Field of disclosure

The disclosure relates to a system comprising a wind turbine blade mould being kept substantially horizontal, and a plurality of digital cameras arranged with at least partly overlapping fields of view above the mould for providing stereo imaging thereof to enable 3D reconstruction.

Background

Fiber-reinforced composite materials are still prevalent in design and manufacturing of fiber composite structures for wind turbine generators, in particular wind turbine blades. The production of such fiber composite structures normally comprises a layer deposition process, in which a plurality of lengths of fabric are placed on a mould in a particular stacking sequence. A process where each lengths of fabric is placed at individually predetermined placement position on the layup surface of said mould and subsequently infused with a curable liquid resin in order to form a composite material according to design specifications defined for achieving a certain performance goal by prior modelling, structural testing or similar expertise.

Design specifications may be based on two- and three-dimensional models of physical objects and their physical interactions, such as structural load analysis encompassing also approximations of the inhomogeneous and anisotropic material properties arising from fiber filaments within the fabric, the cured resin and other components within the composite structure. In particular, it may be necessary to combine such models with empirical data by also testing aspects of a physical counterpart to yield sufficiently accurate predictions of the mechanical properties of a fiber composite structure, such as a wind turbine blade with respect to strength, deformations due to structural loads and material fatigue due to load variations.

Due to the complexity of the structure of wind turbine blades formed in composite structures, the accuracy of the model prediction of the mechanical properties of the manufactured structure may however still require substantial testing of specimens of the composite structure as part of the design and approval process, e.g. when a new or amended design specification is in development. Such testing is costly as well as time consuming, so improved prediction tools may be needed to cut down on testing. This type of performance prediction as a tool to improve design specifications is however fraught with difficulties and impediments in terms of achieving a sufficient accuracy, and developing a unified model can also be very time consuming. A typical engineering practice is then to accept a reduced, marginally acceptable accuracy of each model prediction and to include in the design specification a wider standard- defined safety margin, e.g. with respect to the maximal structural load the fiber composite structure must be able to withstand without failure.

Objective of the disclosure

One objective of the present invention is to provide tools and methods of improving the modelling accuracy of composite structures for wind turbines, in particular for wind turbine blades, to reduce time and costs of developing design specifications for such composite structures. Furthermore, improving the modelling accuracy may enable reduction of the necessary safety margins included in the design specification, which may lead to reduction in material consumption and mass of the composite structures.

Description of the disclosure

Disclosed herein is a system comprising a wind turbine blade mould extending substantially horizontally in a longitudinal direction of the mould, and a plurality of digital cameras arranged with at least partly overlapping fields of view above the mould.

The mould may have an upwardly facing surface for supporting fabric materials placed thereon. It is understood that, as fabric materials are placed on the mould, the topmost fabric materials form part of the upwardly facing surface for supporting fabric materials.

The plurality of digital cameras may be arranged in a pattern, so that a set of digital images depicting at least each point of the upwardly facing surface of the mould in a specific mould section is capturable by at least two of the digital cameras. The specific mould section may comprise an area of the upwardly facing surface of the mould of at least 25 m ² and/or at least 5% of the area of the upwardly facing surface of the mould.

The plurality of digital cameras may be three or more digital cameras. It is understood that the plurality of digital cameras may be arranged in a pattern wherein not all digital cameras in said plurality of digital cameras are substantially in line with each other. An elevation angle between a line of sight of each of the at least two digital cameras and a tangent plane at each of said points may be between 25° and 90°.

The plurality of digital cameras may be arranged to capture at least partly overlapping digital images, wherein at least 15% of the upwardly facing surface of the mould capturable by one of said plurality of digital cameras is also capturable by another camera of said plurality of digital cameras, which other camera furthermore is arranged to capture at least 15% of the upwardly facing surface of the mould capturable by yet another camera of said plurality of cameras.

It is understood that one of said plurality of digital camera may capture a digital image depicting a first part of the upwardly facing surface of the mould and another camera of said plurality of digital cameras may capture a digital image depicting a second part of the upwardly facing surface of the mould, the first part and second part overlapping. Yet another camera of said plurality of cameras may capture a digital image depicting a third part of the upwardly facing surface of the mould, the second part furthermore overlapping the third part. Hereby, three partly overlapping images may be provided. It is understood that additional partly overlapping images may be included e.g. by having the plurality of digital cameras depict a fourth, fifth and sixth part of the upwardly facing surface of the mould, said parts overlapping at least one of a first, second or third part, whereby said digital images may be combined into a panoramic image depicting e.g. a substantial portion of the first, second, third, fourth, fifth and sixth part by said overlap. The plurality of digital cameras according to embodiments of the present disclosure may be arranged in a pattern, so that many partly overlapping digital images can be captured with a reasonable amount of overlap between the individual digital cameras, whereby empirical data may be deduced for a specific mould section based thereon.

By at least 15% of a first part of the upwardly facing surface of the mould capturable by one of said plurality of digital cameras also being capturable by another camera of said plurality of digital cameras, at least 15% of said first part is depicted by two partly overlapping images. Likewise, by at least 15% of a second part of the upwardly facing surface of the mould also being capturable by yet another camera of said plurality of cameras, at least 15% of said second part is depicted by two other partly overlapping images. This may allow for multiple partly overlapping images to be reliably combined. Having a plurality of digital cameras arranged with at least partly overlapping fields of view above the mould may provide a panoramic image of the specific mould section. In some embodiments, at least 5% of the upwardly facing surface of the mould capturable by one of said plurality of digital cameras is also capturable by another camera of said plurality of digital cameras, which other camera furthermore is arranged to capture at least 5% of the upwardly facing surface of the mould capturable by yet another camera of said plurality of cameras.

In some embodiments, at least 10% of the upwardly facing surface of the mould capturable by one of said plurality of digital cameras is also capturable by another camera of said plurality of digital cameras, which other camera furthermore is arranged to capture at least 10% of the upwardly facing surface of the mould capturable by yet another camera of said plurality of cameras.

In some embodiments, at least 25% of the upwardly facing surface of the mould capturable by one of said plurality of digital cameras is also capturable by another camera of said plurality of digital cameras, which other camera furthermore is arranged to capture at least 25% of the upwardly facing surface of the mould capturable by yet another camera of said plurality of cameras.

In other embodiments, between 5% and 50% of the upwardly facing surface of the mould capturable by one of said plurality of digital cameras is also capturable by another camera of said plurality of digital cameras, which other camera furthermore is arranged to capture between 5% and 50% of the upwardly facing surface of the mould capturable by yet another camera of said plurality of cameras.

The least two digital cameras may be arranged to capture at least partly overlapping digital images, wherein at least 15% of the upwardly facing surface of the mould captured by one of said at least two digital cameras may also be capturable by another of said at least two digital cameras. By the at least partly overlapping digital images having at least 15% of the upwardly facing surface of the mould in common within their respective fields of view, the partly overlapping digital images may be combined to produce a larger panoramic image, e.g. via image stitching, by aligning said digital images. Aligning two digital images may require identifying a first part of the upwardly facing surface of the mould captured in a first digital image. The part of the upwardly facing surface of the mould captured in a digital image may be identified as a region of pixels depicting said part of the upwardly facing surface via image segmentation. By also identifying a second part of the upwardly facing surface of the mould captured in a second digital image, wherein the first part and second part have at least 15% of their respectively captured visual information in common, said two digital images may be aligned by orienting and rectifying the two digital images so that their overlap match. It is understood that having visual reference points may improve the alignment accuracy, e.g. by providing more visual information in common between said two digital images.

In some embodiments, at least 25%, such as at least 50%, of the upwardly facing surface of the mould captured by one of said at least two digital cameras may also be capturable by another of said at least two digital cameras. It is also contemplated that having at least 25% overlap, such as at least 50%, between the two digital images may allow for good alignment in cases where there are fewer visual reference points or where large image distortion is caused by curvature of the upwardly facing surface.

