Meeting Artificial Intelligence and Machine Learning for Rail Passenger Service Planning under Competition

Meeting Artificial Intelligence and Machine Learning for Rail Passenger Service Planning under Competition

Funding EntityParticipantsDuration
Ministerio de Ciencia e Innovación (MCIN).
Agencia Estatal de Investigación.
• Universidad Politécnica de Cataluña.
• Universidad de Castilla-La Mancha.
• Universidad Rey Juan Carlos.
1/9/2021 — 31/08/2025

Part of the coordinated proyect "Analytics for Competitiveness and Seamless Mobility at Passenger and Freight Sustainable Transport".

Principal Investigator(s)

Research team:

Work team:

 

Abstract:

The efficiency, performance and environmental sustainability of transport systems depends on the infrastructure design, planning, management and operation as well as supply of services, being oriented all these elements to satisfy the expected levels and characteristics of the transportation demand. This research project is oriented to contribute methodologically to adapt the planning, management and operations to the deregulation and irregular demand scenarios caused by the pandemic situation to air, railway and urban mobility and interurban transport. Competing operators and pandemics, in combination, one announced through directives at European level and suddenly the other, demand that the planning and management models in these transport systems be adapted and extended in order to be useful in the evaluation of the measures and plans designed to face climate change and guaranteeing efficient and safe transport. Competitive scenarios arose earlier in air transport than in railway transport and in Spain the first European directives have only just begun to be considered.

Additionally, the White Paper for transport published by the EU in 2011 set the objective of transferring, by 2030, 30% of freight transport by road to rail or water transport, and 50% in 2050. Another aspect is that due to reduction in travel demand, the system of the fees that regulating entities charge to operators for using railway infrastructures needs to be revised. Another component is the behavior of users in irregular or pandemic in order to have acceptable demand predictions for these situations. The impact in air transport has made that the strategic planning of fleet dimensions and capacitydemand interrelations are critical factors for air companies. In the urban context new mobility concepts have recently emerged, (car-sharing systems and new micro-mobility modes among others) making necessary to study the behavior of users in pandemic situations and the impact on these modes.

This project is highly interdisciplinary and multidisciplinary. At the methodological level it will use: a) key Transport Engineering models for design, planning and management of the above mentioned transport systems and their interaction with demand models b) techniques from Operations Research such as mathematical programming techniques (with special emphasis on stochastic and robust optimization) either alone or combined/complemented with Machine Learning (ML) and artificial intelligence techniques and c) social simulation combined with activity-based demand models reinforced with ICT-based data sources. Scenarios of operators competence will be modelled using classical Nash equilibrium formulated as linear integer feasibility problems and the demand models will be embedded using either classical demand models or combined kernel logistic regression and Random Utility Maximization (RUM) models. Fleet management problems and demand-capacity interactions in air transport will be studied using robust/stochastic optimization. In the field of trajectory management, navigation and control systems in air transport, the project aims to hybridize information from ML and sensors and the design of control algorithms for vehicles based on information provided by ML algorithm. Also novel approaches based on optimal state observers, including of Kalman filters, will be developed.