A case study comparing GMM and AE for lightning damage detection in wind turbines using field SCADA data

Condition monitoring is essential for ensuring the reliable operation and effective maintenance of modern complex systems, where nonlinear behaviors, environmental variability, and sensor noise can complicate fault diagnosis. Wind turbines, as representative examples of such systems, have widely adopted power curve-based methods for performance assessment and anomaly detection. However, most existing approaches overlook the heteroscedastic nature of noise, which adds uncertainty to model training and condition monitoring. In this paper, we propose a Bi-level Piecewise Linear Model (Bi-PLM) with physics-informed constraints to improve resilience against potential data contamination. A data-driven procedure, combining a binning method with locally estimated scatterplot smoothing (LOESS), is developed to characterize the noise heteroscedastic structure, which is then incorporated into change point detection (CPD)-based condition monitoring. Experiments on two real-world datasets show that explicitly accounting for heteroscedasticity reduces variance-induced uncertainty in residuals and substantially lowers false positives in fault detection, yielding average increases in the area under the curve (AUC) of approximately 6.0% and 8.4% on the two datasets. Comparative results against benchmark models confirm the robustness and reliability of the proposed method.

A deep mutual learning-based framework for wind turbine blade defect detection in multimodal phased array ultrasonic data.

High AC loss in stator windings is one of the key challenges in realizing fully high-temperature superconducting (HTS) wind turbine generators, which offer significant advantages in compactness and weight compared to conventional machines. Among the sources of AC loss, the perpendicular magnetic field component (relative to the HTS tape surface) is particularly dominant, while the parallel field contributes much less. Therefore, minimizing the perpendicular field in the stator windings is essential when designing a fully HTS generator. In this work, we present a 5 MW-class, air-cored, axial-flux fully HTS wind turbine generator modeled using the finite element method (FEM). The T-A formulation and interpolated J<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">c</sub>(B) characteristics are used to simulate the electromagnetic behavior of the HTS windings. A moving mesh approach is employed to simulate the rotor movement. Simulation results show that the outer-stage stator windings experience significantly higher AC loss than the inner-stage windings, despite delivering the same power output. By optimizing the rotor end-winding geometry, the perpendicular magnetic field in the outer-stage windings was significantly reduced, resulting in a 77% decrease in total AC loss across the HTS stator windings.

Preparation of a polyurethane/Ni foam composite material for the leading edge of wind turbine blades and investigation of its rain erosion behavior

Liquefaction response of monopile offshore wind turbines under different loading conditions during operating and shutdown states

Time-varying cause tracing-supplemented risk prediction framework for wind turbines under complex uncertain conditions

Vibration control for TLP wind turbines using a tuned mass inerter system: Design, optimization and performance evaluation

• BS7910 FCGR for R < 0.5 is up to 3.6 × over-conservative for S355 in free corrosion. • Removing this conservatism could improve remaining life by 109%. • For the investigated database, statistical weighting had a negligible impact. • HAZ data fit a bilinear curve more closely, while BM data fit a linear one. • S355J2 + N subgrade displayed the highest FCGR, while S355G10 + M showed the lowest. The recommendations for fatigue crack growth rate (FCGR) in British Standard BS 7910 are based on experimental data which are multiple decades old, and the FCGR laws were constructed using data from multiple structural steel grades. The present study sought to gather more recent FCGR data for S355 steel in a marine free-corrosion environment, which can be applied for use in offshore wind turbine structural integrity life prediction. The collected data were filtered to include only valid Stage II (Paris region) data in the analyses. A method for fitting a bilinear regression curve was developed along with bias reducing weighted regression methods. The resulting bilinear trend shows up to 3.6 times lower FCGR relative to the BS 7910 recommendations for the free-corrosion environment. The analysed S355 data indicated that a bilinear fit best described the FCGR in the heat affected zone, whereas a simple linear fit was appropriate for the base metal. The S355 subgrades, J2 + N, G8 + M and G10 + M, presented considerable variance in mean FCGR, with J2 + N having the highest FCGR and G10 + M the lowest. When considered in the context of offshore wind turbine service life, the overall S355 FCGR reduction relative to the BS 7910 free corrosion curve results in a just over double life expectancy for a 4 × 20 mm surface semi-elliptical flaw size, based on the conservative upper bound analysis. This increase in life could prove invaluable for corrosion protection system repair timescales, considering the need to carefully plan the maintenance time window to avoid adverse weather conditions.

