ARTICLE IN PRESS JID: INS [m3Gsc;February 12, 2016;2:27] Information Sciences xxx (2016) xxx–xxx Contents lists available at ScienceDirect Information Sciences journal homepage: www.elsevier.com/locate/ins Robust deadlock control for automated manufacturing systems with an unreliable resource Yunchao Wu, Keyi Xing ∗ , Jianchao Luo, Yanxiang Feng Q1 State Key Laboratory for Manufacturing Systems Engineering and Systems Engineering Institute, Xi’an Jiaotong University, Xi’an 710049, China article info Article history: Received 7 September 2014 Revised 22 November 2015 Accepted 24 January 2016 Available online xxx Keywords: Automated manufacturing system (AMS) Robust control Deadlock Petri net Unreliable resource abstract So far, most of deadlock control policies are proposed based on the assumption that au- tomated manufacturing systems (AMSs) have no unreliable resources. While in real man- ufacturing systems, resource failure is inevitable and will reduce the number of available resources. This paper focuses on the robust deadlock control problem for AMSs with an unreliable resource. Petri nets are used to model the unreliable systems, and a subclass of the ordinary and conservative Petri nets, known as system of simple sequential process with resources (S 3 PR), is studied. A resource failure and recovery net is added to describe the resource failure and recovery. To prevent each siphon from being emptied, the concept of constraint set for a strict minimal siphon is introduced. By limiting the number of to- kens in each constraint set, a robust deadlock controller is devised. It is proved that our controller can guarantee the liveness of the controlled system no matter one resource fails or not. Finally, some examples are provided to illustrate the validity of the proposed robust deadlock controller. © 2016 Elsevier Inc. All rights reserved. 1. Introduction 1 Deadlock-free resource allocation has been an active research area of automated manufacturing systems (AMSs) for the Q2 2 several past decades. Researchers have made tremendous progress in computational, structural and control aspects of the Q3 3 deadlock-free resource allocation problem [1–5,7–9,13–15,17–20,22,23,25–27,29]. Three kinds of strategies, deadlock preven- 4 tion, deadlock avoidance, and deadlock detection and recovery, are usually used to handle deadlock problems [17,29]. Dead- 5 lock prevention policies [2–5,8,14,18–20,22] usually use offline computational mechanisms to design a controller to ensure 6 that deadlocks never occur. While in deadlock avoidance policies [26,27], a state is monitored online and resource is granted 7 to process only if the resulting state is safe. In deadlock detection and recovery policies, however, resources are granted to a 8 process without any check, the statuses of resource allocation and request are examined periodically to determine whether 9 the system is in deadlocks, and if a deadlock is found, the system recovers from it by aborting one or more deadlocked 10 processes. 11 Until now, most of deadlock control policies prevent deadlocks on condition that resources in considered systems are 12 assumed not to fail, whereas this assumption is unrealistic for most real manufacturing systems. In case of resource failure, 13 the existing deadlock control policies are always no longer in force and deadlocks may occur, then the system has to stop 14 ∗ Corresponding author. Tel.: +86 13571897199. E-mail addresses: wuyunchao@stu.xjtu.edu.cn (Y. Wu), kyxing@mail.xjtu.edu.cn (K. Xing), luo770826336@stu.xjtu.edu.cn (J. Luo), fengyxss@stu.xjtu.edu.cn (Y. Feng). http://dx.doi.org/10.1016/j.ins.2016.01.049 0020-0255/© 2016 Elsevier Inc. All rights reserved. Please cite this article as: Y. Wu et al., Robust deadlock control for automated manufacturing systems with an unreliable resource, Information Sciences (2016), http://dx.doi.org/10.1016/j.ins.2016.01.049