Timely detection of surface damages on wind turbine blades is imperative for minimizing downtime and avoiding possible catastrophic structural failures. With recent advances in drone technology, a large number of high-resolution images of wind turbines are routinely acquired and subsequently analyzed by experts to identify imminent damages. Automated analysis of these inspection images with the help of machine learning algorithms can reduce the inspection cost. In this work, we develop a deep learning-based automated damage suggestion system for subsequent analysis of drone inspection images. Experimental results demonstrate that the proposed approach can achieve almost human-level precision in terms of suggested damage location and types on wind turbine blades. We further demonstrate that for relatively small training sets, advanced data augmentation during deep learning training can better generalize the trained model, providing a significant gain in precision.

A single crystal of GdBaCo 2 O 5.47 (2) has been studied by means of X-ray diffraction. Appearance of superstructure reflections at T = 341.5(7) K gives an evidence of continuous transition to the phase with unit cell doubled along the shortest edge a 1 . Critical exponent for the order parameter is found to be 0.33(1) = β .The superstructure reflections are about 2-4 orders of magnitude weaker than the basic ones. Their systematic extinction indicates the crystal symmetry change from to . The integrated intensities allow to calculate displacements of atoms from the positions in the high-temperature phase. The cobalt-ligand distances in the ordered phase are discussed in terms of the spin-state/orbital ordering of Co Pmmm Pmma 3+ ions. PACS numbers: 61.10.+Nz; 61.50.Ks; 61.66.Fn Since discovery of giant magnetoresistance in the oxygen-deficient layered perovskites RBaCo 2 O 5+x , where R is a rare earth [1], these materials attract high interest. Their orthorhombic structure at x ≈ 0.5 with the unit cell a 1 ≈ a p , a 2 ≈ 2a p , a 3 ≈ 2a p , where a p is parameter of the pseudocubic perovskite cell, is usually described by the Pmmm space group [2-5]. One can imagine the structure as a sequence of stacking plains [CoO 2 ][BaO][CoO 2 ][RO x ] along [0,0,1], which results in two types of the cobalt environment: CoO 5 pyramids and CoO 6 . octahedra. The nominal valence of cobalt at x = 0.5 is 3+. It is known [6] that the Co 3+ ion has a non-magnetic, or low-spin ground state (LS, ) as well as two excited states, the intermediate-spin (IS, ) and the high-spin (HS, ) ones. The energy differences are small enough to gain the excited states by the thermal fluctuations or due to the lattice change, which results in spin state transitions [7]. A metal-insulator (MI) transition has been also found, with the transition temperature for GdBaCo 0 6 2 g g e t 1 5 2 g g e t 2 4 2 g g e t 2 O 5+x 350 K < T MI < 370 K being dependent on the oxygen content of 0.4 < x < 0.47 [1, 2 8-10]. The transition is of the first order, which is indicated by a hysteresis of 8 K in resistivity for TbBaCo 2 O 5.4 [8].In spite of numerous studies of these materials, neither Co 3+ spin state nor nature of the MI transition is finally established. A spin-state transition coupled with the orbital degrees of freedom is suggested to be a driving force for the MI transition. The distribution of the IS e g orbitals (3x 2 -r 2 ) in pyramids and (3y 2 -r 2 ) in octahedral sites on cooling has been suggested as an origin of transition on the basis of structural studies of TbBaCo 2 O 5.5 [11]. On the other hand, it has been concluded that the transition to metallic phase in GdBaCo 2 O 5.5 is due to excitation of the LS-state electrons into e g band of the Co HS-state in octahedra; with Co in pyramids having IS both sides of T MI [9]. This conclusion has been made because of octahedron expansion of about 0.012(4) Å and simultaneous pyramid shrinking. The spin-state as well as the orbital ordering among one type of coordinating polyhedra was considered in a num...

Abstract. In the present study an evaluation of the potential for bend-twist coupling effects in wind turbine blades is addressed. A method for evaluation of the coupling magnitude based on the results of finite element modeling and full-field displacement measurements obtained by experiments is developed and tested on small-scale coupled composite beams. In the proposed method the coupling coefficient for a generic beam is introduced based on the Euler-Bernoulli beam formulation. By applying the developed method for analysis of a commercial wind turbine blade structure it is demonstrated that a bend-twist coupling magnitude of up to 0.2 is feasible to achieve in the baseline blade structure made of glass-fiber reinforced plastics. Further, by substituting the glass-fibers with carbon-fibers the coupling effect can be increased to 0.4. Additionally, the effect of introduction of bend-twist coupling into a blade on such important blade structural properties as bending and torsional stiffness is demonstrated.

Impact fatigue caused by rain droplets, also called rain erosion, is a severe problem for wind turbine blades and aircraft. In this work, an assessment of impact fatigue on a glass fibre reinforced polymer laminate with a gelcoat is presented and the damage mechanisms are investigated. A single point impact fatigue tester is developed to generate impact fatigue damage and SN data. Rubber balls are repeatedly impacted on a single location of the coated laminate. Each impact induces transient stresses in the coated laminate. After repeated impacts, these stresses generate cracks, leading to the removal of the coating and damage to the laminate. High-resolution digital imaging is used to determine the incubation time until the onset of coating damage, and generate an SN curve. An acoustic emission sensor placed at the back of the laminate monitors changes in acoustic response as damage develops in the coated laminate. The subsurface cracks are studied and mapped by 3D X-ray computed tomography. A finite element method model of the impact shows the impact stresses in the coating and the laminate. The stresses seen in the model are compared to cracks found by 3D tomography. The damage is also evaluated by ultrasonic scanning.

No abstract