|
楼主 |
发表于 2015-7-8 14:13:24
|
显示全部楼层
asax 发表于 2015-7-7 18:02
我来解释一下,实体到达可以通过两种方法模拟:
一是到达间隔时间拟合,也就是楼上孙工所说的对到达间隔时 ...
谢谢版主的解释。我懂啦。我之前没注意在Arena里面到达率必须是用小时来做单位,所以产生的到达数总是比理论上少很多。我昨天也直接给David Kelton发邮件问了这个问题。然后他很友好的回复了我,跟版主解释的一样。我再这把他的回复贴出来供大家参考吧。
Ryan -- As the book states, it is a requirement of the Arena software that rates for nonstationary Poisson arrival processes be expressed (via the Schedule data module) in arrivals per HOUR, not per any other time unit. In Model 5-2, we think that the arrival rate could change every HALF hour, since we have data on that (for instance, counts of the number of calls in each half-hour period in the table, hopefully averaged over many days for each half-hour period over multiple days, for better estimates). So we'd take the average number of calls in each half-hour period and DOUBLE those values to transform into calls per HOUR, in order to arrive at the data in the table. For instance, in the first half-hour period from 8:00-8:30, the rate is given as 20 per HOUR, so that in this half-hour period we'd expect about 10 arrivals, in the next half-hour period from 8:30-9:00 we'd expect about 17.5 arrivals, etc.
If we instead thought that the arrival rate might change every 15 minutes, we'd multiply our observed counts by 4 to transform, again into arrivals per HOUR.
I trust this clarifies things for you -- David Kelton |
评分
-
查看全部评分
|