穩定狀態再生模擬的多變量統計評估

Multivariate Statistical Analysis for Regenerative Simulation

陳得發(Der-Fa Robert Chen)鄭文仁(Wen-Jen Cheng)

《管理科學學報》 6卷1期 (1989/05) Pp. 43-56

https://doi.org/10.6504/JOMS.1989.06.01.05

摘要

應用模擬方法來研究系統的行為時，常常因為模擬結果，資料之間彼此有相關性，而使其平均值的變異數膨脹或縮小，因此求得的信賴區間不可靠。
本篇論文是著利用再生法來處理多便亮的模擬結果，希望克服資料相關的問題。試驗的模擬模式計有M/M/1單站等候模式、Tandem queue串聯等候模式、Central Server Computer Model中央服務者計算機模式，(s,S)存貨管理模式。

並列摘要

The autocorrelation among output data from a simulation study makes the variance inflated or shrinked. Therefore the confidence interval established this way is on longer correct. In this paper, we tried to develop a regenerative method to overcome the autocorrelation problem for a multivariate steady –state simulation. The models we used to test the performance of this method are M/M/l queue, tandem queue, central server computer model and (s,S) inventory model.