ON THE ENHANCEMENT OF TIMED PETRI NET THEORY AND ITS APPLICATIONS TO SOME ASPECTS OF COMPUTER SYSTEMS

Abstract

This thesis is an attempt towards the development of a new Petri net extension, the Advanced Petri Wet. It is a more powerful modelling and analysis tool th^n conventional Petri 7)ets, which have some inherent limitations, such as lack of descriptive qualities and somewhat complex analytical representations. These limitations have prompted and motivated the author to attempt the postulation of this Advanced Petri "Met (APN). The APN possesses a greater amount of descriptive capability and analytical power as compared to conventional nets. The fundamental components of APN are places, transitions, input/output arcs and tokens as in the case of conventional nets. But, unlike in conventional nets, these components exhibit some special characteristics, which are described in detail. As a result of these features, simple linear algebraic techniques have been evolved for the analysis of the APN. The reachability, tree and the matrix equation, which are the popular analysis techniques of conventional nets, have been extended and modified suitably to reflect the features of the APN. Special emphasis has been eiven to time based analysis techniques for the APN. A new time characteristic equation has been devised to provide a direct mechanism through which performance evaluation of APN models can be conveniently carried out, giving the total execution time of the systems modelled by the net. i The APN has been found to be amenable for application in wide ranging areas like Super computer pipelines, Automated flexible manufacturing systems, Computer communication and computer controlled data acquisition systems. The thesis also includes development of a description language for the automation of analysis techiiques of Petri tists in general and of APN in particular. This description language has been developed on the same lines as the hardware description languages for computer hardware. This description language also helps in the standardization and generalization of Petri »ets.