WinTurCoM
CoDAlab Research Line
WinTurCoM
Wind Turbine Condition Monitoring & Structural Health Monitoring with AI
Since 2009, WinTurCoM has developed mathematical models and artificial intelligence to detect, diagnose and predict faults and structural damage in wind turbines. Our methods are validated on operational data from real wind farms, onshore and offshore, to raise availability and cut operation and maintenance costs.
Research line active since
2009
National competitive funding as PI since 2015
€660,118
5 projects
Institutions
Universitat Politècnica de Catalunya (UPC)
Projects coordinated with IKERLAN
Lead and Co-PI
Yolanda Vidal
Francesc Pozo
Challenge
Energy, Environmental & Green Transition
Flagship project
PhIE-WinTurCoM
AEI, 2026-2030
WinTurCoM brings together applied mathematicians, engineers and data scientists from UPC, industrial partners and international universities. We work where mathematical modelling meets operational wind farm data: SCADA, condition monitoring systems (CMS), vibration and strain measurements.
Our goal is to move wind energy from corrective and time-based maintenance to fully predictive maintenance. We cover the full chain: fault detection, diagnosis, prognosis and remaining useful life (RUL), for both rotating components (main bearing, gearbox, generator, pitch) and support structures (offshore jacket foundations, blades).
Validated on real data
Our strain virtual sensor for offshore jacket supports was validated with data from the Alpha Ventus offshore wind farm, and our SCADA-based early fault detection methods have been tested on in-production wind farms, predicting failures months in advance.
Areas of expertise
Team
Current members are listed on the Group Members page. Former members: Bryan Puruncajas (PhD 2023), Diego A. Sandoval (PhD 2021), Christian Tutivén (PhD 2018, now external collaborator).
Meet the team →AI for wind turbine health, from data to decisions
We build data-driven and physics-informed models that learn the healthy behaviour of a wind turbine and flag deviations early.
Many of our methods need only healthy data, so they can be deployed in wind farms where no failures have been recorded. Most use only existing SCADA data, a low-cost option for older turbines without dedicated monitoring systems.
Methods. Deep Learning, Autoencoders and Variational Autoencoders, Transformers, Physics-Informed Neural Networks, Explainable AI, Conformal Prediction, Contrastive Learning, Reinforcement Learning, Virtual Sensing, Health Indices.
What we monitor
Data and infrastructure
Work over time, 2022-2026
Selected outputs by year:
Gearbox fault detection under variable speed. A speed-robust unsupervised detector (Energy and AI).
Explainable generator fault detection. Early detection using only SCADA data from an operational wind farm (IET Renewable Power Generation).
SCADA-vibration fusion. Fusion with contrastive learning (Structural Control and Health Monitoring).
Reinforcement learning pitch control. Control with double deep Q-networks (ISA Transactions).
International review. State of the art of drivetrain condition monitoring (Wind Energy Science).
Open science platform. A platform for structural health monitoring (Wind Energy Science).
Health index monitoring. Variational autoencoders; first prize in the WeDoWind ASCE-EMI Structural Health Monitoring Challenge (Wind Energy Science).
Strain virtual sensor. For jacket supports, with Alpha Ventus data (Mechanical Systems and Signal Processing).
Reviews. SCADA-based monitoring, and prognostics and health management (Reliability Engineering & System Safety).
Rotor imbalance detection. Unsupervised, output-only vibration data analysis (Energy and AI).
Bearing degradation assessment (Advanced Engineering Informatics).
Gearbox early fault detection. Mel-frequency cepstral coefficients (Structural Control and Health Monitoring).
Bolt-loosening detection. In jacket supports, with an open dataset (Structural Health Monitoring, Data in Brief).
Gearbox monitoring. Industrial SCADA and vibration data (Energy Reports).
Transformer models. Damage localisation in jacket supports (Structural Control and Health Monitoring).
Main bearing anomaly detection (Wind Energy Science).
AI for wind turbine condition monitoring. Overview (Energies).
