基于基本信標的自動制造系統(tǒng)活性控制器設計
發(fā)布時間：2018-12-12 03:45
【摘要】：柔性制造系統(tǒng)是由計算機數(shù)控機床和物料傳輸系統(tǒng)構成的可以高效地生產(chǎn)中小批量不同產(chǎn)品的自動制造系統(tǒng)。死鎖是設計柔性制造系統(tǒng)控制器時必須考慮和解決的問題。在有些情況下,死鎖造成的局部或整體的系統(tǒng)停頓并不單純地降低生產(chǎn)率,更可能造成重大經(jīng)濟損失甚至災難性后果。因此,柔性制造系統(tǒng)中死鎖的描述、分析及控制對系統(tǒng)控制的實現(xiàn)和運行至關重要。目前,基于Petri網(wǎng)的自動制造系統(tǒng)死鎖處理方法大致分為三種策略：死鎖檢測與恢復、死鎖避免和死鎖預防。死鎖檢測與恢復策略是一種在線檢測及恢復死鎖狀態(tài)的機制,一旦檢測到死鎖發(fā)生,便通過在線策略重新分配共享資源使系統(tǒng)恢復到正常狀態(tài)。死鎖避免是一種在線資源分配機制,通過在線策略合理分配系統(tǒng)資源使系統(tǒng)不能進入死鎖狀態(tài)。死鎖預防策略通過離線機制控制系統(tǒng)對共享資源的使用請求,保證死鎖不會發(fā)生。 一般地,死鎖預防可以通過為網(wǎng)系統(tǒng)添加控制器來預防死鎖發(fā)生。衡量Petri網(wǎng)活性控制器的性能指標主要分為結(jié)構復雜性、計算復雜性及行為許可性三個方面。信標作為表征Petri網(wǎng)結(jié)構特性的重要對象,與Petril網(wǎng)系統(tǒng)的死鎖密切相關,這一點在普通Petri網(wǎng)和一般Petri網(wǎng)中均成立。為了獲得最大許可行為的、結(jié)構簡單的及計算復雜度較低的Petri網(wǎng)活性控制器,許多研究者提出了大量基于信標分析控制技術的死鎖控制策略。因此,Petri網(wǎng)中信標的提取及分析方法是系統(tǒng)死鎖控制中一種非常重要的技術手段。由于Petri網(wǎng)信標求解的固有復雜性,任何依賴完全枚舉信標的死鎖預防算法在理論上必然為指數(shù)復雜度。本文立足于Petri網(wǎng)的信標分析及基本信標提取技術,研究柔性制造系統(tǒng)的活性Petri網(wǎng)控制器設計方法。主要完成以下工作： 1.行為許可性是衡量Petri網(wǎng)活性控制器性能的重要指標之一。針對一類普通Petri網(wǎng)-S3PR網(wǎng)系統(tǒng),提出了關于初始標識的S3PR網(wǎng)最大許可行為活性控制器存在性的充分條件�；诨拘艠死碚�,提出了最大許可行為活性監(jiān)督控制器存在性的判定算法。該算法基于基本信標與信標復合的計算,關于網(wǎng)規(guī)模為多項式復雜度。在S3PR網(wǎng)中,一定存在可以得到最大許可行為的活性控制器的初始標識。同時,結(jié)合給出的充分條件,提出了S3PR網(wǎng)最大許可行為活性控制器迭代設計算法。在每一步迭代中,該算法運用混合整數(shù)規(guī)劃的死鎖檢測方法獲得一個信標并加以控制。大量實例驗證了該方法的有效性。本研究可視為對基本信標理論的進一步應用。 2.在柔性制造系統(tǒng)中,基于信標的死鎖控制往往面臨計算復雜性和結(jié)構復雜性問題。為了降低控制器的結(jié)構復雜度,基本信標理論最初成功地應用于普通Petri網(wǎng)中。但考慮到在一般Petri網(wǎng)系統(tǒng)中資源分配具有更為復雜的情況,基本信標的選取及其可控性的分析仍需要進一步改進。為了改進基本信標在一般Petri網(wǎng)中的擴展應用,針對一類一般Petri網(wǎng)-WS3PR網(wǎng)系統(tǒng),首次引入了擴展信標的概念。擴展信標是一個能夠反映其原始信標中資源復雜使用的多集�？紤]WS3PR網(wǎng)系統(tǒng)的拓撲結(jié)構和權值信息,提出擴展基本信標概念,并給出了改進的基本信標定義,通過該定義可以得到一組與系統(tǒng)結(jié)構緊密相關的基本信標集合。同時,基于圖論,在WS3PR網(wǎng)中引入初始資源加權有向圖,并提出了結(jié)構同構網(wǎng)系統(tǒng)的概念。對于結(jié)構同構的不同網(wǎng)系統(tǒng),可以共用一組基本信標。 3.定義了一類一般Petri網(wǎng)-GLS3PR網(wǎng)系統(tǒng),該網(wǎng)系統(tǒng)可以較好地建模柔性制造系統(tǒng),同時具有便于分析的結(jié)構特性。針對GLS3PR網(wǎng)提出了一種高效的嚴格極小信標和基本信標提取方法。通過結(jié)構分析引入了擴展信標概念,并由此改進了基本信標定義�；趫D論,在普通網(wǎng)中可利用初始資源有向圖來計算所有的嚴格極小信標。本文針對GLS3PR網(wǎng)系統(tǒng),引入初始資源加權有向圖。在滿足一定的約束條件下,從初始資源加權有向圖中求取一組約束導出子圖�？紤]GLS3PR網(wǎng)結(jié)構,利用這些子圖,建立了提取所有嚴格極小信標及其擴展信標的方法。此外,根據(jù)信標和約束導出子圖之間的關系,給出了擴展基本信標個數(shù)的上確界。大量實驗結(jié)果表明,通過該方法獲得的基本信標結(jié)構緊湊且更適用于GLS3PR網(wǎng)系統(tǒng)。 4.為了改善活性控制器的結(jié)構復雜性和行為許可性,針對一類一般Petri網(wǎng)-S4PR網(wǎng)系統(tǒng),提出一種有效的死鎖預防策略。通過分析一般Petri網(wǎng)系統(tǒng)發(fā)現(xiàn),一般Petri網(wǎng)系統(tǒng)的死鎖可以通過非充分標識信標來反映,而基本信標理論的提出為設計結(jié)構簡單的活性控制器提供了可行方向。由此,在S4PR網(wǎng)系統(tǒng)中將非充分標識信標劃分為基本信標和從屬信標,從屬信標的可控性可以通過對其基本信標的合理控制來實現(xiàn)。為了獲得一組結(jié)構緊湊且適用的基本信標集合,通過結(jié)構分析在S4PR網(wǎng)系統(tǒng)中引入擴展基本信標概念,然后結(jié)合松弛的信標最大可控性條件來滿足信標的可控性。該方法僅對基本信標添加控制器,即可獲得活性受控網(wǎng)系統(tǒng)。此外,通過設計控制庫所輸出弧撤弧算法來獲得許可行為更優(yōu)的活性控制器。
[Abstract]:The flexible manufacturing system is an automatic manufacturing system which is composed of a computer numerical control machine tool and a material transmission system, which can efficiently produce small and small batches of different products. Deadlock is a problem that must be taken into account and solved when designing a flexible manufacturing system controller. In some cases, the local or global system pause caused by the deadlocks does not simply reduce the productivity, and may result in significant economic losses and even catastrophic consequences. Therefore, the description, analysis and control of the deadlock in the flexible manufacturing system is of great importance to the realization and operation of system control. At present, the deadlock processing method of automatic manufacturing system based on Petri net is divided into three strategies: deadlock detection and recovery, deadlock avoidance and deadlock prevention. The deadlock detection and recovery strategy is a mechanism of on-line detection and recovery of the deadlock state. Once the deadlock has been detected, the system is re-allocated to the normal state through the on-line policy reallocating the shared resource. The deadlock avoidance is an on-line resource allocation mechanism, and the system resources are reasonably distributed through the online policy so that the system can not enter the deadlock state. The deadlock prevention strategy guarantees the use request of the shared resource by the off-line mechanism control system, and ensures that the deadlock does not occur. In general, deadlock prevention can prevent deadlocks by adding a controller to the network system The performance index of the