Fabrication of photothermal superhydrophobic composite coatings based on MWCNTs/SiO₂ and anti-icing/deicing performance of wind turbine blades

Characterization, experimental validation and life cycle assessment of wind energy industrial waste in asphalt mixtures.

ABSTRACT Anthropogenic soundscape disturbance can constrain acoustic signalling, with potential consequences for reproductive behaviour in wild animals. We used passive acoustic monitoring to investigate how habitat, weather, and human-associated soundscape disturbance are associated with roaring activity of Iberian red deer (Cervus elaphus hispanicus) during the rut on Lousã Mountain, central Portugal. Twenty-nine AudioMoth recorders collected 150-s recordings every 7.5 min over 14 days, yielding 88,022 manually validated roars. We quantified anthrophony as relative acoustic energy in the 1–2 kHz band and modelled roar counts using a negative binomial generalised linear mixed model. Roaring activity was highest in shrublands and other open habitats, and decreased with increasing wind speed, rainfall and temperature. Importantly, roaring declined with increasing anthrophony and increased with distance from wind turbines, suggesting a possible avoidance of a broader disturbance gradient rather than only a purely acoustic effect. Daily roaring counts also tended to be lower towards the end of the week. Our results highlight that soundscape disturbance associated with human infrastructure and activity may influence both the location and timing of males’ vocalisations during the rut, demonstrating the value of soundscape metrics for identifying rutting areas potentially sensitive to disturbance and informing appropriate mitigation strategies.

A reduced order analytical model to reconstruct tower bending moments time series of onshore wind turbines validated with experimental data

Wind power plants are frequently placed in natural ecosystems, but their impacts on plant communities are rarely considered. Therefore, it is unknown how far potential impacts extend into adjacent vegetation and how long they persist. To address this, we surveyed vegetation at different distances to roads at three wind power plants in Norway that were commissioned 4, 12, and 19 years ago. We then used Grime's CSR strategies to document functional shifts in plant community composition and Ellenberg Indicator Values (EIVs) to identify the abiotic gradients driving these shifts. We found that shifts in plant community composition were related to road distance and time since disturbance. At the youngest site, the proportion of plants with ruderal strategies was significantly increased within 10.4 m of roads, effectively expanding the footprint of roads by more than two-fold. At the oldest site, this impact was reduced to 2.8 m, suggesting that the original stress-tolerant communities recovered at a rate of 0.5 m per year. Increased ruderality near roads was associated with plant communities indicating higher nutrient availability and more reactive soils. This study provides novel knowledge regarding the spatial and temporal impact of wind energy development on plant communities. As road construction appears to shift community composition toward ruderal dominance by increasing nutrient availability, we recommend keeping road- and construction areas to a minimum. Overall, this can reduce the footprint of wind power plants and ensure that the transition to renewable energy does not come at the expense of ecosystems.

The progress of industrialization and rising electricity demand have significantly impacted various aspects of our lives, leading to a prioritization of green electricity sources such as wind turbines. However, managing the waste from end-of-life turbine blades poses a significant challenge, particularly as the first generation of wind turbines in Europe, primarily made of glass fibre-reinforced thermosets, reaches the end of service life. Unfortunately, landfilling and incineration remain common disposal methods in many EU countries. Co-processing end-of-life blades in cement manufacturing offers a promising alternative by enabling energy and material recovery. This method aligns with sustainable development and circular economy goals. Cement production, known for its high CO 2 emissions and extensive raw material consumption, can benefit from substituting ground or shredded turbine blades, which contain organic and inorganic components. This study conducts a comparative life-cycle assessment of incineration and co-processing across different European countries to identify the disposal method with the least environmental impact. The evaluation focuses on carbon footprint reduction and water conservation as key benefits. Findings reveal that while incineration poses risks to global warming, it can help preserve water. In contrast, co-processing yields environmental benefits across all assessed categories by conserving raw materials for cement production. The normalized results indicate that human health is the most impacted area of protection for both processes, with resource benefits influenced by the type of energy produced in the examined countries. The Global Warming Potential values for co-processing are similar across the four countries (−525 kg CO2-eq for Germany, −523 kg CO2-eq for Spain, −533 kg CO2-eq for Italy, −494 kg CO2-eq for the Netherlands), representing an environmental benefit from substituting the raw mineral feedstock with EoL turbine blades. Moreover, the results show that, in terms of Human Health and Ecosystems, co-processing produces environmental benefits, whereas incineration generates damage (positive impact values) due to emissions that are harmful to living species regardless of geographical location. • Comparative life-cycle quantitative assessment between different waste management possibilities for EoL wind turbine blade. • The assessment is carried out in different European countries. • Description of the co-processing recycling procedure of EoL turbine blades is presented. • The environmental impacts have been presented in terms of GWP, AWERE, human health, natural resource, and ecosystem.