GRU-based early main bearing fault detection. With SCADA data (IEEE/ASME Transactions on Mechatronics).
Siamese networks. For jacket damage diagnosis (Mathematics).
Next steps
Within PhIE-WinTurCoM (2026-2030), we will build a population-based, fleet-level health management framework that combines physics-informed and explainable AI, health indices and uncertainty-aware decisions, moving towards TRL 6 through quasi-operational deployment with IKERLAN.
Funded projects of the line between 2015 and 2030, drawn on a common time axis. PI: Yolanda Vidal; co-PI: Francesc Pozo, unless stated otherwise. Total competitive national funding as PI since 2015: €660,118 (5 projects).
Funded research projects
PhIE-WinTurCoM
2026 - 2030
Physics-Informed and Explainable AI for Wind Turbine Condition Monitoring and Structural Health Monitoring. Fleet-level, population-based health management of wind farms; coordinated with IKERLAN.
Funded by: Agencia Estatal de Investigación, PID2025-168502OB-C21 (€143,375)
AIWinTurCoM
2022 - 2027
Artificial Intelligence for Wind Turbine Condition Monitoring and Structural Health Monitoring. Deep learning for predictive maintenance and early structural damage detection; coordinated with IKERLAN.
Funded by: Agencia Estatal de Investigación, PID2021-122132OB-C21 (€103,697)
DTWinTurCoM
2022 - 2025
Digital Twins for Wind Turbine Condition Monitoring (Ecological and Digital Transition call).
Funded by: Agencia Estatal de Investigación, TED2021-129512B-I00 (€181,815)
Deep learning for wind turbine fault diagnostics
2020 - 2022
Industrial contract. Comparative study of deep learning for fault diagnosis on a worldwide wind farm database.
Funded by: CDTI contract with ITESTIT S.L., Smartive (€50,000)
FloWinTurCoM
2018 - 2022
Intelligent monitoring, pitch and structural damping control for floating offshore wind turbines; coordinated with IKERLAN.
Funded by: Agencia Estatal de Investigación, DPI2017-82930-C2-1-R (€96,800)
Fault detection and fault-tolerant control for offshore wind energy plants
2015 - 2018
Fault detection and fault-tolerant control strategies; coordinated with IKERLAN.
Funded by: Agencia Estatal de Investigación, DPI2014-58427-C2-1-R (€134,431)
Doctoral projects
Shun Wang
PhD Candidate, EAWE PhD Scholar. Thesis defence: 23 November 2026
AI for wind turbine condition monitoring (compendium thesis)
Jokin Cuesta
Industrial PhD, UPC / IKERLAN. Thesis defence: 20 November 2026
Physics-informed prognostics and remaining useful life (RUL) estimation
Qianling Wang
PhD Candidate, CSC Grant
Physics-informed inversion of drivetrains; quantification of retrofit gains
Cristian Velandia
PhD Candidate
SCADA-vibration fusion for gearbox diagnostics; static yaw misalignment
Ricardo M. Prieto
PhD Candidate, UPC / ESPOL
Autoencoders and adaptive conformal inference for fault detection
Carlos López Tamayo
PhD Candidate, ESPOL (co-supervised)
Reinforcement learning for wind turbine maintenance
More than 80 JCR-indexed journal articles; h-index 31 and more than 2,900 citations (Google Scholar, Y. Vidal). Representative journal papers, 2022-2026.