The plurality of digital cameras may be arranged to capture at least partly overlapping digital images, wherein a first digital image capturable by one of said plurality of digital cameras partially overlaps a second digital image capturable by another camera of said plurality of digital cameras by at least 50 pixels measured as a number of pixels in a direction of least overlap, which second digital image furthermore partially overlaps a third digital image capturable by yet another camera of said plurality of cameras by at least 50 pixels measured as a number of pixels in a direction of least overlap. Direction of least overlap may be determined between two partially overlapping images as that direction in which the extent of a region of pixels depicting the same visual information is smallest, i.e. the direction in each of said images which provides the least amount of pixels useful for correlating reference points between said two digital images for image stitching or the like. The number of pixels may be measured by counting pixels along the direction of least overlap in each of images, wherein the smallest number is used. By arranging the plurality of digital cameras in a pattern according to embodiments of the present disclosure, this allows for capture of multiple pairs of partially overlapping images, wherein each pair overlaps at least 50 pixels measured as a number of pixels in a direction of least overlap within said pair. The partially overlapping digital images are hereby overlapped in such a way that each point of the upwardly facing surface of the mould may be capturable by at least two of the plurality of digital cameras, which allows for combining said digital images by image stitching or the like, thus providing empirical data which includes depth information for at least a specific mould section.

In some embodiments, a first digital image capturable by one of said plurality of digital cameras partially may overlap a second digital image capturable by another camera of said plurality of digital cameras by at least 100 pixels, such as by at least 200 pixels, measured as a number of pixels in a direction of least overlap, which second digital image furthermore partially overlaps a third digital image capturable by yet another camera of said plurality of cameras by at least 100 pixels, such as by at least 200 pixels, measured as a number of pixels in a direction of least overlap.

The elevation angle between a line of sight and a tangent plane may be measurable as 90° minus an angle of incidence of a ray of light impinging a point on the upwardly facing surface of the mould from a light source placed at the relevant camera and pointed at said point. The angle of incidence may be relative to a surface normal.

In one aspect, the digital cameras may be arranged in a pattern being a regular grid.

The set of digital images may be a non-empty set comprising one or more images.

Using so-called stereo imaging where two cameras each capture digital images which depict the same points from non-coincident orientations i.e. a difference in elevation angle and/or rotation angle may result in 3D imaging of the surface. It is preferable that two such cameras configured to capture two or more digital images depicting the same point from two different viewpoints may be arranged to do so with a sufficiently large parallax between their respective lines of sight, such that depth information extracted via stereo imaging based on the two or more digital images provide a high degree of depth resolution. Generally, points at the upwardly facing surface or the layup surface may appear relatively indistinguishable from other points, e.g. when glass fiber fabrics are arranged to cover parts of the mould, due to substantially uniform appearances. It is understood that 3D imaging of the surface via stereo imaging using digital cameras may require identifying and correlating reference points in two or more digital images depicting the same point from two different viewpoints to derive said depth information. Such reference points may include a plurality of highly localized visual features at the upwardly facing surface, e.g. fiducial markers, structured light, contrasting fabrics etc.

A low elevation angle may result in focal or perspective distortion between multiple points to be depicted in a single image. Thus an elevation angle roughly perpendicular to the surface to be imaged may be preferred to reduce any distortion, however at the same time, the points to be 3D-imaged should be depicted from two non-coincident orientations for obtaining sufficient depth information. In one aspect, the elevation angle of the at least two cameras may be between 45° and 85° on average for at least each point of a specific mould section. This may provide a sufficiently low focal or perspective distortion so that loss of information is minimized when subsequently correcting for said distortion by post-processing the digital images.

Furthermore, the elevation angle, the rotation angle, the distance between the camera and the tangent plane or a combination hereof may be different for the at least two cameras, typically being substantially different for all of the plurality of digital cameras.

One effect of this embodiment may be to achieve 3D images of the wind turbine blade during the fabrication process i.e. , the structure of the wind turbine blade may be 3D imaged layer-by-layer such that a complete 3D-model of the actual structure of the blade may be obtained.

The actual structure of wind turbine blade may be at variance with the modelled structure. The process of building a wind turbine blade includes acts of arranging a plurality of sheets, mats and/or lengths of fabric comprising a plurality of substantially linear fiber filaments arranged within the fabric on a curved layup surface, distributing resin between the abutting layers of fabric and performing these tasks with a high level of ad hoc manual handling and adjustment during the process. The modelled structure includes an idealized description of how to build the structure. The description may include tolerances for the placement of each sheet, specifications for using dry sheets or sheets pre-impregnated with resin and amounts of resin to be used during vacuum assisted resin transfer moulding. Typically, the description does not specify layup steps.

The discrepancies of the modelled structure and the actual structure may include folds in the lengths of fabric, faults in the fiber filaments or distribution hereof within the fabric. Furthermore, one wind turbine blade may be build using the maximum tolerances, while another may be built exactly according to the description. Thus, an actual structure of a wind turbine blade, subject for testing, is known in details. The test results may as a consequently be linked directly to the actual structure of the wind turbine blade and not merely a theoretical model structure.

Furthermore, the actual 3D imaged structure of a wind turbine blade may be beneficial for monitoring and analyzing the wind turbine blade when in use. This may include monitored behavior of the wind turbine blade in use or observed failures arising during use. These observations may be compared and analyzed for the impact and degree of impact of defects or variances in the construction of the wind turbine blade. The actual 3D imaged structure of a wind turbine blade and subsequent monitoring of the blades performance may be used for improved modelling of the composite structure.

The embodiment is in no way limited to wind turbine blades but may be used for other composite structures. This may lead to a reduction in time and costs of developing design specifications for such composite structures. Furthermore, improving the modelling accuracy may enable reduction of the necessary safety margins included in the design specification, which may lead to reduction in material consumption and mass of the composite structures. The present disclosure relates to a method for systematically collecting data about the specific characteristics of a fibre composite wind turbine blade as it is being manufactured, i.e. while the length of fabric are being placed. This allows for a more thorough comparison to structural models, which can be systematically adjusted and optimized to better fit said data, whereby a reduction of safety margins is possible.

The actual 3D imaged structure of a composite structure, i.e. specific documentation for each structure, may be used for:

• Documentation of quality requirements.

• Improved blade production by subsequent analysis of discrepancies and their impact on and relation to parameters such as lifetime, wear and tear, specific damages and fractures.

• Verification of building models.

• Establishing quality requirements and production tolerances.

• Establishing critical deviations for specific sections or parts of the composite structure and optimizing the structure of the wing based on the findings of specific critical or non-critical parts and faults to determine permissible deviations and possibly reduced material consumption.

Reducing material consumption may be of great interest as a reduced consumption e.g. of glass fiber fabrics may result in lighter constructions of the finished wind turbine blade. This may lead to constructions having a greater extend without breaking due to weight of the structure itself, or conversely, to cheaper construction in support thereof. Measured data collections may form a statistical basis for changing the construction of the structure to reduce the difference between the modelled construction and the real construction of the structure. Better modeling supported by measurement may also be beneficial for establishing quality and safety requirements, and for fulfillment thereof. Data collections of the manufacturing of each blade may also be important to have on record e.g. if warranty claims is made to that specific blade in the future. Correlation between one failure instance and “as build” documentation may be directly observable in these data collections. Should such a correlation exist, data for other blades may be screened for similar features so that other affected blades can be tested or recalled.

In some embodiments, the plurality of digital cameras are arranged in a pattern at positions being substantially equidistant to the upwardly facing surface of the mould, such as by each digital camera having a smallest distance to said upwardly facing surface approximately equal to the smallest distance of another digital camera to said upwardly facing surface, such as being within ±50%, such as between 75% and 125%.

In one embodiments of the system, the plurality of digital cameras may be arranged in a pattern substantially encompassing, removed from and/or substantially peripheral to the upwardly facing surface of the mould. For example, each digital camera may be affixed at a position away from the boundary or periphery of the mould to provide a substantially unhindered view of at least a part of the upwardly facing surface of the mould, optionally also providing a peripheral view of other parts of the upwardly facing surface of the mould. A periphery of the mould may be a circular, semi-circular or arc like spatial region at the outer limits or edge of a mould having a substantially circular cross-section, such as e.g. at a root section. The digital cameras arranged in a pattern being substantially peripheral to the upwardly facing surface of the mould may provide that the digital cameras have a substantially unhindered view of at least a part of the upwardly facing surface of the mould via a space providing a mould distance between the cameras and the mould, so that an operator may work unhindered in said space.

