The passengers' flow and station dwell time estimation are important tasks for mass transit planning. However, classical methods are difficult to apply into some practical achievements. This paper presents a new approach that models passengers' flow and its effect on passenger alighting and boarding time in mass transportation systems in the presence of uncertainties. The applied technique combines origin destination matrices approach with the application of artificial intelligence. This new approach allows the inclusion of some intuitive knowledge provided by a fuzzy logic inference motor to predict the flow demand of passengers' trips, alighting and boarding time passenger cars in explicit stations.

In this work a direct method to measure the stability of metro system lines with respect to a previously constructed time schedule is presented. For this purpose we first model saturation effects using a real time discrete space state representation and then apply a Lyapunov-based stability analysis considering time delays of trains as disturbances. As a result we have been able to define a new set of indexes that relate time delays with the validity of the actual time schedule when falling inside a particular 'stability area'. Results obtained in a simulated environment show that the new stability indexes are able to evaluate quantitatively and qualitatively the effects of saturation in metro lines as well as predict the need for rescheduling

En este trabajo, se propone un nuevo índice basado en el método directo de Lyapunov para el diseño de un algoritmo de reprogramación en tiempo real para líneas de metro. En este estudio se utiliza una versión modificada de un modelo de espacio de estados en tiempo real discreto, que considera los efectos de saturación en la línea de metro. Una vez que el modelo de espacio de estados se ha obtenido, el método directo de Lyapunov se aplica con el fin de analizar la estabilidad del sistema de la línea de metro. Como resultado de este análisis no sólo se propone un nuevo índice de estabilidad, sino también la creación de tres zonas de estabilidad para indicar el estado actual del sistema. Finalmente, se presenta un nuevo algoritmo que permite la reprogramación del calendario de los trenes en tiempo real en presencia de perturbaciones medianas.

No abstract