Work Packages

Big Data Infrastructure for System Monitoring

Months 1–48 · 24 Man-Months

Design and implement a big data analytics infrastructure to support Digital Twin systems. This includes data collection methodologies, high-performance computing setup, and data analysis frameworks. The infrastructure adopts a Lambda architecture with speed, batch, and serving layers for real-time and historical data processing, with message queuing for loose coupling between components.

Calibration of Physics-Based Models

Months 1–30 · 27 Man-Months

Develop data assimilation techniques (Kalman filters, statistical methods) for calibrating physics-based models with large variate data streams. This work package also covers surrogate model learning through adaptive machine learning, integration of physics-based and data-driven models, and multi-task machine learning approaches.

DT Foundational Models & Transfer Learning

Months 19–30 · 12 Man-Months

Define foundation models for Digital Twins and develop transfer learning approaches for model adaptation. The goal is to enable knowledge reuse across similar physical systems, reducing the need for full retraining when applying DT methodologies to new configurations.

Causal Discovery & Inference

Months 13–48 · 24 Man-Months

Infer causal models from multivariate data to add interpretability to surrogate models. This includes counterfactual reasoning capabilities and validation of causal interventions, enabling "what-if" scenario analysis and understanding of cause-effect relationships within the Digital Twin.

Forecasting & Simulation

Months 13–42 · 24 Man-Months

Enable what-if scenario analysis through multi-variate multi-step-ahead forecasting with uncertainty characterization. This work package focuses on visualization of scenario impacts and robust prediction under various operating conditions.

Case Studies Validation

Months 1–48 · 36 Man-Months

Validation through three progressively complex case studies:

• WP6.1 (M1–M12): Small-scale IoT devices — Arduino-based robotic arm, wind turbine, smart farm mockup
• WP6.2 (M12–M36): TrafficTwin — mobility Digital Twin for smart cities with simulation, calibration, and 3D visualization
• WP6.3 (M19–M48): EnergyTwin — wind farm Digital Twin with real-time data, forecasting, and maintenance prediction