In one embodiment of the system, the plurality of digital cameras may be arranged in a pattern by being placed at points along a curvilinear surface being substantially remote or peripheral from and equidistant to the upwardly facing surface of the mould, such as by the curvilinear surface being a substantially cylindrical surface arranged along the mould with the cylindrical surface being on a periphery of a root section of the mould. By peripheral arranged may also be understood that the cameras are arrange such that the upwardly facing surface of the mould to be captured is within line of sight of at least two cameras. Hereby, extraction of depth information from two or more digital images in the plurality of digital images may be possible via computer stereo vision.

Arranging the cameras according to one or more embodiments of the present disclosure may ensure that each point of the surface to be captured is captured with sufficient quality and height/depth information obtain a 3D image. The inventors have seen good indication that by capturing high-resolution digital images of glass fiber fabrics continuously or at least regularly during the layup process of multiple layers of glass fiber fabric onto the wind turbine blade mould, and additionally using said high- resolution digital images to also compose high-resolution 3D images thereof, such a camera arrangement may provide valuable data for improving the modelling accuracy of composite structures for wind turbines and in particular for wind turbine blades. It is furthermore an advantage of the system that capturing valuable data for improving the modelling accuracy of composite structures for wind turbines does not require manual measurement steps nor does it impinge on the workers carrying out the layup process.

It is understood that obtaining information of edges, wrinkles or other irregularities in fabric materials may be detectable from conventional imaging based on either intensity information, depth information or a combination thereof. This information may also be processed, e.g. via statistical analysis on a large database over multiple manufacturing instances, to reveal manufacturing issues related to the layup process as such.

In one embodiment of the system, the plurality of digital cameras may be arranged along a substantially horizontal plane. For example, the plurality of digital cameras may be arranged along a substantially horizontal support structure below a ceiling.

This arrangement of the cameras may ensure a multipurpose installation of the cameras usable for a variety of mould shapes. Furthermore, the curvature of the mould and the change of the curvature to be captured during the construction of the wind turbine blade may be such that a planar arrangement of cameras is sufficient to provide good fidelity of intensity and depth information based on two or more images.

Furthermore, mounting of the cameras in a substantially horizontal plane may be obtained in an easy and straightforward manner, e.g. in a construction hall, without performing extensive calculations of the camera arrangement and adjusting the camera positions accordingly.

In one embodiment of the system, the plurality of digital cameras are arranged in a two-dimensional array.

This arrangement of the cameras may likewise ensure a multipurpose installation of the cameras usable for a variety of mould shapes. Furthermore, each camera being placed on a node of a two-dimensional grid, such as a regular grid, in the array may provide that recording the spatial position and orientation of each camera is made simpler, thereby allowing faster installation or rearrangement of the entire pattern.

In a further embodiment, the two-dimensional array may comprise at least four rows of digital cameras, each row extending generally in a transversal direction to the mould and the rows being arranged with a mutual longitudinal distance in the longitudinal direction of the mould.

By capturing high-resolution digital images of glass fiber fabrics using a plurality of digital cameras arranged in a two-dimensional array comprising at least four rows, the system is enabled to capture a set of digital images of a substantive part of the mould along its longitudinal extent, such as a root section of the wind turbine blade mould. It is understood that having a plurality of rows being substantial parallel to each other and provided at a mutual distance to each other, such an arrangement of the plurality of digital cameras is enabled to capture digital images of a larger section of the mould.

In further embodiments of the system with the plurality of digital cameras arranged in a two-dimensional arrays comprising multiple rows of digital cameras, the digital cameras may be arrange according to the following further embodiments.

In a further embodiment, each row of digital cameras may comprise at least two digital cameras with a mutual distance in the transversal direction, such as a plurality of rows comprising three or more digital cameras.

By capturing high-resolution digital images of glass fiber fabrics using a plurality of digital cameras arranged in a two-dimensional array comprising four or more rows, with each row comprising at least two digital cameras with a mutual distance in the transversal direction, the system is enabled to capture a set of digital images of a substantive part of the mould both along its longitudinal extent and in the transverse direction, where the mould is generally more curved and thus harder to fully cover. It is understood that having at least two digital cameras in each row, with the rows being substantial parallel to each other and provided at a mutual distance to each other, such an arrangement of the plurality of digital cameras is enabled to capture digital images of a more curved section of the mould.

The inventors have seen good indications that by capturing high-resolution digital images of glass fiber fabrics using a plurality of cameras arranged in at least four rows with each row comprising at least two digital cameras provided at mutual distances to each other, such a camera arrangement may provide valuable data for improving the modelling accuracy of composite structures for wind turbines. In particular for wind turbine blades having curved sections with a curvature radius less than 3 meters. It is furthermore an advantage of the system that capturing valuable data for improving the modelling accuracy of composite structures for wind turbines can be carried out using a substantially fixed installation of the digital camera arrangement i.e. without requiring rearrangement of the digital cameras during the layup process e.g. due to the mould.

In a further embodiment, each digital camera may have a distance to neighboring digital cameras in the same row of between 2.5 meters and 15 meters or 100% of the smallest mould distance of said neighboring digital cameras to the mould, whichever is greater. It is understood that, for example, with a smallest distance of said neighboring digital cameras to the mould being 10 meters, this range is 2.5 meters to 15 meters; as it is written here, the “whichever is greater” clause applies to upper limits of the range.

In one aspect, each digital camera may have a distance to neighboring digital cameras in the same row between 2.5 meter and 10 meters. In another aspect, the distance to neighboring digital cameras in the same row may be between 5 meters and 10 meters or 75% of said smallest distance of said neighboring digital cameras to the mould, whichever is greater. It is understood that, for example, with a smallest distance of said neighboring digital cameras to the mould being 15 meters, this range is 5 meters to 11.25 meters. Preferably, the smallest distance is not less than 2.5 meters.

In one aspect, the smallest distance of said neighboring digital cameras to the mould may be 4 meters or greater. In a further embodiment, the at least two digital cameras in the same row may have a mutual distance in the transversal direction between 5 and 15 meters, such as between 7 and 12 meters, such as approximately equal to the smallest distance of said at least two digital cameras to the mould.

In a further embodiment, for each row, two neighboring digital cameras in said row may have a mutual distance in the transversal direction between 1 and 10 meters, such as between 3 and 8 meters, such as between 6 and 10 meters.

By two neighboring digital cameras in a respective row having a mutual distance in the transversal direction between 1 and 10 meters, the two digital cameras in the same row of digital cameras may provide that a set of digital images depicting at least each point of the upwardly facing surface of the mould in a part of the specific mould section may be capturable by said two digital cameras from different viewing angles. Further, by the two neighboring digital cameras having such a mutual distance in each row, two digital cameras from two different rows are also provided at a mutual distance to each other, thereby potentially also contributing to such points being suitably capturable.

The inventors have seen good indications that particularly with a substantial portion of the rows having two digital cameras, having two neighboring digital cameras in each row with a mutual distance in the transversal direction between 6 and 10 meters may provide a pattern which represents a cost-effective tradeoff between cost and fidelity.

In particular, for a mould distance of the plurality of digital cameras to the mould is between 8 and 12 meters with two digital cameras per row dispersed with a mutual distance in the transversal direction between 6 and 10 meters, a range of viewing angles from said cameras to the mould of about 30° to 75° is provided. The range of viewing angles to the mould being e.g. about 30° between a first line of sight to a first camera having a mould distance of 12 meters and a second line of sight to a second camera with a transversal distance of 6 meters to the first camera and placed at about the same mould distance and 75° for cameras 10 meters apart, 8 meters to the mould. Arranging cameras to capture digital images from a range of different viewing angle may allow for detection of three-dimensional aspects of the upwardly facing surface with increased accuracy and reliably capturing such aspects by one of the cameras.

In a further embodiment, each row may have a distance to neighboring rows in the longitudinal direction between 1 meter and 50% of a longitudinal extent of the mould. The longitudinal extent of the mould being measured from the root end to the tip end of said mould.

