Deployment of large scale wind turbine parks, in particular offshore, requires well organized operation and maintenance strategies to make it as competitive as the classical electric power stations. It is important to ensure systems are safe, profitable, and cost-effective. In this regards, the ability to detect, isolate, estimate, and prognose faults plays an important role. One of the critical wind turbine components is the gearbox. Failures in the gearbox are costly both due to the cost of the gearbox itself and also due to high repair downtime. In order to detect faults as fast as possible to prevent them to develop into failure, statistical change detection is used in this paper. The Cumulative Sum Method (CUSUM) is employed to detect possible defects in the downwind main bearing. A high fidelity gearbox model on a 5-MW spar-type wind turbine is used to generate data for fault-free and faulty conditions of the bearing at the rated wind speed and the associated wave condition. Acceleration measurements are utilized to find residuals used to indirectly detect damages in the bearing. Residuals are found to be nonGaussian, following a t-distribution with multivariable characteristic parameters. The results in this paper show how the diagnostic scheme can detect change with desired false alarm and detection probabilities.
Effective fault identification and management are critical for efficient wind turbine operation. This research presents a novel adaptive- system designed to enhance fault tolerance in wind turbines, specifically addressing common issues such as pump wear, hydraulic leakage, and excessive air content in the oil. By combining with adaptive control, the proposed technique effectively controls rotor speed and power, ensuring reliable performance under various conditions. The controller employs an adjustable gain and an integrated sliding surface to maintain robustness. We validate the controller’s performance in the FAST (Fatigue, Aerodynamics, Structures, and Turbulence) simulation environment using a 5-megawatt wind turbine under high wind speeds. Simulation results demonstrate that the proposed adaptive- outperforms traditional adaptive- and adaptive control schemes, particularly in the presence of faults, unknown disturbances, and turbulent wind fields. This research highlights the controller’s potential to significantly improve the reliability and efficiency of wind turbine operations.
The probability density of crest amplitude and duration that exceeds a given level is used in many theoretical and practical problems in engineering that are subjected to fluctuating loads such as wind and wave loads. The presently available joint distributions of amplitude and period are limited to excursion through a mean-level or to describe the asymptotic behavior of high level excursions. This paper extends the knowledge by presenting a theoretical derivation of probability of wave exceedance amplitude and duration for a stationary narrow-band Gaussian process. A density function is suggested that has the salient feature to depend only on the three lowest spectral moments m 0 , m 1 , and m 2 and desired level of exceedance, H. It does not require any condition on the autocorrelation function. This paper shows how increase in H, increases the correlation between excursion periods and amplitude. This paper also shows that how the accuracy of the proposed joint distribution relates to spectral width parameter, v, and that accuracy increases for higher levels of H, especially for a spectrum describing a physical phenomenon such as a sea state spectrum. It was demonstrated that the marginal distribution of amplitude is Rayleigh distributed, as expected, and that the marginal distribution of excursion duration works for asymptotic and non-asymptotic levels. Results demonstrate that the established distribution fits well with ideal narrow-band Gaussian processes as well as the sea states at three European sites - in the Atlantic Ocean and the North Sea. The suggested model is found to be a good replacement for the existing empirical distributions.
Accurate characterization of anisotropic turbulence in wind-turbine wakes requires spectral models that retain both directional information and its variation across scales. Although one-dimensional spectra are widely used to compare measured and simulated wake turbulence, they collapse the underlying wavevector structure and therefore cannot fully describe the geometry of anisotropic energy organization. We propose a two-dimensional spectral framework based on a bidirectional, scale-dependent tensorial remapping of the planar wavenumber domain. The model represents the strong anisotropy observed at energy-containing scales and its gradual relaxation toward smaller-scale isotropic behavior. This transition is described by two directional stretch factors that depend jointly on the streamwise and spanwise wavenumbers. The corresponding Jacobian correction ensures that the spectral density is transformed consistently under wavevector mapping. The scale dependence of the directional stretching is represented by using constrained Bernstein polynomial surfaces. A cumulative softplus parameterization guaranties boundedness, smoothness, and monotonic decay throughout the optimization. The model is applied independently to large-eddy simulation datasets representing wind-turbine wakes under high- and low-turbulence inflow conditions, together with a reference atmospheric boundary-layer case without turbine forcing. Parameters are estimated using a staged, coarse-to-fine constrained optimization of the planar spectra of the streamwise and wall-normal velocity components. Component-energy constraints, Jacobian admissibility, and surface-smoothness regularization are included to maintain physical consistency and numerical robustness. The fitted spectra are also used to generate statistically equivalent synthetic turbulent fields to evaluate spatial correlations, integral scales, structure functions, and one-point statistics. The resulting framework provides an interpretable and computationally practical method for quantifying how turbine wakes and inflow turbulence intensity modify the directional, scale-dependent organization of turbulent energy.