Unreliable workstations or resources pose challenges in deadlock control for automated manufacturing system (AMS) since resources failures have negative impacts on system activities and make the situation more complex. There are two types of robust deadlock-free control policies, absorbing type and distributing type. Each of them has its advantage as well as its drawback. Based on two modified Banker's Algorithms (BAs), this paper develops a robust supervisory control policy for avoiding deadlocks in a class of AMSs with multiple unreliable resources. A state and the corresponding produced state are tested by the two BAs in two stages, respectively, in order. The state is judged to be a feasible state if and only if both the two tests can be passed. Our policy is of distributing type and capable of handling simultaneous multiple resource failures.
展开▼