If each row has a distance to neighboring rows in the longitudinal direction between 1 meter and 50% of a longitudinal extent of the mould, the pattern of digital cameras may provide a mutual distance between the rows of digital cameras so that the arrangement of digital cameras represents a cost-effective tradeoff between a cost of adding more rows of cameras and increased fidelity of data thereby achieved.

In one aspect, the distance to neighboring rows in the longitudinal direction may be between 2 meters and 25% of said longitudinal extent.

In one aspect, the distance to neighboring rows in the longitudinal direction may be between 3 meters and 10% of said longitudinal extent.

It is contemplated that, to obtain a 50% overlap of the field of view at an object surface for cameras with an angle of view of ‘x’ degrees, the mutual transversal or longitudinal distance ‘d’ to the neighboring camera should be at least: d = tan(x/2) * object_distance, the object_distance being the distance from the camera to the closest point on the object. Accordingly, to implement the system with an object_distance of 10-12 meters to a wind turbine blade mould having a root diameter of 3.5-4 meters, and a 90° angle of view, the mutual distance ‘d’ in a longitudinal direction should be 6-8.5 meters or less. It is understood that, when the mould is configured for producing a half-shell for a wind turbine blade to be joined with another half-shell, only the radius should be subtracted.

In one embodiment of the system, the specific mould section comprises a given extent in the longitudinal direction of the mould, such as a longitudinal section spanning a distance of at least two meters along said longitudinal direction of the mould.

By arranging the plurality of digital cameras in a pattern, so that a set of digital images depicting at least each point of the upwardly facing surface of the mould in a specific mould section is capturable by at least two of the digital cameras, and wherein the specific mould section comprises a given extent in the longitudinal direction of the mould, the system may allow a more focused image capturing and data processing effort directed at only a given extent in the longitudinal direction of the mould. For example, the specific mould section comprises may comprise longitudinal section spanning a distance of at least two meters along said longitudinal direction of the mould, such as the innermost two meters of the root section of the mould, for which data may be most vital for performance prediction to improve design specifications.

In one embodiment of the system, the specific mould section comprises at least the inner 25% or 10 meters of a longitudinal extent of the mould measured from the root end of said mould. It is to be understood that it is the lowest value of 10 meters or the inner 25% of a longitudinal extent of the mould that is to be comprised in the specific mould section. The inventors have seen good indications that the inner 10 meters or 25% of a longitudinal extent of the mould measured from the root end of said mould may be the most important part to include in the specific mould section for acquiring data to produce models which yield sufficiently accurate predictions of the mechanical properties of at least the fiber composite structure of the finished wind turbine blade. It is contemplated that the root section of a wind turbine blade in particularly is subject to higher structural loads during the operation of an assembled wind turbine, and therefor may be the most important aspect to focus on when gathering data e.g. for modelling. Alternatively or additionally, the so-called sparcap may also be of high importance for analyzing structural loads along the length of the mould despite being about 1m wide.

Compared to e.g. structural static load testing, vibrational analysis etc., a system with a plurality of digital cameras arranged in a pattern according to an embodiment of the present disclosure may provide the advantage of allowing the gathering of such data for sections relating directly to the structurally relevant parts of the wind turbine blade.

In one embodiment of the system, the plurality of cameras may comprise at least 10 digital cameras. It is understood that having a greater number of digital cameras, such as at least 10 digital cameras, allows for arranging the cameras in a pattern spanning a substantially fixed volume above the mould being increasingly densely packed with said digital cameras, each placed substantially in said fixed volume above the mould. By arranging the digital cameras more densely in a pattern, the set of digital images depicting at least each point of the upwardly facing surface of the mould in a specific mould section may provide a higher resolution, e.g. measured as the area depicted per pixel at the upwardly facing surface. Alternatively or additionally, having a greater number of digital cameras may provide an increased depth perception related to a 3D reconstruction of the visible part of the upwardly facing surface via stereo imaging. As the mould is generally curved, it may also be necessary to arrange a greater number of digital cameras at or near larger curved sections of the mould, such as near the root or mid sections where the wind turbine blade is substantially cylindrical, the mould e.g. a half-shell or equally cylindrical, having a substantially circular cross-section.

In one aspect, the plurality of cameras may comprise 20 or more digital cameras.

In another aspect, the plurality of cameras may comprise between 50 and 100 digital cameras. The inventors have seen good indications that having between 50 and 100 digital cameras may provide a nearly optimal tradeoff between cost of digital cameras, which may constitute the majority of the cost of establishing the entire system, and the quality of data captured by the arrangement of digital cameras for improved modelling. Typically, an arrangement of 50 and 100 digital cameras may be adequate for capture of useful data for wind turbine blade moulds 50 to 100 meters in longitudinal extent.

In one embodiment of the system, each digital camera may capture a digital image with an image resolution of at least 8 megapixel, such as an image larger than 3264 by 2468 pixels in dimensions. Preferably, the image resolution is 16 megapixel or higher. In another embodiment of the system, the image resolution may be 4 megapixel or more, which can be e.g. realized by having more digital cameras closer to the mould. For example, for two partly overlapping images with image resolutions of 4 megapixel, having at least 15% of the upwardly facing surface of the mould depicted in one of said two images also be depicted by another of said two digital images may correspond to about 0.6 megapixel of visual information in common, or e.g. about 3000 x 200 pixels.

In one aspect, the digital image may be in any raster image format comprising a plurality of pixels arranged in a fixed number of rows and columns, e.g. with at least 4290 pixels along one dimension and at least 2800 pixels along another dimension.

In one embodiment of the system, each digital camera may be configured to capture a digital image comprising a plurality of pixels corresponding to one pixel in said digital image depicting a substantially rectangular area of less than 25 mm ² measured at the upward facing surface of the mould, such as less than 5 mm ² . The plurality of pixels of the digital image may correspond to one pixel depicting a square smaller than 3 by 3 mm in dimensions at said distance. By each pixel depicting a substantially rectangular area of less than 25 mm ² , such as less than 5 mm ² , the digital image provide a pixel density sufficiently high for capturing small details of the glass fiber fabric placed on the upward facing surface of the mould. This may be achieved despite the digital camera being positioned at a distance from the mould so as not to obstruct or impinge the layup process.

In one embodiment of the system, the plurality of digital cameras may be arranged such that each field of view at the mould partially overlaps with the fields of view of 10 or less other digital cameras. In another embodiment of the system, the plurality of digital cameras may be arranged such that each field of view at the mould partially overlaps with the fields of view of between 2 and 10 other digital cameras.

The inventors have seen good indications that each digital camera being arranged such that the field of view of said digital camera overlaps with the field of view of 10 or less other digital cameras may provide a nearly optimal tradeoff between the cost of adding more digital cameras, and the increased depth perception related to a 3D reconstruction of the visible part of the upwardly facing surface via stereo imaging. It is understood that the field of view of one camera may overlap with the fields of view of one or more of the other digital cameras based on at least two digital cameras being arranged to capture at least partly overlapping digital images.

In one embodiment of the system, the distance of the plurality of digital cameras to the mould may be between 4 and 15 meters. In one aspect, the distance may be between 8 and 12 meters. The distance may be determined as the median value of the shortest distance measured from each digital camera to the mould.

For example, the median value of the shortest distances measured from ‘N’ respective digital cameras to the same mould is determined as the median value calculated for ‘N’ measurements of the shortest distance measured for a first digital camera to the mould, then for the second digital camera to the mould etc.

For e.g. three cameras, a first camera with a shortest distance of 9 meters to the mould, a second camera with a shortest distance of 6.5 meters to the mould and a third camera with a shortest distance of 8 meters to the mould, the median value is determined to be 8 meters (being the ‘middle’ value). It is understood that the shortest distance from a digital camera to the mould is measured in a straight line from the lens of said digital camera to the closest integral part of the upwardly facing surface of the mould, such as e.g. where a glass fiber fabric is placed. In one embodiment of the system, each digital cameras may have a focal axis angled substantially towards the mould. The angle of mutual intersection of said focal axis with the focal axis of another digital camera having partially overlapping fields of view may be between 15° and 60°. In one aspect, the angle of mutual intersection may be between 30° and 50°. In another aspect, the angle of mutual intersection may be about 45°. The angle of mutual intersection may be measured in a vertical plane defined by the two cameras. In this way, despite two differently oriented digital cameras generally having focal axes which do may not intersect in three-dimensional space as such, their angle of mutual intersection may be defined as the angle of mutual intersection of two other axes being the respective projection of these focal axes onto said vertical plane.

