy to solve stochastic dynamic pick-up and delivery problem with general demands distributions.