Projects

ASTRAL

AI-Based Analysis and Monitoring of Multi-Scale Critical Systems

Funding: IFAB

Enabling technology: Artificial Intelligence, Big Data Analytics, LLM

Sustainable Development Goals

The project focuses on developing artificial intelligence algorithms for diagnostic, prognostic and health management solutions for complex cyber-physical systems. It responds to the growing demand for more accessible and reliable systems capable of operating under complex lifecycle conditions. The project also aims to develop digital twins of the cyber-physical systems involved. These digital replicas will provide data to the AI algorithms, supporting the continuous monitoring of system reliability and model accuracy.

Centro Nazionale di Ricerca in High Performance Computing Big Data e Quantum Computing (ICSC), progetto finanziato dall’Unione Europea – NextGenerationEU – e dal Piano Nazionale Ripresa e Resilienza (PNRR) – Missione 4 Componente 2.

Objectives

The project’s main objective is to develop advanced Predictive Health Management solutions for complex systems, with a particular focus on: anomaly detection; root cause analysis; estimation of Remaining Useful Life.

Initial Challenge

The project addresses a common challenge in predictive maintenance and complex system analysis: the detection of anomalies in time-series data, a crucial aspect of Predictive Health Management.

The main difficulty lies in the nature of the data available for training machine learning models, which tend to be heavily skewed towards either nominal or anomalous samples. This imbalance makes it difficult to apply supervised learning effectively using the original datasets.

Solution

Developing effective machine learning models for identifying anomalies in time-series data can significantly improve system reliability and maintenance processes. One of the key challenges is the management of imbalanced datasets dominated by nominal samples. The project addresses this issue through unsupervised learning approaches and the generation of synthetic anomalies. These methods are designed to improve the ability of machine learning models to detect and predict anomalous behaviour.

The project also aims to integrate these techniques with Root Cause Analysis processes in order to understand the origins of system failures and prevent their recurrence. Another crucial component is the use of artificial intelligence to estimate Remaining Useful Life, which is essential for asset management and maintenance planning.

Finally, the solutions are intended for deployment both in data centre environments and on edge computing devices. The machine learning models will therefore be optimised to operate effectively in resource-constrained environments. Overall, the project aims to develop advanced machine learning and AI technologies for a broad range of applications, improving the detection, analysis and prevention of failures in complex systems.

Benefits

The creation of a national network bringing together leading research centres will provide a clear direction for the implementation of health management and reliability solutions for mission-critical systems. The project also aims to build a secure platform capable of leveraging large volumes of data and signals through artificial intelligence and high-performance computing technologies, supporting applications in both space missions and industrial environments. In addition, it will strengthen national expertise in this field by building on experience gained in other industrial sectors, including weather platforms for large-scale data processing.

IFAB’s Role

IFAB participated in the project as a facilitator, helping to create the conditions required for its successful development. Its contribution focused on connecting complementary organisations, opening access to specialised expertise and resources, and establishing the operational and collaborative framework within which the project could take shape and progress.