Robust supervision using shared-buffers in automated manufacturing systems with unreliable resources q Hao Yue a,b,c , Keyi Xing b , Hesuan Hu d , Weimin Wu a,⇑ , Hongye Su a a The State Key Laboratory of Industrial Control Technology and the Institute of Cyber-Systems and Control, Zhejiang University, Hangzhou, Zhejiang 310027, PR China b The State Key Laboratory for Manufacturing Systems Engineering and the Systems Engineering Institute, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, PR China c Lab of Granular Computing, Minnan Normal University, Zhangzhou, Fujian 363000, PR China d School of Electro-Mechanical Engineering, Xidian University, Xi’an, Shaanxi 710071, PR China article info Article history: Received 6 October 2013 Received in revised form 11 January 2015 Accepted 29 January 2015 Available online 9 February 2015 Keywords: Deadlock avoidance Robust control Automated manufacturing systems Control policy abstract It has been an active area of research to solve the modeling, analysis, and deadlock control problems for automated manufacturing systems (AMSs). So far, all the system resources are assumed to be reliable in most of the existing approaches for deadlock-free and nonblocking supervisory control. However, many resources of AMSs are subject to failure in the real world. In order to develop a more practical and appli- cable supervisor, this work takes into consideration of multiple unreliable resources in a class of AMSs. On the basis of two variants of Banker’s Algorithm, this paper presents a robust supervisory control policy to avoid deadlock and blocking in these systems. The policy tries to make the best use of buffers of the shared resources to achieve the control objectives. Our controller is qualified to handle simultaneous multi-resource failures. By using formal language and automata theory, we establish its correctness. Moreover, our proposed method is verified via an AMS example, and we make comparison studies between our policy and some of the other similar type of policies in the literature. Ó 2015 Elsevier Ltd. All rights reserved. 1. Introduction Automated manufacturing systems (AMSs), which have evolved for decades, are computer-controlled production facilities adapt- able to variable production plans and goals. Recently, modern information technology has been increasingly and extensively applied to AMSs research. On the basis of the up-to-date tech- nology in information science and engineering, researchers addressed a number of issues regarding modeling, simulation, con- trol, and scheduling for AMSs (Ahmad, Huang, & Wang, 2011; Choi & Ko, 2009; Ferrarini & Piroddi, 2008; Hsueh, 2010; Huang, Pan, & Zhou, 2012; Huang, May, Wu, & Huang, 2013; Jeng, Xie, & Peng, 2002; Roszkowska & Reveliotis, 2013; Wang & Wu, 1998; Wu, Zhou, & Li, 2008; Xing, Han, Zhou, & Wang, 2012; Yalcin & Namballa, 2005). As a consequence, this field develops rapidly, and manufacturers can benefit much in reducing cost, increasing productivity, and improving products’ quality so as to meet the global market competition. Supervisory control is often treated in a logical domain. It aims to constrain system behavior to the legal scope (including dead- lock-free zones). Therefore, supervisory control is one of the funda- mental needs to establish and maintain AMSs in manufacturing engineering theory and applications. Generally, control logic is for- mally modeled by using formalisms such as automata and Petri nets. With the help of the previously set up system model, system designers or control engineers develop control software to meet all kinds of desired requirements such as deadlock-free operation. Finally, executable control codes are generated and the control goals are fulfilled. Actually, the whole procedure demonstrates a typical successful application of state-of-the-art supervisory con- trol theory to contemporary manufacturing systems. AMSs usually exhibit high degrees of process concurrency, resource sharing, and resource competition. The resources in AMSs include computer numerically controlled machine tools, workstations, fixtures, robots, automated guided vehicles, and other material-handling devices. Various parts entering the system compete for limited resources. As a result, this may in turn lead to dead states or deadlocks (Fanti & Zhou, 2004; Li, Wu, & Zhou, 2012) under improper control or supervision. Deadlock is a highly http://dx.doi.org/10.1016/j.cie.2015.01.028 0360-8352/Ó 2015 Elsevier Ltd. All rights reserved. q This manuscript was processed by Area Editor Manoj Tiwari. ⇑ Corresponding author at: The State Key Laboratory of Industrial Control Technology and the Institute of Cyber-Systems and Control, Zhejiang University, No. 38, Zheda Road, Hangzhou, Zhejiang 310027, PR China. Tel.: +86 571 87952369. E-mail addresses: yuehao1980@mail.xjtu.edu.cn (H. Yue), kyxing@sei.xjtu.edu. cn (K. Xing), huhesuan@gmail.com (H. Hu), wmwu@iipc.zju.edu.cn (W. Wu), hysu@iipc.zju.edu.cn (H. Su). Computers & Industrial Engineering 83 (2015) 139–150 Contents lists available at ScienceDirect Computers & Industrial Engineering journal homepage: www.elsevier.com/locate/caie