FAIRE: Federated Artificial Intelligence for Remaining useful life Edge analytics
Revolutionising Industrial Operations with FAIRE: Federated AI for Predictive Maintenance
Author: Konstantina Tsioli, Pavlos Stavrou
February 20th 2025
At CORE Innovation Days in January, CORE unveiled a groundbreaking demonstration of FAIRE (Federated Artificial Intelligence for Remaining Useful Life Edge Analytics), a cutting-edge solution that combines AI, edge computing, and federated learning to address critical challenges in industrial operations.
This innovative approach not only enhances operational efficiency but also ensures data privacy and scalability, making it a game-changer for industries like manufacturing, energy, and pharmaceutical.
What is FAIRE
FAIRE is a ground-breaking solution based on the MODUL4R and RE4DY EU projects. FAIRE is a federated AI solution designed to optimise industrial processes by leveraging edge computing and federated learning.
It enables real-time data processing and predictive analytics, while keeping sensitive data secure and on-premise. FAIRE showcased how it can be applied to predictive maintenance for CNC machines, but its applications extend far beyond this use case.
Key FAIRE Features
Edge Computing: This solution utilises edge devices deployed directly on the shop floor to collect and process data locally. This reduces latency, minimises bandwidth usage, and ensures real-time insights without relying on constant cloud connectivity.
In the demo, two edge devices were connected to CNC machines, collecting data relevant to tool wear and predicting the Remaining Useful Life (RUL) of milling tools.
Remaining Useful Life (RUL): is a predictive tool that estimates the time left before a machine or component fails or requires maintenance, based on real-time data and historical performance patterns. In the context of FAIRE, the RUL model predicts tool wear in CNC machines, enabling proactive maintenance and reducing downtime while ensuring data privacy and security.
Federated Learning: FAIRE employs federated learning to enable collaborative intelligence across multiple machines or factories. Instead of sharing raw data, only model parameters (e.g., insights and updates) are sent to a central server, ensuring data privacy and compliance with regulations like GDPR. This approach allows machines to "learn" from each other, improving prediction accuracy and operational efficiency without compromising sensitive information.
Data Privacy and Security: By keeping data on-premise and sharing only model updates, FAIRE ensures that proprietary information remains secure. This is particularly important for industries with strict data protection requirements.
Scalability and Flexibility: FAIRE’s architecture is designed to scale effortlessly. As new machines or edge devices are added to the network, they can seamlessly integrate into the federated learning ecosystem, enhancing the system’s overall intelligence and resilience.
Predictive Maintenance for CNC Machines
The demonstration of FAIRE solution focuses on a real life application: predictive maintenance for CNC machines. Here’s how it worked:
Data Collection: Two edge devices were connected to two CNC machines, collecting real-time data on tool wear and machine performance using industrial protocols like OPC-UA and MQTT.
Local Processing: The edge devices preprocessed the data locally, running AI models to detect anomalies and predict RUL. Results were displayed on monitors, providing operators with actionable insights.
Federated Learning: Model updates from each edge device were aggregated to a central server to update the global model. The updated model was then sent back to the edge devices, enhancing their predictive accuracy.
Real-Time Insights: Operators then could monitor tool wear and RUL in real time, enabling proactive maintenance and reducing downtime.
The benefits of FAIRE
FAIRE offers numerous benefits for industrial operations:
Smarter Machines: Continuous learning and adaptation improve machine performance and operational efficiency.
Enhanced Data Privacy: Sensitive data remains on-premise, ensuring compliance with data protection regulations and/or requirements.
Cost Optimisation: Reduced data transmission and proactive maintenance minimise operational costs.
Collaborative Intelligence: Federated learning enables machines to learn from each other, improving model accuracy across the network.
Scalability: The solution can easily scale to include additional machines or factories, making it suitable for large industrial networks.
Application areas
While the demonstration of FAIRE solution involved an example of CNC machines, its capabilities extend to various industries:
Pharmaceutical: In a sector where protecting sensitive and production data is paramount, this solution safeguards data privacy and security.
Automotive: Enhance predictive maintenance for automotive production lines.
Aerospace: Improve the performance and reliability of aircraft components.
Energy and Smart Grids: Monitor and optimise power grid equipment like transformers and substations.
Mining: Optimise the operation of heavy machinery like excavators and drilling equipment.
