Research Lines

WinTurCoM addresses the challenge of energy, environmental and green transition by maximising wind turbine reliability and lifespan. The line develops AI-based frameworks for resilient energy infrastructures, validated with real data from operational wind farms — reaching Technology Readiness Level 5 (TRL5) through active industrial collaborations under non-disclosure agreements with Ocean Winds and RAVE Alpha Ventus.

Key contributions include data-driven condition monitoring and structural health monitoring (SHM) of offshore support structures: transformer-based virtual sensing for dynamic-response reconstruction, bolt loosening identification, and SCADA-based early detection and prognosis of mechanical degradation in drivetrain components.

The forward work plan shifts from isolated asset monitoring to a population-based, data-centric framework — deploying graph neural networks to model wind farms as collaborative sensor networks, integrating explainable AI to translate latent features into physical failure modes, and progressing toward TRL6 through weekly quasi-operational deployment.

AIWinTurCoM · AEI

Deep & machine learning for predictive maintenance in wind turbines · 2022–2026 · €103,697

DTWinTurCoM · AEI

Digital twins for wind turbine condition monitoring · 2022–2024 · €181,815

STOR-HY · Horizon Europe

Resilience of EU energy grid · hydropower SHM · 2024–2028

The Control line focuses on modelling and designing feedback systems for stability and robustness, contributing to resilient industrial infrastructures aligned with the energy and green transition challenge. The group has validated advanced techniques including sliding mode control, model predictive control, and barrier-function-based adaptive approaches, while systematically enhancing classical PID controllers with mathematical rigour.

To facilitate global dissemination, the development of low-cost experimental platforms has been prioritised. A recent convergence with WinTurCoM has initiated the study of thermoplastic composites — designing robust control strategies for systems with embedded sensors and coupled thermal-mechanical dynamics, progressing from modelling toward TRL 4–5 validation via Hardware-in-the-Loop systems and physics-informed AI.

The line is structured around four complementary expertise nodes: AI-driven monitoring (Mujica), embedded hardware and feedback (Acho), piezoelectric energy harvesting for system autonomy (Pujol-Vázquez), and decentralised control strategies for large-scale multi-actuator systems (Rubió).

SNAPSHOT · AEI

AI-driven systems for manufacturing process & SHM of composites · 2025–2028 · €100,000

Unite Energy · Horizon Europe

European doctoral network on chemical energy storage via hydrogen · 2024–2027

E-ROTORS · Unite! Seed Fund

Industry 5.0 modules for Master's & PhD students · 2025–2026

The SHM line develops data-driven and AI-based methodologies for detecting, localising and assessing structural damage in engineering systems, with emphasis on real-world industrial deployment. The line bridges signal processing, machine learning and structural dynamics to provide robust monitoring solutions where direct inspection is costly or impractical.

Key contributions span multivariate statistical process control, principal component analysis-based damage indices, and neural network approaches for damage classification in composite structures, truss systems and mechanical components. A strong methodological thread connects the SHM line to WinTurCoM — sharing data-driven frameworks — and to the Control line through cyber-physical monitoring integration.

The line maintains active international partnerships through the Latin American Workshop on Structural Health Monitoring (LatamSHM), the European Workshop on SHM (EWSHM) and the International Workshop on SHM (IWSHM), with collaborative research spanning institutions in Colombia, Spain, USA and Europe.

SNAPSHOT · AEI

AI-driven SHM for manufacturing process & composites · PI: L.E. Mujica, L. Acho · 2025–2028 · €100,000

LatamSHM Initiative

Latin American Workshop on SHM — inaugural edition Cartagena 2023, second edition Santiago 2026

STOR-HY · Horizon Europe

Sensor-based condition monitoring for hydropower plants · 2024–2028

CellsiLab brings together experts from UPC and the Hospital Clínic of Barcelona / IDIBAPS, aligned with the Knowledge and Quality of Life challenge. Its dual objective is to develop computational models from microscopic images for automatic recognition of blood cells in peripheral blood, and to support clinical laboratory medicine through automated diagnosis of haematological diseases.

Since 2022, deep learning models have been developed for: detection of blast and atypical lymphoid cells in lymphoma and leukaemia; identification of neutrophil abnormalities; generation of synthetic cell images (SyntheticCellGAN) to address low-prevalence classes; and normalisation of image colouration across laboratories. A complete system — CellsiMatic — has been validated in a multicenter proof-of-concept study across five hospitals and is currently deployed on an industrial platform.

The forward plan focuses on consolidating CellsiMatic as a clinical decision-support tool, developing new models for MDS detection and CAR-T therapy monitoring, and advancing through iterative multicenter validation, data harmonisation using diffusion models, and technology transfer pathways with industrial partners including Sysmex R&D Center Europe.

xAI-HEALTH · AEI

Explainable deep learning in medical image analysis · UPC + Hospital Clínic + Hospital Sant Joan de Déu · 2024–2027 · €128,750

CellsiMaticPlus · AEI

Proof of concept & valorisation roadmap for leukaemia/lymphoma diagnosis · 2022–2024 · €90,000

Sysmex R&D Contract

AI models for MDS/LMMC detection & CAR-T monitoring · Sysmex R&D Center Europe · 2025–2027 · €50,000

43

JCR papers
since 2022

24

in Q1
journals

€843K

funding as
PI since 2022

20+

PhD theses
completed