The rapid expansion of global wind power infrastructure has urgently necessitated accurate identification of wind turbines (WTs). In this study, we propose a deep learning detection framework constrained by multi-source geospatial data for automated global WT identification using remote sensing imagery. By integrating land-use, topography, anthropogenic buffers, and wind power density constraints, we construct a global WT installation suitability map that serves as a spatial constraint, which effectively reduces the search space by 38.99%. An iterative sample refinement strategy is then employed, where dataset expansion and model training are jointly performed through a feedback loop. The final YOLOv8 model achieves precision values of 96.3% and 90.1% on the training and validation sets, respectively. Within the global latitude range of ±70°, 404,392 WTs are detected, including 48,688 previously unrecorded in OpenStreetMap. The detection results are consistent with known wind resource distributions, predominantly concentrated in China, the United States, and India. Detection performance varies across terrain types, with the highest accuracy observed in inland plains (95.61%). The proposed framework provides a robust and scalable approach for global wind power infrastructure mapping and monitoring, thereby offering valuable insights for energy planning.

After one century from the original patent of J.M. Darrieus on a lift-driven “turbine having its rotating shaft transverse to the flow of the current”, vertical-axis wind turbines (VAWTs) are stuck in a dichotomy between being the most fertile field of research for aerodynamicists and a technology that still fails in turning into an industrial reality, with many companies going in and out of the market. After the research on VAWTs restarted in the early 2000s, significant progress was made in understanding their complex aerodynamics and modeling them with simulations of different fidelity, leading in turn to the definition of design guidelines able to reduce the performance gap with respect to horizontal-axis wind turbines (HAWTs) to less than ten percentage points. This did not persuade the consolidated HAWT industry to leave the most efficient and reliable horizontal-axis archetype, especially after the rush towards diffused onshore energy production by VAWT started in the 2010s failed. However, recent discoveries are putting again Darrieus VAWTs in the spotlight as a possible game changer in offshore wind energy, where their faster wake recovery could bring to farms with unprecedented energy density in a market where competition for marine space is strong. Different to other reviews made to date, the present study not only presents the evolution of VAWTs over the last decades, but also critically analyses the lessons learnt made along the way and highlights the recent discoveries that are opening after a century new prospects for VAWT technology. • Critical analysis of the evolution phases of Darrieus VAWTs. • Annotated overview of recent discoveries and new design paradigms. • A new perspective on unsteady aerodynamics in VAWTs. • Analysis of the prospects of flow entrainment for offshore applications.

Hybrid graph neural network with FFT and genetic optimization for fault detection in wind turbines

Recent advances in carbon based anti-icing materials: new insights from Hansen Solubility Parameter analysis.

The transition to sustainable energy positions wind power as a key renewable solution. As demand grows, wind turbines are deployed across diverse terrains. However, wind’s stochastic nature and environmental variability complicate power forecasting, affecting grid stability. The study leverages data-driven techniques to enhance wind power forecasting using high-resolution SCADA system time-series data. Key operational parameters include wind speed, rotor speed, generator speed, nacelle orientation, ambient temperature and power output. A comparative analysis evaluates traditional machine learning models—Linear Regression, Decision Trees, Random Forests, Gradient Boosting and Support Vector Machines—against deep learning models like Long Short-Term Memory (LSTM) networks and a novel Recurrent Neural Network (RNN) architecture. The core contribution is an optimized Bidirectional LSTM-RNN model with permutation layers and attention. These layers capture long-range dependencies and nonlinear interactions in wind data. The structure improves long-range dependency capture and nonlinear interaction modeling. Bidirectionality enables learning from both past and future time steps, while attention mechanisms highlight critical temporal features. Experimental results demonstrate the proposed model’s superior performance, achieving a Mean Absolute Error (MAE) of 0.0994 and Root Mean Square Error (RMSE) of 0.1390, significantly outperforming traditional models (e.g., Random Forest: MAE 86.44, RMSE 220.30) and basic LSTM models (MAE 14.48, RMSE 15.27). Robust cross-validation confirms its ability to generalize across different temporal segments. Feature importance analysis improves interpretability, supporting informed decision-making in wind farm operations. The framework is scalable, modular and well-suited for real-time forecasting applications. The work presents a reliable deep learning model for wind power forecasting, enabling intelligent, data-driven energy management in modern power systems.