Imagine if your industrial machinery could tell you it's going to break 24 hours before it actually happens. This project builds a smart 'health monitor' that lives directly inside the power hardware, rather than in a distant cloud server. It uses a mix of physics and AI to spot tiny warning signs in high-performance semiconductors, preventing sudden crashes and expensive downtime.
Industrial plants face high costs due to periodic maintenance shutdowns and the inherent reliability risks of modern GaN and SiC semiconductors. Current failure analysis often relies on cloud servers, which are too slow or resource-heavy for real-time embedded power systems.
An automated, cloudless fault-prediction system integrated into power converters. This includes compact AI models deployed on RISC-V and ARM-cortex architectures and high-accuracy sensors for real-time monitoring.
If you are an EV manufacturer dealing with unpredictable power module failures in high-voltage systems — this project developed AI models for ARM-cortex MCUs with ASIL-D certifications that provide real-time residual lifetime estimation. This ensures safety and reduces unexpected vehicle breakdowns.
If you are an equipment builder dealing with high maintenance costs for industrial drives — this project developed a short-term fault prediction system with a 12-24h horizon. This allows operators to schedule power-saving shutdowns during idle times without risking a failure during the next start-up.
If you are a component maker dealing with the reliability limitations of III/V-semiconductors — this project developed compact failure models and specialized sensors to monitor GaN and SiC devices. This adds a layer of intelligence to the hardware, making the components more attractive to risk-averse industrial buyers.
Based on available project data, specific pricing or implementation costs are not provided.
Yes, the project targets industrial drives and power converters, specifically aiming to lower exercise costs by reducing the need for periodic maintenance shutdowns.
Based on available project data, the IP and licensing terms are not specified in the provided summary.
The AI is deployed locally on resource-constrained embedded devices, specifically RISC-V-based systems for industry and ARM-cortex MCUs for automotive use.
The project period runs from 2023-09-01 to 2026-08-31, suggesting the technology is currently in development.
The consortium is heavily industry-driven with a 59% industry ratio, comprising 20 companies and 9 SMEs across 11 countries. This strong commercial presence, combined with 9 universities and 5 research centers, indicates a high focus on practical application and market integration rather than pure theoretical research.
Contact the University of Bologna (ALMA MATER STUDIORUM)
Contact us to connect with the R-PODID consortium for early adoption of AI-driven power electronics monitoring.