Petri net active controller is mainly divided into three parties: structure complexity, computational complexity and behavior licensing. As an important object for characterizing the structure of the Petri net, the beacon is closely related to the deadlock of the Petri net system. In order to obtain the maximum permissible behavior, the structure is simple and the computational complexity is lower, and many researchers have put forward a lot of deadlock control methods based on the beacon analysis control technology. Therefore, the extraction and analysis method of the object of the Petri net is a very important technical hand in the system deadlock control segment. Because of the inherent complexity of the Petri net beacon solution, any deadlock prevention algorithm that relies on the full enumeration of the beacon must theoretically be complex in theory In this paper, based on the beacon analysis of the Petri net and the basic beacon extraction technology, the design party of the active Petri net controller of the flexible manufacturing system is studied. Method. The following work is mainly completed 1. The behavior permissibility is an important means to measure the performance of the Petri net's active controller. In this paper, for a class of common Petri net-S3PR network system, the existence of the maximum permissible behavior controller of the S3PR network with initial identification is proposed. In this paper, based on the basic beacon theory, the existence of the maximum permissible behavior monitoring controller is put forward. The algorithm is based on the calculation of the basic beacon and the beacon composite, and the network size is a polynomial. Complexity. In the S3PR network, there must be the beginning of the active controller which can get the maximum permissible behavior. The initial identification is given. At the same time, the iterative setup of the maximum allowable behavior controller of the S3PR network is put forward in combination with the sufficient conditions. in each iteration, the algorithm uses a deadlock detection method of mixed integer programming to obtain a beacon and add to control. A large number of instances validate the method Effectiveness. This study can be considered as an entry to the basic beacon theory Step application. 2. In the flexible manufacturing system, the deadlock control based on the beacon is often faced with the computational complexity and structure complexity problem. In order to reduce the structural complexity of the controller, the basic beacon theory is successfully applied to the ordinary Pe in that tri-network, the analysis of the selection of basic beacon and its controllability still need to be given in view of the more complex resource allocation in the general Petri net system In order to improve the extended application of the basic beacon in the general Petri net, the extension of the general Petri net-WS3PR network system is introduced for the first time. the concept of a beacon. The extended beacon is a complex that can reflect the resources in its original beacon Considering the topological structure and weight information of the WS3PR network system, the concept of extended basic beacon is put forward, and the improved basic beacon definition is given. At the same time, based on the graph theory, the initial resource weighted digraph is introduced in the WS3PR network, and the structure isomorphic network is proposed. The concept of a system. For different network systems with a homogeneous structure, one can share one. Group basic beacon. 3. A class of general Petri net-GLS3PR network system is defined, which can be used to model the flexible manufacturing system. The structure characteristics of the analysis are analyzed. An efficient and strict minimum beacon and base are proposed for GLS3PR network. The method for extracting the beacon comprises the following steps of: introducing an extended beacon concept through a structure analysis, Basic beacon definition. Based on graph theory, the initial resource digraph can be used to calculate all the basic beacon. The system of GLS3PR is introduced in this paper. The resource-weighted digraph is obtained from the initial resource-weighted digraph under certain constraint conditions. Group-constrained derived sub-graphs. Taking into account the GLS3PR network structure, using these subgraphs, all strictly minimal beacons are established and all the strictly minimal beacons are extracted. The method of expanding the beacon. In addition, according to the relation between the beacon and the constraint-derived sub-graph, the extended basic signal is given. A large number of experimental results show that the basic beacon obtained by this method is compact and more suitable for GL The S3PR network system. 