Ph.D. received on: 8/7/1997

E-mail: simona@dist.unige.it

Tutor: Prof.  Riccardo Minciardi, Department of Communication, Computer, and System Sciences, University of Genova
 
 

 ___________________________________________________________________________________________________________

Performance optimization of manufacturing systems represented as Petri nets   ___________________________________________________________________________________________________________

Advisor:

Prof. Riccardo Minciardi, Department of Communication, Computer, and System Sciences, University of Genova

Summary:

An effective approach to the problem of modelling, analysing, and optimizing manufacturing systems performance is proposed in this thesis; such an approach is based on the use of Petri nets, a specific modelling formalism suitable for complex systems. In particular, both the transient and the steady-state behaviour of production systems are studied by exploiting some interesting results relevant to a particular class of Petri nets, the Timed Event Graphs (TEGs).
 The manufacturing system model here considered and represented by means of TEG's is a very general model including assembly/disassembly operations, the possibility of performing the same operation on different machines, machines able to implement different operations with the relevant set-up times, and machines provided with input and output buffers characterized by limited or unlimited sizes. For such a model, the main performance indices to be taken into account are the system productivity and the cost of the system work-in-progress, whereas the decisional aspects to be defined are relevant to the dimensions of lots of products for each operation and each machine, the initial system configuration, the buffer sizes, and the working sequences of each machine.
 The proposed model, some basics on Petri nets, and the way of representing the considered model through TEG's, are the subjects of the first chapters of the thesis. Then, the decisional problems typically faced when dealing with manufacturing systems are reported, and a survey is provided about the existing works coping with Petri nets applied to production systems modelling, analysis, and control. Finally, the optimization problems stated and solved during the Ph.D studies are presented in details and discussed, thus giving a deep insight on the main innovative aspects of the thesis.
 The first optimization problem considered regards the maximization of the system productivity and contemporary minimization of the cost of the system work-in-progress; such a problem has been solved through a lexicographic procedure. Great attention is here devoted to the definition of an innovative algorithm to solve the mixed continuous-integer optimization problem which constitutes the first step of the overall procedure. Then, the same problem is faced also for production processes including a material handling system. In this case, the multiobjective optimization problem comprehends also the optimal dimensioning of the material handling system and it is again solved through a lexicographic procedure. Finally, the manufacturing systems behaviour, both in the transient and in the steady-state, is analyzed by means of a new algebraic structure (the Max-Plus algebra), which is ad-hoc defined to allow a thorough study of the dynamic evolution of discrete-event systems. It is to be noted that the problems stated in this work and the corresponding solution algorithms have all been experimentally tested; the most significant results are reported and discussed throughout the thesis.
 

MINOR

Modelling, simulation, and control of freeway systems

Advisor: Prof. Ezio Volta, University of Genova

The main objective of this research work regards the possibility of exploiting hybrid modelling methodologies to represent the dynamic behaviour of freeway systems. The designed model is composed of a discrete-time modelling level, which represents the regular system functioning, and of a discrete-event framework describing exceptional conditions perturbing the regular behaviour of the system. On the basis of the proposed model, which has also been validated using real data, suitable control strategies aimed at solving traffic congestion situations have been designed and tested by means of an ad-hoc realized simulation tool.
 

 _______________________________________