In one embodiment of the system, the plurality of digital cameras may be stationary with respect to each other and/or the mould.

In one aspect, the digital cameras may be rigidly affixed to a mount unit suspended below a ceiling. This may e.g. provide faster installation or rearrangement of cameras without impinging on the manufacturing operations of various machinery, e.g. cranes.

In one embodiment, the system may further comprise a processing module having a computer-readable memory. The processing module may be configured to calculate a three-dimensional position of a plurality of points at the upwardly facing surface of the mould in said specific mould section, and provide a three-dimensional reconstruction of at least parts of the upwardly facing surface of the mould, based on the set of digital images and storing said three-dimensional reconstruction to the memory.

In a further embodiment, the system may further comprise one or more fiducial markers of a predetermined size, such as Quick Response (QR) codes, placed at or printed on the mould or fabric materials placed thereon. The processing module may further be configured to detect said one or more fiducial markers in the set of digital images. The calculating the three-dimensional position of the plurality of points may be based at least in part on an apparent size of each fiducial marker in each digital image. It is contemplated that using AprilTag or the like as fiducial markers may provide more precise detection of 3D position, orientation and identity of tags relative to the camera. In yet a further embodiment, the system may further comprise an interface device connected to the processing module. The interface device may be configured to receive data, such as data representing the set of digital images, from the plurality of digital cameras and to provide the three-dimensional reconstruction from the memory after being calculated.

In one aspect, the three-dimensional reconstruction may be provided to an external party or client.

Further disclosed herein is a method of obtaining a three-dimensional scanning of a wind turbine blade mould extending substantially horizontally in a longitudinal direction of the mould using a plurality of digital cameras arranged with at least partly overlapping fields of view above the mould.

The mould may have an upwardly facing surface for supporting fabric materials placed thereon. Alternatively or additionally, fabric materials placed on the mould may form part of the upwardly facing surface for supporting fabric materials.

The method may comprise an act of obtaining from at least two of the digital cameras a set of digital images depicting at least each point of the upwardly facing surface of the mould in a specific mould section.

An elevation angle between a line of sight of each of said two digital cameras and a tangent plane at each of said points may be between 25° and 90°.

The plurality of digital cameras may be arranged to capture at least partly overlapping digital images, wherein at least 15% of a first part of the upwardly facing surface of the mould capturable by one of said plurality of digital cameras is also capturable by one other camera of said plurality of digital cameras, which other camera furthermore is arranged to capture at least 15% of a second part of the upwardly facing surface of the mould capturable by yet another camera of said plurality of cameras.

Alternatively or additionally, the at least two digital cameras may be arranged to capture at least partly overlapping digital images wherein at least 50% of the upwardly facing surface of the mould captured by one of said at least two digital cameras is also capturable by another of said at least two digital cameras.

The method may further comprise an act of calculating a three-dimensional position of a plurality of points at the upwardly facing surface of the mould in said specific mould section, thereby providing a three-dimensional scanning of at least parts of the upwardly facing surface of the mould, based on the set of digital images.

The effects and advantages of this method may include those already described either individually or in combination for disclosed embodiments of the system comprising a wind turbine blade mould extending substantially horizontally in a longitudinal direction of the mould, and a plurality of digital cameras arranged with at least partly overlapping fields of view above the mould.

Alternatively, the disclosed method may be for obtaining a three-dimensional scanning of a wind turbine blade or parts thereof, wherein the wind turbine blade or parts thereof are separated from the mould prior to scanning.

In yet another alternative, the disclosed method may be for imaging a fabric material placed on a wind turbine blade mould or on previously placed layers of fabric material, such that a three-dimensional scanning of the construction of the wind turbine blade layer-by-layer can be obtained. This may include obtaining information of comprised edges, wrinkles or other irregularities in said fabric material detectable from said imaging based on either intensity information, depth information or a combination thereof. Intensity information may include pixel color intensity, greyscale intensity etc.

In one embodiment of the method, the act of calculating a three-dimensional position of the plurality of points may include performing stereo photogrammetry. The three- dimensional position of the plurality of points may be calculated relative to a known three-dimensional reference point, such as a reference point at the root end. The three-dimensional position of the plurality of digital cameras may likewise be known or calculated relative to said three-dimensional reference point, e.g. by prior calibration.

In one embodiment, the method may further comprise an act of detecting one or more fiducial markers, placed at or printed on the mould or fabric materials placed thereon, in the set of digital images.

In a further embodiment of the method, the act of detecting one or more fiducial markers may include determining an apparent size and spatial orientation of each fiducial marker, thereby localizing each fiducial marker in relation to each of the relevant digital cameras providing a digital image depicting said fiducial marker. In yet a further embodiment of the method, the act of calculating the three-dimensional position of the plurality of points may be based, at least in part, on an apparent size of each fiducial marker in each digital image.

In one embodiment of the method, the method may comprise a further act of positioning the mould in relation to the plurality of digital cameras to put a specific mould section within the at least partly overlapping fields of view.

In one aspect, the specific mould section may be substantially centered in the at least partly overlapping fields of view.

In one embodiment of the method, the method may comprise a further act of selecting or omitting a specific mould section, or subsections thereof, based on a known position and orientation of the mould in relation to unobscured lines of sight of each of the plurality of digital cameras.

In one embodiment of the method, the method may comprise a further act of arranging the plurality of digital cameras in a pattern along the longitudinal and transversal directions of the mould, preferably providing a substantially even distribution of digital cameras above a substantially horizontal plane, such as a parting plane of a two-part mould.

Further disclosed herein is a non-transitory computer readable medium storing a computer program product comprising instructions which, when executed by a hardware processor of a processing module in a system comprise a processing module, may be configured for provisioning the method of obtaining a three- dimensional scanning of a wind turbine blade mould.

Alternatively, the processing module may be configured for provisioning the method of obtaining a three-dimensional scanning of a wind turbine blade or parts thereof, wherein the wind turbine blade or parts thereof are separated from the mould prior to scanning.

In yet another alternative, the processing module may be configured for provisioning the method for imaging a fabric material placed on a wind turbine blade mould or on previously placed layers of fabric material, such that a three-dimensional scanning of the construction of the wind turbine blade layer-by-layer can be obtained. This may include obtaining information of comprised edges, wrinkles or other irregularities in said fabric material detectable from said imaging based on either intensity information, depth information or a combination thereof.

Further disclosed herein is the use of a system for obtaining a three-dimensional scanning of a wind turbine blade or parts thereof, wherein the wind turbine blade or parts thereof may be separated from the mould prior to scanning.

The system may be according to any of the previously described embodiments of the system comprising a wind turbine blade mould extending substantially horizontally in a longitudinal direction of the mould, and a plurality of digital cameras arranged with at least partly overlapping fields of view above the mould.

The effects and advantages of this use may include those already described either individually or in combination for the disclosed embodiments of the system and for any of the disclosed methods for obtaining a three-dimensional scanning or 3D imaging.

The camera array may also be used in later production steps, e.g. finalisation, where fiducial markers can be used to quantify blade deformation during curing of the resin, release of the blade from the mould and under subsequent non-destructive testing of the blade to reveal structural characteristics without moving the blade into a testing rig. This may provide the advantage of reducing costs of quality assurance or inspection.

The above mentioned system may be a system comprising a wind turbine blade mould extending substantially horizontally in a longitudinal direction of the mould, and a plurality of digital cameras arranged with at least partly overlapping fields of view above the mould. The system may further comprise a source of structured light, preferably monochromatic structured light, such as a laser projection system.

Further disclosed herein is the use of a system for imaging a fabric material placed on a wind turbine blade mould, wherein edges, wrinkles or other irregularities in said fabric material may be detectable from said imaging based on either intensity information, depth information or a combination thereof.

The system may be according to any of the previously described embodiments of the system comprising a wind turbine blade mould extending substantially horizontally in a longitudinal direction of the mould, and a plurality of digital cameras arranged with at least partly overlapping fields of view above the mould.

