Procedura fuzzy per la scelta degli interventi di ripristino delle condizioni di sicurezza di una rete stradale

A lot of problems of transport engeneering are to face ambiguity, vagueness, subjectivity: • subjective valuations usually affect both a driver’s route choice and the mode of transportation among concurrent systems; • the valuation of the choice of the planoaltimetrical lay-out, of the level of service, of safety standards in relation to drivers’ perceptions and reactions, depend on the subjective judgment of road designers. Even when a stochastich and not deterministic approach is chosen the traditional methods of decision on above matters are based on binary logic, which, because of its inner nature, cannot effectively deal with vagueness, ambiguity and subjectivity. Fuzzy logic is by far more adherent to the peculiarities of these problems, and its recent development has permitted a lot of interesting achievements in transportation engeneering, both during the planning phase and during its operation regulating. The analysis of the infrastructural safety component of a road network, the choice of accident – prone locations, the individuation of possible improvement techniques issues, the rating of related costs without any detailed project, the assesment of benefits, the costs of safety improvement and driving confortare problems which may be successfully solved by fuzzy sets theory tecniques. The basic problem is determining the kind of countermeasure to be feasibly carried out within the budget. The safety costs cannot be defined deterministically and benefit of alternatives cannot be measured without fuzziness. When evaluating certain alternatives, factors such as feeling of safety, driving comfort, etc. must be taken into account. This research working aims to formalyze a “clustering fuzzy method for the accident - prone locations in an infrstructural network of the Salerno Province, where more than 300 accidents - prone locations have been spotted. We have previously realized that the risk analysis of recorded accidents is not reliable, when made on a stocastic base both because of lack in data base, where only accidents causing grave personal injures are generally included and even because the really dangerous events are relatively rare in comparison with the total of the accident-prone situations: all that led us to research for the danger rate through the reading of the perceptions of qualified users. The infrastructural cause of dangerousness, which has been perceived by correlating appropriate variables of each point such as longitudinal slope, road paving condition, urbanization degree, speed limit, flow regulation, has been identified through the “fuzzy clustering” algorithms.