HAL will be down for maintenance from Friday, June 10 at 4pm through Monday, June 13 at 9am. More information
Skip to Main content Skip to Navigation

Supervision in Multi-Modal Transportation System

Abstract : Without any doubt, modern multimodal transportation systems are vital to the ecological sustainability and the economic prosperity of urban agglomerations, and in doing so to the quality of life of their many inhabitants. Moreover it is known that a well-functioning interoperability of the different modes and lines in such networked systems is key to their acceptance given the fact that (i) many if not most trips between different origin/destination pairs require transfers, and (ii) costly infrastructure investments targeting the creation of more direct links through the construction of new or the extension of existing lines are not open to debate. Thus, a better understanding of how the different modes and lines in these systems interact through passenger transfers is of utmost importance. However, acquiring this understanding is particularly tricky in degraded situations where some or all transportation services cannot be provided as planned due to e.g. some passenger incident, and/or where the demand for these scheduled services deviates from any statistical long term-plannings. Here, the development for and integration of sophisticated mathematical models into the operation of such systems may provide remedy, where model-predictive supervision seems to be one very promising area of application which we consider here. Model-predictive supervision can take several forms. In this work, we focus on the model-based impact analysis of different actions, such as the delayed departure of some vehicle from a stop, applied to the operation of the considered transportation system upon some downgrading situation occurs which lacks statistical data. For this purpose, we introduce a new stochastic hybrid automaton model, and show how this mathematically profound model can be used to forecast the passenger numbers in and the vehicle operational state of this transportation system starting from estimations of all passenger numbers and an exact knowledge of the vehicle operational state at the time of the incident occurrence. Our new automaton model brings under the same roof, all passengers who demand fixed-route transportation services, and all vehicles which provide them. It explicitly accounts for all capacity-limits and the fact that passengers do not necessarily follow efficient paths which must be mapped to some simple to understand cost function. Instead, every passenger has a trip profile which defines a fixed route in the infrastructure of the transportation system, and a preference for the different transportation services along this route. Moreover, our model does not abstract away from all vehicle movements but explicitly includes them in its dynamics, which latter property is crucial to the impact analysis of any vehicle movement-related action. In addition our model accounts for uncertainty; resulting from unknown initial passenger numbers and unknown passenger arrival flows. Compared to classical modelling approaches for hybrid automata, our Petri net-styled approach does not require the end user to specify our model's many differential equations systems by hand. Instead, all these systems can be derived from the model's predominantly graphical specification in a fully automated manner for the discrete time computation of any forecast. This latter property of our model in turn reduces the risk of man-made specification and thus forecasting errors. Besides introducing our new model, we also develop in this report some algorithmic bricks which target two major bottlenecks which are likely to occur during its forecast-producing simulation, namely the numerical integration of the many high-dimensional systems of stochastic differential equations and the combinatorial explosion of its discrete state. Moreover, we proof the computational feasibility and show the prospective benefits of our approach in form of some simplistic test- and some more realistic use case.
Document type :
Complete list of metadata

Contributor : Abes Star :  Contact
Submitted on : Wednesday, March 8, 2017 - 10:47:06 AM
Last modification on : Saturday, April 16, 2022 - 3:14:13 AM
Long-term archiving on: : Friday, June 9, 2017 - 12:50:53 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01419126, version 3


Simon Theissing. Supervision in Multi-Modal Transportation System. Modeling and Simulation. Université Paris Saclay (COmUE), 2016. English. ⟨NNT : 2016SACLN076⟩. ⟨tel-01419126v3⟩



Record views


Files downloads