The effects and advantages of this use may include those already described either individually or in combination for the disclosed embodiments of the system and for any of the disclosed methods for obtaining a three-dimensional scanning or 3D imaging.

In one aspect, the use may include using structured light, preferably monochromatic structured light.

Brief description of the drawings

Figure 1 illustrates one embodiment of the system comprising a mould of a wind turbine blade and a plurality of digital cameras.

Figure 2 illustrates two embodiment of two digital cameras arranged with partly overlapping fields of views of the upward facing surface of the mould.

Figure 3 illustrates an embodiment of the points on an upward facing surface of a mould comprising two half-shells and a mandrel.

Figure 4A illustrates two embodiment of an image of an upward facing surface of a fabric material. Figure 4B illustrates two embodiments of contrast enhancing an image to detect details of a fabric material, such as e.g. the images illustrated in Fig. 4A.

Figure 5 illustrates multiple embodiments of glass fiber fabrics and structures herein.

Figure 6 illustrates another embodiment of a pattern of digital cameras arrange above a mould.

Figure 7 illustrates another embodiment of a pattern of digital cameras arrange above a specific mould section, such as e.g. the a section of the mould illustrated in Fig. 6.

Figure 8 illustrates multiple embodiments of a plurality of digital cameras, viewed from above, arranged above a mould to capture at least partly overlapping digital images. References

1 system

10 mould

11 mandrel

12 longitudinal direction

14 transversal direction

16 upwardly facing surface

18 mould section

20 digital camera

22 field of view

24 elevation angle

26 line of sight

30 fabric materials

32 tow direction

40 pattern

42 two-dimensional array

44 row of digital cameras

46 longitudinal distance

48 transverse distance

50 digital image

52 direction of least overlap

60 root end

62 tip end

64 longitudinal mould extent

68 mould distance

70 point

72 tangent plane

74 surface normal

80 longitudinal section

90 focal axis

92 angle of mutual intersection of two focal axes

100 three-dimensional scanning

102 three-dimensional position

104 fiducial marker

106 hardware processor

108 computer-readable memory Description of examples

Various examples are described hereinafter with reference to the figures. Like reference numerals refer to like elements throughout. Like elements will, thus, not be described in detail with respect to the description of each figure. It should also be noted that the figures are only intended to facilitate the description of the examples. They are not intended as an exhaustive description of the claimed invention or as a limitation on the scope of the claimed invention. In addition, an illustrated example needs not have all the aspects or advantages shown. An aspect or an advantage described in conjunction with a particular example is not necessarily limited to that example and can be practiced in any other examples even if not so illustrated, or if not so explicitly described.

In the drawings, thicknesses of a plurality of layers and areas are illustrated in an enlarged manner for clarity and ease of description thereof. When a layer, area, element, or plate is referred to as being “on” another layer, area, element, or plate, it may be directly on the other layer, area, element, or plate, or intervening layers, areas, elements, or plates may be present there between. Conversely, when a layer, area, element, or plate is referred to as being “directly on” another layer, area, element, or plate, there are no intervening layers, areas, elements, or plates there between.

Further when a layer, area, element, or plate is referred to as being “below” another layer, area, element, or plate, it may be directly below the other layer, area, element, or plate, or intervening layers, areas, elements, or plates may be present there between. Conversely, when a layer, area, element, or plate is referred to as being “directly below” another layer, area, element, or plate, there are no intervening layers, areas, elements, or plates there between.

The spatially relative terms “lower” or “bottom” and “upper” or “top”, "below", "beneath", "less", "above", and the like, may be used herein for ease of description to describe the relationship between one element or component and another element or component as illustrated in the drawings. It will be understood that the spatially relative terms are intended to encompass different orientations of the device in use or operation, in addition to the orientation depicted in the drawings. For example, in the case where a device illustrated in the drawings is turned over, elements described as being on the “lower” side of other elements, or "below" or "beneath" another element would then be oriented on “upper” sides of the other elements, or "above" another element. Accordingly, the illustrative term "below" or “beneath” may include both the “lower” and “upper” orientation positions, depending on the particular orientation of the figure. Similarly, if the device in one of the figures is turned over, elements described as “below” or “beneath” other elements would then be oriented ’’above” the other elements. The exemplary terms “below” or “beneath” can, therefore, encompass both an orientation of above and below, and thus the spatially relative terms may be interpreted differently depending on the orientations described.

The terminology used herein is for the purpose of describing particular examples only and is not intended to be limiting. As used herein, the singular forms “a,” “an,” and “the” are intended to include the plural forms, including “at least one,” unless the content clearly indicates otherwise. “At least one” is not to be construed as limiting “a” or “an.” It will be further understood that the terms “comprises," "comprising,"

"includes" and/or "including," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

It will be understood that, although the terms “first,” “second,” “third,” and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. Thus, “a first element” discussed below could be termed “a second element” or “a third element,” and “a second element” and “a third element” may be termed likewise without departing from the teachings herein.

"About" or "approximately" as used herein is inclusive of the stated value and means within an acceptable range of deviation for the particular value as determined by one of ordinary skill in the art, considering the measurement in question and the error associated with measurement of the particular quantity (i.e. , the limitations of the measurement system). For example, "about" may mean within one or more standard deviations, or within ± 30%, 20%, 10%, 5% of the stated value.

Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by those skilled in the art to which this invention pertains. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined in the present specification. Figure 1 is a graphical illustration of one embodiment of the system 1 comprising a mould 10 for a wind turbine blade and a plurality of digital cameras 20.

The mould 10 extends substantially horizontally in a longitudinal direction 12 and a transversal direction 14. The mould comprises an upwardly facing surface 16 for supporting fabric materials 30 to be placed thereon. The fabric material is not illustrated in this figure. The upwardly facing surface 16 is illustrated as being divided into specific mould sections 18 illustrated by the dotted lines. The mould sections has a longitudinal section spanning 80, which can be different for each of the sections 18. The mould has a root end 60 and a tip end 62 and a longitudinal mould extend 64, extending from root end to tip end.

The plurality of digital cameras 20 are arranged with at least partly overlapping fields of view 22 at the upward facing surface 16 of the mould 10. The digital cameras are arranged above the upward facing surface 16 of the mould 10. The digital cameras are furthermore arranged, such that each point of the upward facing surface 16 of the mould in a specific mould section18 is capturable by at least two of the digital cameras 20, i.e. a set of digital images 50 depicting at least each point of the upwardly facing surface of the mould in a specific mould section 18 may be captured by at least two of the digital cameras 20. Cameras 20 may be fixed or gimbal-mounted with a zoom lens.

The plurality of digital cameras 20 may be arranged in a pattern 40, here simply a pattern in the form of a 2x3 regular grid defined by two near-orthogonal axes.

Figure 2 is a graphical illustration of two embodiment of the system 1 having two digital cameras 20 arranged with partly overlapping fields of views 22 of the upward facing surface 16 of the mould 10.

Figure 2A illustrates a mould 10 with a convex upward facing surface16. Figure 2B illustrates a mould 10 with a concave upward facing surface 16. Both embodiments are illustrated from a side illustrating a cross section of the system 1 along the transversal direction of the mould 10. The two cameras 20 in both embodiments are arrange with a transversal distance 48 between the cameras and a mould distance 68 from the cameras to the mould. The field of view 22 of both cameras is illustrated by the outer lines of sight 26. Both cameras have a focal axis 90 slightly angle towards each other and substantially towards the mould, such that they forms an angle of mutual intersection of the two focal axes 92.

Figure 2B illustrates the elevation angle 24 between a line of sight 26 of a digital camera 20 and a tangent plane 72. The tangent plane 72 is linked to a specific point. For the line of sight 26 coincident with the focal axis, the elevation angle 24 is 90°. For the line of sights diverging from the focal axis 90, the elevation angle 24 is lower and hence for both illustrated embodiments, the elevation angle 24 decreases for the line of sights taken further away from the focal axis 90.

The overlapping fields of view 22 on the mould surface 16 encompasses a number of points 70 on the upwardly facing surface of the mould to be depicted by both cameras 20. The points on an upward facing surface is further illustrated in the embodiments in figure 3.

