Various techniques have been developed and assessed in order to perform traffic predictions, however, most of the past researches on this subject has focused mainly on tools based on the traditional and well established physical-mathematical models, while the testing and evaluation of the most recent data-driven approaches have not yet been addressed in a systematic manner.

This thesis, both from a personal interest in the subject of transport planning and mobility, deepened during the course of my studies, tries at least in part to fill that gap, focusing on the evaluation of a DST for incident detection and traffic forecasting, which uses artificial neural network techniques. The work undertaken in order to achieve the results discussed in these pages was partly held during a four months internship conducted at the Birmingham City Council, partner of the Opticities Project under which the examined DST has been developed. The activities carried out during this collaborative experience were part of the in-itinere phase of the project. They consisted of the tool experimentation as well as the collection of the feedback provided by the end-users participating in the trials, through the administration of a highly structured questionnaire.

The achieved outcomes, presented in this dissertation together with the methodology followed to obtain them, are promising and indicate the importance of the development of DSTs based on sophisticated short-term traffic prediction models. In fact, they can not only effectively support the traffic management but also lead traffic operators and other key stakeholders to a better understanding of the network behaviour and its performance. As a consequence, the integration of such technologies in the traffic management centres could allow rethinking the traditional reactive and undifferentiated mitigating actions, which can take place only after the disruption occurrence, to face the mobility issues in a more proactive way, by implementing preventive strategies tailored to the current needs. A brief introduction to the structure of the thesis will be presented below. The first chapter describes the European transport policy, which represents the background of the Opticities Project. Moreover, the UK’s approach to ITS will be framed, with the aim to allow a better understanding of the particular historical and legislative context in which the considered DST is developed.

The second chapter instead, focuses on Decision Support Systems, discussing the theoretical and technological framework from which they have evolved, their state of the art and the future prospects. In the third chapter, the potentiality of DSTs’ employment in traffic management and, more in detail, in the short-term traffic provision, are discussed. Two typologies of forecasting techniques (mesoscopic modelling and ANNs) are presented and, in order to provide examples of their real applications, the Opticities project and the DSTs implemented within it in Birmingham and Lyon are introduced. The fourth chapter enunciates the thesis objectives and defines the methodological structure followed to achieve the predefined goals, describing in detail all the steps that have made it up. The results of the analysis performed to validate the DST performances and evaluate the perception of the users about it are finally presented and discussed in the fifth chapter.