Evaluation of deadlock control designs in automated manufacturing systems

Abstract

Petri nets are an effective way to model, analyze, and control deadlocks in automated manufacturing systems (AMS). There are three important criteria in designing and evaluating a liveness-enforcing supervisor for a system to be controlled: behavioral permissiveness, structural complexity, and computational complexity. A maximally permissive supervisor can lead to high utilization of system resources. A supervisor with a simple structure can decrease the hardware and software costs. As for the computational complexity, means that a deadlock control policy can be applied to large systems. The objective of this paper is to design liveness-enforcing supervisors for different flexible manufacturing systems, simulate the controlled systems, and estimate the utilization of resources and throughput of the system. The siphon control methods (Strict Minimal Siphons and Elementary Siphons) are used to solve the deadlock control problems for a number of AMSs with different sizes. Moreover, the paper aims to evaluate the performance of selected methods such as utilization of resources, throughput, and the number of monitors, arcs, and states. Finally, the computational results indicate that the elementary siphons based policy provides better structural complexity and computational complexity than the strict minimal siphons based policy. However, strict minimal siphons based policy leads to better behavioral permissiveness than elementary siphons methods.