A DISTRIBUTED MULTI-CRITERIA APPROACH FOR TRAFFIC REGULATION IN PUBLIC TRANSPORTATION SYSTEMS

Given the dynamic character of public transportation systems, it is difficult to respect accurately theoretical timetables. In real conditions, many disturbances may occur in transportation networks and cause troubles in vehicle schedules. In order to keep transportation systems as stable as possible, a real-time traffic regulation has to be performed by optimizing some regulation criteria that represent exploitation objectives. In this article, we deal with the regulation problem as a multi-criteria optimization one for which we propose a nonaggregative approach based on multi-agent systems. For a given disturbance, the approach firstly ensures an anytime generation of Pareto solutions set by using a distributed Tabu search. Then, the best compromise solution is determined through a multi-criteria evaluation process. Therefore, two multi-criteria evaluation procedures are also suggested: Plurality voting procedure and fuzzy-based procedure. In order to assess the distributed approach, an experimental study was performed on the base of real scenarios.