EDITS

Enhancing Digital Twins with Scalable and Interpretable AI

F.R.S.-FNRS WEL-T Investigator Programme · 2025–2029 · 48 months

WEL-T ULB MLG

Project Overview

EDITS is a research project funded by the F.R.S.-FNRS WEL-T Investigator Programme (Call 2024), running from 2025 to 2029. The project aims to integrate recent advances in Artificial Intelligence and Machine Learning into the Digital Twin domain, focusing on scalable data processing, adaptive learning, multi-step-ahead forecasting, and causal reasoning.

Led by Prof. Gianluca Bontempi at the Machine Learning Group (MLG), Université libre de Bruxelles (ULB), EDITS addresses open challenges in Digital Twin technology by developing AI/ML methodologies for calibration of physics-based models, learning high-dimensional surrogate data-driven models, causal interpretation, big data streaming analytics, and forecasting.

EDITS Vision

Research Aims

Infrastructure

Big Data Analytics

Design a big data analytics infrastructure to support Digital Twin systems in real-time data streaming and processing.

Calibration

Physics-Based Model Calibration

Develop data assimilation and calibration techniques to calibrate physics-based models with large variate data streams.

Transfer Learning

Foundational Models

Define a foundational model approach for transferring knowledge across similar configurations and physical systems.

Causality

Causal Discovery & Inference

Apply causal inference and discovery to add interpretability and counterfactual reasoning to surrogate models.

Forecasting

Simulation & Forecasting

Validate methodology through multi-variate multi-step-ahead forecasting with uncertainty characterization.

Validation

Case Studies

Three real-world case studies: IoT devices, TrafficTwin for smart cities, and EnergyTwin for wind farms.

Explore case studies →

At a Glance

147

Man-Months of research effort across 6 Work Packages

6 Work Packages

From infrastructure to validation, covering the full DT pipeline

View WPs →

3 Case Studies

IoT, Traffic, and Energy — progressively increasing in scale

View case studies →