S3PR的控制器合併理論之加強

Abstract

最近，創新和有計算效率的方法來設計最大活性結構簡單彈性製造控制系統已有所報導。但是，目前還不清楚是否反復的方法，可以實現最小的顯示器配置。早些時候，我們發現，在α-S3PR（簡單的順序過程與資源的系統。所有依賴虹吸管是是強烈地）所需的監視器的數量降到最低不能低於基本虹吸管數量的。這證實了在[]中兩個控制系統例子是一個最小的顯示器配置，因為他們屬於α-S3PR的，他們在每一個例子中的控制器數量等於基本信標數量。這項工作進一步探索了一條新的理論，使上述結果可以擴展到非-α-S3PR的。我們也提出了一個快速的方法來計算無可達性分析的關鍵活性標記及禁止標記（比上面的方法更快的速度及更少標記）。在此基礎上，我們提出了一個快速的方法來合併多個監視器成一個單一的監視器以實現簡單的配置而無須整數線性規劃（ILP）。然而，就Ezpeleta等的例子而言，我們無法實現如其他方法的5個顯示器，。我們將探索背後的理論，回答下面的問題： 1。理論上和在什麼情況下，我們的合併方法（基於虹吸）不能達到最低限度的配置嗎？ 2。在這種情況下，借用稍微扭曲的虹吸的方法，我們怎麼實現最低限度的配置呢？ 這項工作在保持我們在監控控制該領域的領先地位是非常重要的。 Recently, novel and computationally efficient methods to design maximally permissive liveness enforcing supervisors with simple structures for flexible manufacturing systems have been reported. However, it is unclear whether the iterative approach can achieve the minimal monitor configuration. Earlier we showed that the minimal number of monitors required cannot be less than that of basic siphons in α-S3PR (systems of simple sequential processes with resources) with strongly dependent siphons only. This confirms that two of the three controlled systems in [10] are of a minimal monitor configuration since they belong to α-S3PR and their number in each example equals that of basic siphons. This work further explores a new theory so that the above results can extend to non- α-S3PR. We also propose a fast way to compute critical live and forbidden markings (fewer and faster than the above methods) without reachability analysis. Based on that, we proposed a fast method to merge several monitors into a single one to achieve simple configuration without integer linear programming (ILP). However, for the benchmark by Ezpeleta et al., we were unable to achieve 5 monitors as with other approaches. In the ILP approach, when the number of monitors reaches the lower bound, the ILP can be exited to allow a faster solution. We propose to explore the theory behind to answer the following issues: 1. Is it theoretically true and under what condition, our merging method (siphon-based) cannot achieve minimal configuration? 2. In this case, how do we achieve minimal configuration by twisting the siphon-based approach? This work is important to maintain our leading position in the field of supervisor control.