Figure 3 is a graphical illustration of two embodiments of the points 70 on an upward facing surface 16 of a mould 10. Two half-shells of a mould 10 of different shape are illustrated along with a mandrel 11, whereby blades may be cast in one or more parts.

It is understood that different parts of the mould 10 may be present under the cameras 20 during various stages of the layup process of positioning and fastening a plurality of lengths of fabric 30 (not shown) on the mould 10, e.g. first placing fabric 30 onto a first upwardly facing surface 16 of the lower half-shell of the mould 10, then onto a second upwardly facing surface 16 of the mandrel 11, and finally closing the mould 10 with the upper half-shell of the mould 10 so as to support and/or enclose the placed lengths of fabric 30 in preparation for vacuum assisted resin transfer moulding. In some cases, a plurality of the cameras 20 may also be used for flow tracing during resin infusion, e.g. when the blade is cast one half shell at the time using two half-shells of the mould 10.

In figure 3, a given point 70 on the upper half-shell of the mould 10 is shown being captured by the two left-most cameras 20. An elevation angle 24 between a tangent plane 72 defined by a surface normal 74 of the upwardly facing surface 16 at that point 70 and a first line of sight 26 from said point 70 to the leftmost digital camera 20 and is shown being between 25° and 90°, i.e. approximately 75°. Further, the elevation angle 24 between said tangent plane 72 and a second line of sight 26 from said point 70 to the other digital camera 20 is shown also being between 25° and 90°, i.e. about 70°. Figure 4A illustrates two embodiment of an image 50 of an upward facing surface 16 of a fabric material 30.

Building a wind turbine blade can include manufacturing a half-shell of a fiber composite blade. The half-shell may typically be manufactured according to a layup design specifying a two-part mould 10 split along a parting plane. Lengths of material fabrics 30 may be arranged at predetermined placement positions on the upward facing surface 16 of one part of the mould 10 or on top of previously arranged fabric material 30. . A single web length of fabric material 30 may be arrange such that at least a portion of the fabric 30 is within line-of-sight to one or more digital cameras 20 of the system. A single web length of fabric material 30 may furthermore be arrange such that the web fabric 30 may extend over only a single mould section or multiple mould sections, preferably extending a substantial portion of the longitudinal extent of the entire mould.

The digital cameras 20 may capture digital image 50 obtained as individual pixels. The information relating to said individual pixels may be uniquely localized to a corresponding three-dimensional position 102 or region of the fabric 30 depicted. This may be achieved by said pixel by transforming from the position of said pixel in a sensor-centric coordinate system of a specific digital camera 20 to a placement position in three-dimensional space in relation to the mould via the coordinate system of the mould.

The embodiments in figure 4A shows a region of a unidirectional glass fiber fabric having substantial parallel tows. Each tow comprises a plurality of fiber filaments aligned with a tow direction illustrated by the double-arrowed straight line. The spatial separation of two neighboring substantial parallel tows, the tow distance is typically in the order of a few millimeters or less.

The plurality of lengths of fabric material 30 may be cut from substantial identical fabrics having identical or comparable tow distances between neighboring tows. However, due to the very inhomogeneous and anisotropic material properties arising from fiber filaments within said fabrics, even slight deviations in tow distance or the like may be important to detect or image to predict the mechanical properties of the finished fiber composite structure. Such deviations may be detectable relative to the substantially pristine highlighted regions shown in dashed squares in both images. When placing the fabric on a curved surface or during distribution of resin between layers of fabric, other deviations in the arranged fabric compared to the predetermined or modelled placement positions may arise. Such deviations may also be important to detect or image to predict the mechanical properties of the finished fiber composite structure. The circular highlighted sections in the images illustrates a tow spilt and a wrinkle in the fabric, respectively. Fabric wrinkles or other deviations originating in one layer may also “break through” to abutting layers.

Figure 4B illustrates two embodiments of contrast enhancing an image 50 to detect details of the fabric material 30, where the images shown in Fig. 4A have been post processed by increasing the image contrast and applying a black/white threshold. It is apparent that individual tows, markers and fabric defects, such as in the highlighted regions, may be recognized and processed to measure properties of the fabric 30.

Figure 5 illustrates multiple embodiments of information provided by a 3D-scanning 100 of an upward facing surface 16 of a fabric material 30 based on physical aspects. For example, a fabric material 30 comprising tows of fibers uniformly arranged along a tow direction 32, or woven from tows of fibers, may be visually recognizable cf. Fig. 4.

Figure 5A is a graphical illustration of a cross-section of a unidirectional glass fiber fabric having substantial parallel tows, each tow comprises a plurality of fiber filaments aligned with a tow direction illustrated by the arrow. The tow distance being the spatial separation of two neighboring tows is typically in the order of a few millimeters or less. The fabric lengths may be cut from substantial identical fabrics having identical or comparable tow distances between neighboring tows. However, due to the very inhomogeneous and anisotropic material properties arising from fiber filaments within said fabrics, slight deviations in tow distance, arrangement of the fibers within a single tow, arrangement of fibers in different tows or is arranged or combinations hereof may be important features for predicting the mechanical properties of the finished fiber composite structure. The fabric material may be biaxial or triaxial glass fiber or the like.

Figure 5B is a graphical illustration of a cross-section of a woven bidirectional glass fiber fabric having substantial parallel weft tows woven together by warp tows substantial perpendicular to weft tows. Each tow comprising a plurality of fiber filaments aligned to the relevant tow direction. As also discussed in relation to figure 5A, the spatial separation of two neighboring substantial parallel tows may be determined by at least one geometrical parameter being measurable or detectable from above. Such geometrical parameters may be important for accurate modelling.

Figure 5C is a graphical illustration of a fabric wrinkle in a fabric having substantial parallel tows, such as for example illustrated by the circular highlighted region in Fig.

4. The fabric wrinkle is detectable from the visible deviation of neighboring tows from substantially straight lines with some tows instead having a discontinuous jump or shift at the fabric wrinkle. Such a fabric wrinkle may be detectable in a digital image by determining the apparent tow direction 32 of neighboring tows and looking for such discontinuities or other departures from substantially straight lines. Beyond features directly identified in the digital images 50, providing a 3D reconstruction 100 may also reveal detailed information of wrinkles based on local depth deviation in the vicinity of said wrinkle. Depth information may also allow for improved detection. The 3D position 102 of the point may be determined based on fiducial markers 104 on the fabric and by using overlapping fields of view resulting in depth information of a specific point.

Figure 5D is a graphical illustration of a fabric fold line in a fabric having substantial parallel tows. The fabric fold line is likewise detectable from visible deviation of neighboring tows from substantially straight lines with some tows but instead showing only minor deviations at the fabric fold line. While possibly not as severe as fabric wrinkles, fabric fold lines may be signs that the fabric has been subjected to bending or folding, e.g. during an initial stage of layup where an operator could have stepped on the length of fabric, and may therefore still be of interest to detect in a digital image as outlined above.

Figure 5E is a graphical illustration of a tow split in a length of fabric having substantial parallel tows, a tow split was illustrated and highlighted in figure 4A. The tow split may also be detectable from the visible deviation of neighboring tows from substantially straight lines with some tows leaving a central region with an increased tow separation between neighboring tows deviating in different directions perpendicular to the overall tow direction. Larger tow splits may lead to undesirable fabric perforations, e.g. for unidirectional glass fiber fabric which resists tear forces in only one direction.

Figure 5F is a graphical illustration of a tow fracture in a fabric having substantial parallel tows. The tow fracture may not be detectable from a deviation of the substantial parallel tows from substantially straight lines as such, but may instead be detectable by tracing the tows along the tow direction across a discontinuous jump or shift at the tow fracture. It may be contemplated that a significant tow fracture spanning a plurality of neighboring tows may appear as visible edges at the gaps between tow ends especially when the fabric is stretched during placement in the mould.

Figure 5G is a graphical illustration of a fabric deformation having substantial parallel tows. The fabric deformation is shown as a non-uniform combination of tensile and shear deformation particularly in the lower right region of the length of fabric, where neighboring tows are substantially parallel to each other but with each tow having a tow direction or curvature thereof leading said tow to be less parallel to tows further away. Most fabric deformation may be detectable as deviations of the substantial parallel tows from the substantially straight lines the tows form in an un-deformed state, e.g. by misalignment of tows or directional distortion of the fabric. However, some deformation may not be detectable e.g. tensile strain along the tow direction. Furthermore, fabric deformations may also originate from the fabric being placed on or draped over a surface having a curved three-dimensional geometrical shape.