4. In order to improve the structure complexity and behavior permissibility of the active controller, a kind of general Petri net-S4PR network system is put forward. In this paper, by analyzing the common Petri net system, the deadlock of the general Petri net system can be reflected by the non-sufficient identification of the beacon, while the basic beacon theory is the active control with simple design structure. in that S4PR network system, the non-fully identify beacon is divided into a base beacon and a slave beacon, and the controllability of the slave beacon can be in ord to obtain a set of basic beacon which are compact and suitable, an extended basic beacon concept is introduced in a S4PR network system through a structure analysis, the method only adds a controller to the basic beacon, in addition, that license line is obtain by the arc-back-arc algorithm output by the design control library,
【學位授予單位】：西安電子科技大學
【學位級別】：博士
【學位授予年份】：2013
【分類號】：TH165;TP273
本文編號：2373843
[Abstract]:The flexible manufacturing system is an automatic manufacturing system which is composed of a computer numerical control machine tool and a material transmission system, which can efficiently produce small and small batches of different products. Deadlock is a problem that must be taken into account and solved when designing a flexible manufacturing system controller. In some cases, the local or global system pause caused by the deadlocks does not simply reduce the productivity, and may result in significant economic losses and even catastrophic consequences. Therefore, the description, analysis and control of the deadlock in the flexible manufacturing system is of great importance to the realization and operation of system control. At present, the deadlock processing method of automatic manufacturing system based on Petri net is divided into three strategies: deadlock detection and recovery, deadlock avoidance and deadlock prevention. The deadlock detection and recovery strategy is a mechanism of on-line detection and recovery of the deadlock state. Once the deadlock has been detected, the system is re-allocated to the normal state through the on-line policy reallocating the shared resource. The deadlock avoidance is an on-line resource allocation mechanism, and the system resources are reasonably distributed through the online policy so that the system can not enter the deadlock state. The deadlock prevention strategy guarantees the use request of the shared resource by the off-line mechanism control system, and ensures that the deadlock does not occur. In general, deadlock prevention can prevent deadlocks by adding a controller to the network system The performance index of the Petri net active controller is mainly divided into three parties: structure complexity, computational complexity and behavior licensing. As an important object for characterizing the structure of the Petri net, the beacon is closely related to the deadlock of the Petri net system. In order to obtain the maximum permissible behavior, the structure is simple and the computational complexity is lower, and many researchers have put forward a lot of deadlock control methods based on the beacon analysis control technology. Therefore, the extraction and analysis method of the object of the Petri net is a very important technical hand in the system deadlock control segment. Because of the inherent complexity of the Petri net beacon solution, any deadlock prevention algorithm that relies on the full enumeration of the beacon must theoretically be complex in theory In this paper, based on the beacon analysis of the Petri net and the basic beacon extraction technology, the design party of the active Petri net controller of the flexible manufacturing system is studied. Method. The following work is mainly completed 1. The behavior permissibility is an important means to measure the performance of the Petri net's active controller. In this paper, for a class of common Petri net-S3PR network system, the existence of the maximum permissible behavior controller of the S3PR network with initial identification is proposed. In this paper, based on the basic