Using Generalized Stochastic Petri Nets for Preventive Maintenance Optimization in Automated Manufacturing Systems

以廣義隨機斐氏網路探討具串並聯特性自動化製造系統之最佳預防維護週期

張清靠(Ching-Kao Chang) ； 向振良(Chen-Liang Hsiang)

《品質學報》 18卷2期 (2011/04) Pp. 117-135

摘要

具串並聯特性的製造系統，其所組成機器的可靠度會相互影響，因此本文探討以廣義隨機斐氏網路建構該製造系統的複雜模型，以掌握各機器間加工變動情形，另配合週期性的檢測，以確認該系統在不同時間下各機器可靠度變化程度。並於最低維護成本準則及假設不同的維修將導致不同的性能恢復條件下，來決定整個系統最合適維護週期，及各階段所需執行的維護方式。續再以田口法驗證廣義隨機斐式網路方法，應用於決定複雜製造系統之維護策略的正確性。

關鍵字

串並聯 ； 可靠度 ； 廣義隨機斐氏網路 ； 維護成本 ； 田口法

並列摘要

This paper presents periodically preventive maintenance (PM) procedures for an automated manufacturing system (AMS) with series-parallel characteristics. The reliability of machines that comprise of above AMS will be influenced by each other. A system like this may develop latent failures and its reliability may vary, depending on scheduled checks and maintenance. Accordingly, we have modeled the deteriorated behavior of machines of AMS, and the effect of their reliability with PM activities, by using generalized stochastic Petri nets (GSPN) in order to measure the variation in reliability of those machines.
In this study, it is assumed that different levels of PM will lead to different degrees of functional recovery. Different PM are considered simultaneously in order to determine the PM schedule of a given system. Consequently, it can determine an optimal combination of each PM stage with the criterion that the total cost of system maintenance must be minimized. In this paper, we present a realistic example of annual maintenance scheduling for an AMS to demonstrate the effectiveness of our proposed methodology. Furthermore, Taguchi method is used to validate the correctness of the application of GSPN on determination of PM for a complicated system.

並列關鍵字

maintenance ； series-parallel ； generalized stochastic Petri nets GSPN ； Taguchi method