S. Cavalieri

Noise Prediction in Urban Traffic by a Neural Approach

The aim of this paper is to determine functional relationships between the road traffic noise and some physical parameters. Should this goal achieved, it is possible to modify the causes of traffic noise, in order to sensibly reduce it. Correlations are usually derived trough multiple regression analysis. In this paper an alternative solution based on the use of a neural approach is proposed. Its advantage is due to the capability of the neural networks to model non-linear systems such as the one treated in the paper. After an overview about the neural approach, the learning and production phase results are shown and discussed. They point out how good is the approach proposed to model noise pollution in urban areas.

Evaluating worst case response time in mono and multi-master profibus DP

Profibus DP networks are often used as the communication infrastructure for supporting distributed computer-controlled applications, which, very often, impose strict real-time requirements. The aim of this paper is to present a novel methodology for evaluating the worst-case message response time in Profibus DP networks where high and low priority information flows coexist.

Improving flexible semiconductor manufacturing system performance by a coloured Petri net-based scheduling algorithm

The paper describes an approach to scheduling in a flexible semiconductor manufacturing system. The application refers to the production of memories in a new module built in the SGS-Thomson plant in Catania (Italy). The approach uses Petri nets in which the concept of coloured marks has been introduced. Although the scheduling approach has been devised for a particular industrial application, it can be extended to other applications sharing similar features. The paper focuses on the scheduling approach and on the improvement of performance obtainable by its use.

A genetic algorithm for job-shop scheduling in a semiconductor manufacturing system

The paper describes an approach to solving a job-shop scheduling problem in a flexible semiconductor manufacturing system. The application refers to the production of memories in a new module built in the ST Microelectronics plant in Catania (Italy). The approach uses a genetic algorithm, which has been devised for the particular industrial application. The paper focuses on the scheduling approach and on how it improves performance of the plant.

A PN-based scheduler for a flexible semiconductor manufacturing system

The paper describes an approach to scheduling in a flexible semiconductor manufacturing system. The approach uses off-line scheduling as a basis for managing activities but integrates it with online scheduling whenever a modification in operating conditions requires a change in the scheduling previously drawn up. Petri nets (PNs) have been used as formal tool, for modelling and solving the scheduling problem.

Exploiting FDDI communication features to connect FieldBuses in process control environment

The authors deal with the problem of the information exchange between processes in extended fieldbus systems. As is known, although most of the traffic in these systems is within a single bus, there is always a certain amount transiting from one fieldbus to another, often covering a certain distance. There is therefore the problem of finding ways to interconnect the various fieldbuses in such a way as to meet the time constraints imposed by the applications. The paper proposes and then assesses a way of interconnecting even a large number of fieldbuses, introducing delays which are compatible with process control requirements.<<ETX>>

CAN assessment in time-critical cyclic applications through Petri net model

The authors discuss problems relating to use of the CAN (controller area network) fieldbus in time-critical process control applications. Certain characteristics of the CSMA/CD protocol, such as the reduced size of the frames with simplified headers and the use of an approach based on nondestructive collisions, allow optimal exploitation of the communication bandwidth. Moreover, the use of a priority-based approach to solve collisions provides a form of implicit scheduling, similar to the rate monotonic algorithm, which allows traffic to be organized correctly, according to the time constraints of the various variables. In the paper, after presenting the main features of the CAN, the problem of correct traffic scheduling is dealt with. A system model based on Petri nets is presented and evaluated. The results of the simulations performed show that CAN is suitable for several time-critical applications.

MULTI-MASTER PROFIBUS DP MODELLING AND WORST CASE ANALYSIS-BASED EVALUATION

This paper provides an analysis of the real-time behaviour of the multi- master Profibus DP network. The analysis is based on the evaluation of the worst- case message response time and the results obtained are compared with those present in literature, pointing out its capability to perform a more accurate evaluation of the performance of the Profibus network. Copyright

Petri nets and genetic algorithms to increase productivity in FMS

Full exploitation of available resources in a flexible manufacturing system is extremely important to optimize its productivity. The parameters on which to act in order to maximize exploitation of the resources are the sequence of activities performed by each resource and the number of parts processed in each production cycle (WIP). The aim of the paper is to propose a performance optimization strategy which takes both these parameters into account. The strategy is based on a genetic algorithm, whose aim is to explore as far as possible sequences of activities performed by each resource. For each sequence, a heuristic algorithm, the adjustment heuristic algorithm, is applied to find the minimum WIP. In this way, the sequence scenario which offers the minimum WIP is found.

Jitter-based policies to improve asynchronous bandwidth exploitation in fieldbus communication systems

An important problem in time-critical process control networks is to exploit the bandwidth fully and, at the same time, try to minimize the data delivery delay and serve all the stations fairly. The problem is particularly felt in fieldbuses which have to support traffic with hard time constraints. In this paper we present and evaluate some policies which, by allowing a controlled jitter on periodic traffic, allow the performance of asynchronous traffic produced by alarm signals, processing results, human operations, etc to be improved.