beacon theory, the existence of the maximum permissible behavior monitoring controller is put forward. The algorithm is based on the calculation of the basic beacon and the beacon composite, and the network size is a polynomial. Complexity. In the S3PR network, there must be the beginning of the active controller which can get the maximum permissible behavior. The initial identification is given. At the same time, the iterative setup of the maximum allowable behavior controller of the S3PR network is put forward in combination with the sufficient conditions. in each iteration, the algorithm uses a deadlock detection method of mixed integer programming to obtain a beacon and add to control. A large number of instances validate the method Effectiveness. This study can be considered as an entry to the basic beacon theory Step application. 2. In the flexible manufacturing system, the deadlock control based on the beacon is often faced with the computational complexity and structure complexity problem. In order to reduce the structural complexity of the controller, the basic beacon theory is successfully applied to the ordinary Pe in that tri-network, the analysis of the selection of basic beacon and its controllability still need to be given in view of the more complex resource allocation in the general Petri net system In order to improve the extended application of the basic beacon in the general Petri net, the extension of the general Petri net-WS3PR network system is introduced for the first time. the concept of a beacon. The extended beacon is a complex that can reflect the resources in its original beacon Considering the topological structure and weight information of the WS3PR network system, the concept of extended basic beacon is put forward, and the improved basic beacon definition is given. At the same time, based on the graph theory, the initial resource weighted digraph is introduced in the WS3PR network, and the structure isomorphic network is proposed. The concept of a system. For different network systems with a homogeneous structure, one can share one. Group basic beacon. 3. A class of general Petri net-GLS3PR network system is defined, which can be used to model the flexible manufacturing system. The structure characteristics of the analysis are analyzed. An efficient and strict minimum beacon and base are proposed for GLS3PR network. The method for extracting the beacon comprises the following steps of: introducing an extended beacon concept through a structure analysis, Basic beacon definition. Based on graph theory, the initial resource digraph can be used to calculate all the basic beacon. The system of GLS3PR is introduced in this paper. The resource-weighted digraph is obtained from the initial resource-weighted digraph under certain constraint conditions. Group-constrained derived sub-graphs. Taking into account the GLS3PR network structure, using these subgraphs, all strictly minimal beacons are established and all the strictly minimal beacons are extracted. The method of expanding the beacon. In addition, according to the relation between the beacon and the constraint-derived sub-graph, the extended basic signal is given. A large number of experimental results show that the basic beacon obtained by this method is compact and more suitable for GL The S3PR network system. 4. In order to improve the structure complexity and behavior permissibility of the active controller, a kind of general Petri net-S4PR network system is put forward. In this paper, by analyzing the common Petri net system, the deadlock of the general Petri net system can be reflected by the non-sufficient identification of the beacon, while the basic beacon theory is the active control with simple design structure. in that S4PR network system, the non-fully identify beacon is divided into a base beacon and a slave beacon, and the controllability of the slave beacon can be in ord to obtain a set of basic beacon which are compact and suitable, an extended basic beacon concept is introduced in a S4PR network system through a structure analysis, the method only adds a controller to the basic beacon, in addition, that license line is obtain by the arc-back-arc algorithm output by the design control library,
【學位授予單位】：西安電子科技大學
【學位級別】：博士
【學位授予年份】：2013
【分類號】：TH165;TP273
【參考文獻】
相關期刊論文 前1條
1 李志武,王安榮;基于Petri網(wǎng)的柔性制造系統(tǒng)一種預防死鎖方法(英文)[J];自動化學報;2003年05期
，本文編號：2373843
本文鏈接：http://sikaile.net/kejilunwen/jixiegongcheng/2373843.html
教材專著
