A Federated Digital Twin platform for data collection, which is interoperable, multifactorial and on a citizen scale, to monitor and integrate tracking data with AI-based learning, analysis framework and inclusion of simulation methodologies, ‘what if’ analysis and predictive tools.

Federated Digital Twins can be applied to the interpretation and generation of actions according to user needs and the company and to sustainable goals, specifically oriented to the greening of cities and companies, from energy footprint to resource consumption.

Tools of this scope become useful for providing guidelines for decision makers, for user awareness and engagement, and for user acceptance of behavioural guidelines, best practices and impact assessment.

Many metropolitan cities constantly collect information about traffic flows through induction coils placed under the asphalt and/or other systems.

From historical data on traffic flows observed in a large city and made available as open data, the possibility of predicting traffic flows for the following hours has been tested.

Using a Digital Twin for air monitoring and urban-scale pollution forecasting allows you to produce maps that provide information about air quality and other environmental indices.

The instrument produces one-day spatial NO2, PM10, O3 concentration, air temperature, air humidity and BLQ-Air index values forecasts. The tool is developed in the Python programming language and is automated in order to operate every day, while the maps produced are displayed on the e-Globus platform, updated daily.

The results obtained indicate that the forecast output is well in agreement with the measurements observed.

Based on the actual production timing of the orders made in recent years and the bill of materials for each product, a proof-of-concept of a digital twin was developed which can forecast the time required to produce a new production batch according to various parameters.

The use of neural networks and deep learning algorithms can enable the digital twin to improve timing predictions using a large number of input variables and considering hidden links between variables by applying non-linear logic.

This allows you to:
• have more precise information for better scheduling of production orders;
• support the R&D area in the development of new product phases, simulating the time savings that can be obtained by replacing one component with another.