Kevin抛砖引玉之“如何验证模型的有效性”

1# Kevin


Flexsim美国官网上有一个用户建模的经验总结，可以借鉴一下。

https://www.flexsim.com/lessons-from-a-massive-model/

1# Kevin


原帖回复地址链接：https://www.flexsim.com/community/forum/showthread.php?t=3126

Before building a simulation model the important thing is to prepare objectives to clarify what your simulation model wants to achieve and to guide the customer for implementation overtime.If we construct a model that constitute all particulars of the real world and build everything corresponding with your operation,then the model will become too complex and even it complicate the results.The goals should be for the model and not for the real world.

For example some one wants to know that can we increase throughput with the same resources deployed? How extent that batch size should be decreased that will reduce the amount of time spent in the process and decrease the lead time?There are even other factors you might consider like how to release the production orders and do we need to track the inventory of other material(brought from outside) other than the main product(build inside)? Is there human intervention that makes decisions?

If you want to run different scenarios that might comprise things like cycle time,nr of operators,breakdown time,batch sizes and inter-arrival time.You need to know what variables that you must consider.There are just a lot of factors that need to be considered yet to create a reasonable model that can solve a problem.

About what statistics to choose it is upto the modeler to decide what to choose that depends upon the objectives of the model?Good luck!:-)


在开始建模之前，很重要的一件事情就是明确一些你的模型想要达成的目标，以此类指引用户具体的实施。如果我们建立的模型要包括真实世界的所有部分及其相关的所有运作，那么模型将会变得过于复杂，甚至使结果也复杂化。目标应该基于模型而非现实世界。
例如，某人想要知道在同样的资源部署下，我们是否能够提高产出？如何缩小批次数量可以减少流程中所花费的总时间和提前期？还有很多你可能考虑的因素，诸如如何安排产品订单以及我们是追踪其他（外购）原料，还是主要产品（自产）的库存情况？是否有认为的干涉影响决策？
如果你想要尝试包括加工时间、操作员数量、故障时间、产品批次和到达间隔在内的不同方案。你需要清楚哪些变量你必须考虑到。可以解决难题的可靠模型有非常多需要考虑的地方。
关于选择哪些数据是根据建模者指定的模型目标而决定的。祝好运~