Imagine several factories wanting to learn from each other's mistakes without sharing their secret recipes or private data. This system lets them train a shared AI brain that stays updated across different sites. It helps them figure out the perfect mix of raw materials and scrap to make high-quality metal while using less power.
Metallurgy companies struggle to reduce CO2 and energy use because they cannot share sensitive production data with other plants to improve AI models. This prevents them from finding the most efficient mix of raw materials and scrap.
A platform using Federated and Continual Learning. It includes data-driven and physics-based process models for optimizing furnace charges and Life Cycle Assessments.
If you are a steel mill operator dealing with high CO2 emissions and waste — this project developed an AI system that optimizes the mix of input materials and scrap recycling. This allows you to maintain product quality while reducing energy consumption and emissions.
If you are an auto parts manufacturer dealing with inconsistent material quality — this project developed a scalable AI tool for prognostic optimization. It ensures your parts match customer specifications while lowering the environmental footprint of production.
If you are a scrap processor dealing with variable material purity — this project developed a system that integrates scrap recycling into the steelmaking process. It helps determine the best charge mix to reduce waste generation.
Based on available project data, specific pricing or cost structures are not mentioned as this is a research and innovation action.
Yes, the project focuses on big European metallurgy industries and includes a complementary use case for automotive parts to ensure scalability and replicability.
Based on available project data, the IP and licensing terms are not specified, though the project involves a consortium of 11 partners including 7 industry players.
The system uses Federated Learning, which allows AI to learn from data across several factories without requiring the actual data to be shared or moved.
The project period runs until 2025-11-30, with the first integrated demo currently being prepared.
The consortium is heavily industry-led, with 7 out of 11 partners being industrial entities (64% ratio). This strong commercial presence, spanning 8 countries, suggests the technology is being developed with direct market needs in mind rather than purely academic interest. The inclusion of Atos as coordinator indicates a strong focus on the IT infrastructure and deployment side of the AI.
Contact ATOS IT SOLUTIONS AND SERVICES IBERIA SL
Contact us to explore how Federated Learning can decarbonize your metallurgy assets.