JournalWang S., Vidal Y., Zhang J. & Pozo F.: "Robust unsupervised gearbox fault detection under variable-speed operating conditions." Energy and AI, 25, 100862. Q1 Read →
JournalBarrera K. & Vidal Y.: "Explainable early fault detection in wind turbine generators using only SCADA data from an operational wind farm." IET Renewable Power Generation, 20(1), e70316. Read →
JournalEspinoza V., Ormaza C., Tutivén C. & Vidal Y.: "Reinforcement learning-based pitch control for wind turbines using double deep Q-networks." ISA Transactions, 169, 428-435. D1 Read →
JournalKestel K., … & Vidal Y.: "Condition monitoring of wind turbine drivetrains: state-of-the-art technologies, recent trends, and future outlook." Wind Energy Science, 11(6), 2103-2155. Read →
JournalVelandia-Cárdenas C., Vidal Y. & Pozo F.: "Enhancing wind turbine diagnostics with SCADA-vibration fusion, contrastive learning, and linear predictive coefficients." Structural Control and Health Monitoring, 4023580. Q1 Read →
JournalWang S., Vidal Y. & Pozo F.: "Fault detection in wind turbines using health index monitoring with variational autoencoders." Wind Energy Science, 10(12), 2841-2863. First prize, WeDoWind ASCE-EMI Structural Health Monitoring for Wind Energy Challenge Read →
JournalWang S., Vidal Y. & Pozo F.: "Recent advances in wind turbine condition monitoring using SCADA data: a state-of-the-art review." Reliability Engineering & System Safety, 267, 111838. D1 Read →
JournalCuesta J., Leturiondo U., Vidal Y. & Pozo F.: "A review of prognostics and health management techniques in wind energy." Reliability Engineering & System Safety, 260, 111004. D1 Read →
JournalEncalada-Dávila Á., Vidal Y. & Tutivén C.: "Strain virtual sensor for offshore wind turbine jacket supports: a time series transformer approach validated with Alpha Ventus wind farm data." Mechanical Systems and Signal Processing, 231, 112653. D1 Read →
JournalValdez R., Palacios E., Tutivén C. & Vidal Y.: "Bolt-loosening detection in offshore wind turbines' jacket-type supports." Structural Health Monitoring. Read →
JournalCampoverde-Vilela L., Feijóo M.C., Vidal Y., Sampietro J.L. & Tutivén C.: "Anomaly-based fault detection in wind turbine main bearings." Wind Energy Science, 8(4), 557-574. Read →
JournalEncalada-Dávila Á., Moyón L., Tutivén C., Puruncajas B. & Vidal Y.: "Early fault detection in the main bearing of wind turbines based on gated recurrent unit (GRU) neural networks and SCADA data." IEEE/ASME Transactions on Mechatronics, 27(6), 5583-5593. D1 Read →
Books, datasets, patent and theses
Books and chapters
2026 Vidal Y. Artificial Intelligence for wind turbine predictive maintenance (book chapter). Springer. DOI →
Open datasets (CORA)
2025 Replication data: detection of wind turbine rotor imbalance using unsupervised output-only vibration data analysis. DOI 10.34810/data2304
2024 Replication data: structural health monitoring of jacket-type support structures in offshore wind turbines, bolt loosening detection dataset. DOI 10.34810/data1011
Patent
Acho L., Vidal Y., Tutivén C., Rodellar J. & Pozo F.: method and analog electronic circuit for fault detection in the pitch hydraulic system of a wind turbine.
P201730664, Spain. Priority date 6 May 2017. Owner: UPC.
Doctoral theses
2023 Bryan Puruncajas: Detection and diagnosis of faults and damage in wind turbines.
2021 Diego A. Sandoval: Diagnosis of low-speed bearings via vibration-based entropy indicators and acoustic emissions.
2018 Christian Tutivén: Fault detection and fault tolerant control in wind turbines.
Working with industry and the community
Services for industry
Early fault detection and diagnosis from existing SCADA data, with no new sensors
Analysis of CMS and vibration data for drivetrain components (main bearing, gearbox, generator)
Structural health monitoring and virtual sensing for offshore support structures
Prognosis and remaining useful life estimation for maintenance planning
Joint R&D projects, industrial PhDs and technology transfer agreements
Community and editorial roles
Associate Editor of Wind Energy, Wind Energy Science and IET Renewable Power Generation
Theme Lead on the WESC 2027 scientific committee
Chair of the Anomaly Detection session at TORQUE 2026
Opportunities
We welcome PhD candidates, postdocs and companies interested in wind turbine condition monitoring.
Contact: Yolanda Vidal, yolanda.vidal@upc.edu
Industrial and academic partners
Selected awards