Structural Control of Large-Scale Flexibly Automated Manufacturing Systems

Abstract

Current strategic and technological trends in discrete-part manufacturing require extensive and exible automation of the underlying production systems. However, even though a great deal of work has been done to facilitate manufacturing automation at the hardware component level, currently there is no adequately developed control methodology for these environments. In particular, it has recently been realized by the manufacturing community that any successful attempt towards the extensive automation of these environments requires the establishment o f logically correct and robust behavior from the underlying system. The resulting line of research is complementary to the most traditional performance-oriented control, and it has come to be known as the structural control of the contemporary exibly automated production environment. In this chapter, we initially discuss the role of structural control in the broader context of realtime control of exible manufacturing systems FMS, and in continuation, we present the major results of our research program on manufacturing system deadlock, currently the predominant structural control problem in the manufacturing research literature. In the concluding section, we also indicate how these results should be integrated in contemporary FMS controllers, and highlight directions for future research in the FMS structural control area.

Relative Value Function Approximation for the Capacitated Re-Entrant Line Scheduling Problem

“…However, in [11], it is also shown that in many practical cases (e.g., when the capacity of a pertinently selected set of buffers is greater than 1), the problem can be resolved in polynomial time through one-step look-ahead deadlock detection. For the remaining cases, one can either (i) employ polynomialcomplexity criteria / tests that will seek to identify a strongly connected component of the safe region that further contains the system empty state [3], or (ii) opt to ignore the complexity concern, and proceed to the identification of the entire safe region, by generating and trimming, with respect to the system initial empty state, the state transition diagram representing all the possible evolution of the buffer space allocation taking place in the underlying system; we refer the reader to [18], [3] for the algorithmic details. 2.…”

“…After having finished the processing of its current stage at a certain station, the part waits in its allocated buffer for transfer to the next requested station. Due to the finite buffering capacity, this transfer should be authorized by a structural control policy (SCP) [3] ensuring that (i) the destination workstation has available buffering capacity, and (ii) the transfer is safe, i.e., it is still physically possible from the resulting state to process all running jobs to completion.…”