基于Arena的车间作业排序问题建模方法及其仿真优化系统设计

Modeling & simulation optimization systems design for job shop scheduling based on Arena

针对车间作业排序问题的固有复杂性和目标函数难于解析求解等特点,建立了一个优化与仿真的集成系统框架,并提出了一种新的建模求解思路:首先,以通用仿真工具Arena为平台,提出虚拟抢占规则,实现了车间作业排序问题的仿真建模。然后,以贪婪随机自适应搜索算法为基础,结合高级语言VB,利用面向对象编程思想,通过Arena类库,设计了一个通用的车间作业排序问题的仿真优化系统框架,从而实现了优化和仿真的外部集成。在该框架下,可引入各种随机因素,提高对实际系统的建模与求解能力。最后,通过实例验证了该方法的有效性。

Considering the inherent complexity of job shop scheduling problems, especially its difficulty in analytical solution, an integration system framework based on optimization and simulation was established. A new modeling solution philosophy was proposed for the problems. By this way, a general-purpose and powerful simulator, Arena, was employed for modeling job shop problems firstly. And a kind of virtual preemption rule was recommended to improve the modeling capability. Then an efficient optimization procedure Greedy Randomized Adaptive Search Procedure （GRASP） was utilized to find the optimal solution. Based on Arena class library and its interface with VB, the interaction between Arena and GRASP could be effectively implemented, and an optimization ＆ simulation integrated framework was thus achieved. With this framework, various stochastic factors could be conveniently introduced into the model, hence the capability for modeling the real-life systems was greatly improved. Finally, an example was analyzed to demonstrate the effectiveness of the method.