Distributed fuzzy control of flexible manufacturing systems

Abstract

A flexible manufacturing system (FMS) consists of a set of machines that are connected via tracks over which parts may be transported from one machine to another for processing. As parts arrive at a machine via the tracks, they are put in a buffer (queue) where they are held before they are processed. There is a local controller (scheduling policy) at each machine which uses the machine's buffer levels to decide which part type to process next; hence the overall controller for the FMS is physically distributed across the entire FMS with local schedulers at each machine. In this paper the authors show how to design a fuzzy controller for a single machine and show via simulation that its performance is comparable to conventional schedulers. In addition, the authors introduce an adaptive fuzzy controller which can automatically synthesize itself (or tune itself if there are machine parameter variations) to achieve good throughput rates for the single machine as compared with conventional schedulers. Next it is shown via simulations that by using such adaptive fuzzy controllers in a distributed fashion, one obtains a distributed fuzzy controller (DFC) which can automatically synthesize itself and lower the maximum buffer level more effectively than conventional schedulers. Finally, the authors illustrate the ability of the DFC and conventional schedulers to automatically tune themselves in case there are unpredictable machine parameter changes in an FMS. These final results show that while sometimes the DFC performs in a superior fashion, better scheduling policies are needed to guarantee high performance FMS operation in case there are unpredictable machine parameter changes.