Specific Objectives

1 - Prescriptive Asset Management

·         Develop AI-based (Machine Learning in particular) reliable and interpretable tools and methodologies aiming at prescribing actions in intelligent railway assets management exploiting both data and domain knowledge taking into consideration the related uncertainty;

·         Study and develop the metrics and methods/tools to assess the accuracy of developed prescriptive analytics solutions;

2 - Multi-Objective Decision Optimisation

·         Enhance IAMS with advanced mathematical modelling and AI tools by adding:

o   Optimal asset management strategies subject to domain constraints;

o   Optimal maintenance planning and scheduling policies;

o   Sustainable alternative multi-modal transport options during rail unavailability.

·         Propose and evaluate a variety of state-of-the-art (including AI-based) optimisation algorithms to address jointly infrastructure maintenance planning and maintenance operations optimisation, identifying the optimal share between complexity and accuracy.

3 - Advanced context-driven Human Machine Interfaces (HMIs)

·         Survey and evaluate current state-of-the-art context-driven HMI approaches to be used by IMs to support decision-making processes also borrowing from more advanced industrial sectors (e.g. finance and aviation).

·         Define and design HMIs leveraging on AI and multi-objective optimisation unlocking the potential of IMs to:

o   Appropriately access the models and visualise prescribed options;

o   Understand why and how the model predicts or prescribes something (“opening the black-box”) via explainable AI (XAI);

o   Steer the model to comply with IMs’ preferences;

o   Assess alternatives through speculative execution.