One- or two-dimensional aspects of a fabric deformation may be detectable in a digital image by determining at least the apparent tow directions, the curvature of said tows and mutual angle to each other, e.g. based on multiple fiducial markers 104 shifting.

Figure 5H is a graphical illustration of another visual aspect in the form of an edge cut into a length of fabric having substantial parallel tows of fibers. The fabric edge may be detectable as a linear alignment of neighboring tow ends of neighboring tows, but may also be visible as a clear boundary towards a visibly different background (shown here as darker). It is understood that a 3D placement position 102 of any reference point 70 on a fabric material lengths may be determined from a boundary position derived from said fabric edge, e.g. by performing edge detection on a digital image of said length of fabric using algorithms known in the art. It is further contemplated that markers or ink may be printed onto at least a portion of said fabric edge to provide a greater visual contrast at the boundary position, e.g. by in-woven dyed fibers with high contrast.

Figure 6 illustrates one embodiment of a pattern 40 of digital cameras 20 arranged above a mould 10 and a processor module 106 providing a 3D scanning 100 thereof.

The cameras 20 are arranged facing the upward facing surface 16 of the mould and arranged in a two-dimensional array 42. The illustrated array comprises four rows 44 of digital cameras 20, each row extends in a transversal direction 14 to the mould 10 and the rows 44 are separated by a longitudinal distance 46 in the longitudinal direction 12 to the mould 10.

The array is illustrated with one configuration having two digital cameras 20 in each row 44 separated by a transverse distance 48 in the transversal direction 14 of the mould.

In the figure, another configuration of the array is indicated, having four digital cameras 20 in each row 44. However, other configuration having other numbers of rows and/or number of digital cameras in each row may be used. It is hereby understood that the number of cameras in each row may vary as needed along the length of the blade, which may provide adequate camera coverage of the mould with a minimum of cost.

The cameras 20 may furthermore be arrange such that the longitudinal distance between neighboring rows are the same throughout the array. Likewise may the cameras in each row be arranged such that the transverse distance 48 between neighboring cameras 20 within one row 44 or throughout the array 40 are the same.

Figure 7 illustrates another embodiment of the system 1 with a pattern 40 of digital cameras 20 arrange above the mould 10.

The cameras 20 are arranged facing the upward facing surface 16 of the mould and arranged along a curvilinear plane being substantially peripheral to the upwardly facing surface 16 of the mould. Hence, the cameras 20 are arranged such that the entire upward facing surface 16 is covered by the field of view of the array of cameras 20.

The field of view of each camera 20 is illustrated by the line of sights 26 illustrated with solid lines for the outer cameras and with dashed lines for the middle camera(s). The dashed lines extending backwards from the cameras indicates that multiple rows can be arranged alone the longitudinal direction of the mould.

In the figure, the overlap of fields of view for each camera is illustrated by the hatched areas. As illustrated additional cameras may be arranged along the curvilinear plane of the camera array 40 to obtain a complete overlap of at least two fields of view across the entire surface 16. Figure 8A illustrates an embodiments of a plurality of digital cameras 20 arranged pattern 40 above a mould 10 to capture at least partly overlapping digital images 50 of a specific mould section 18 of said mould 10. The field of view 22 of the plurality of digital cameras 20 is shown as pyramid or prism shapes extending from each digital camera 20 downwards towards the mould 10 producing digital images 50 at the plane of said mould 10. The plurality of digital cameras 20 are shown arranged to capture three partly overlapping digital images 50. A first hatched area illustrates the overlap between a first digital image 50 and a second digital image 50, wherein about 20% of the upwardly facing surface 16 of the mould 10 capturable by one of said plurality of digital cameras 20 to produce said first digital image 50 is also capturable by another camera 20 of said plurality of digital cameras 20 to produce said second digital image 50. A second hatched area illustrates the overlap between the second digital image 50 and a third digital image 50, wherein said other camera 20 is furthermore arranged to capture about 20% of the upwardly facing surface 16 of the mould 10 capturable by yet another camera 20 of said plurality of cameras 20 to produce said third image 50. The partially overlapping digital images 50 are mutually connectable by their overlap being capturable by at least two of the digital cameras 20, and the inventors have seen indication that reliable image stitching can be achieved with at least 15% of overlap.

Alternatively or additionally, a direct of least overlap 52 shown at the first hatched area illustrates a partially overlap of at least 50 pixels, such as at least 100 pixels, such as at least 200 pixels, measured as the number of pixels in said direction of least overlap. Likewise, a direction of least overlap 52 shown at the second hatched area illustrates a partially overlap of at least 50 pixels, such as at least 100 pixels, such as at least 200 pixels, measured as the number of pixels in said direction of least overlap. Hereby, an image stitching type algorithm can e.g. combine the first, second and third image 50 to produce a panoramic image and depth information for at least these hatched areas.

Figure 8B illustrates an embodiments of a plurality of digital cameras 20 arranged in a substantially two-dimensional pattern 40 above a mould 10 to capture at least partly overlapping digital images 50 of a specific mould section 18 of said mould 10. Six rows 44 of digital cameras 20 shown in the pattern 40 thus form a two-dimensional array 42, with two digital cameras 20 in each row 44 provided at a transversal distance 48 from each other, and each row provided at a longitudinal distance 46 from each other. The partially overlapping digital images 50 are shown to overlap in such a way that each point 70 (not shown) of the upwardly facing surface 16 of the mould 10 is capturable by at least two of the digital cameras 20, thus providing empirical data which includes depth information for at least a specific mould section 18 (not shown) of the mould. It is noted that the partly overlapping digital images 50 are shown here in an idealized way, where effects of perspective on the field of view 22 (not shown) of each digital camera 20 and distortion of said images 50, e.g. due to curvature of the mould 10, are omitted. In this example, empirical data may be obtainable for the upwardly facing surface 16 of the mould 10 by all, or almost all, points 70 (not shown) of the upwardly facing surface 16 being capturable by at least two digital cameras 20 via partially overlapping images.

Figure 8C illustrates another embodiments of a plurality of digital cameras 20 arranged and oriented in three-dimensions in a substantially two-dimensional pattern 40 above a mould 10 to capture at least partly overlapping digital images 50 of a specific mould section 18 of said mould 10. Because each digital camera 20 is oriented at an angle to the mould 10 and each other, the field of view 22 of each digital camera 20 is shown to correspond to capturing a digital image 50 depicting a trapezoid area, or generally a four-sided polygon area, in the plane of the upwardly facing surface 16 of the mould. Further, the plurality of digital cameras 20 shown here are placed at different mounting positions being substantially equidistant to the upwardly facing surface 16 of the mould 10, so that each digital camera 20 has a smallest distance 68 (not shown) downwards to the upwardly facing surface 16 of the mould 10 which is about equal to the smallest distance 68 for the other digital cameras 20 to said upwardly facing surface 16. This may provide that all partly overlapping images 50 captured from the plurality of digital cameras 20 depict points 70 (not shown) at said upwardly facing surface 16 with each pixel in said partly overlapping images 50 depicting a substantially rectangular area of less than 25 mm ² at the upward facing surface 16 of the mould 10. Particularly, with the digital cameras 20 configured to capture two or more partly overlapping images 50 with a substantially identical image resolution, placing digital cameras 20 at positions substantially equidistant to the mould 10 may provide that pixels from different digital cameras 20 depict substantially the same size of rectangular area at the upward facing surface 16 of the mould 10, which allows the partly overlapping digital images 50 to be combined equally into a larger panoramic image 50 (not shown) via image stitching. This may e.g. require that a first digital image 50 capturable by one of said plurality of digital cameras 20 partially overlaps a second digital image 50 capturable by another camera 20 of said plurality of digital cameras 20 by at least 50 pixels, such as at least 100 pixels, such as at least 200 pixels, measured as a number of pixels in a direction of least overlap 52, as shown for the hatched area illustrating an overlapping region of said first and second digital images 50 depicting the upwardly facing surface 16.