Non-destructive testing method for a batch of industrial parts using X-ray tomography

Abstract

The invention relates to a parts inspection by X-ray tomography which includes: the acquisition, with a beam blocking grid (4) interposed between an X-ray source (1) and a part of the batch, of calibration intensity images according to different projection angles, the grid comprising X-ray attenuation elements (EA) supported by a support panel and, for each projection angle, the determination of a scattered radiation image from the calibration intensity image acquired according to that projection angle;the replacement of the grid with a replacement panel having an X-ray attenuation identical to that of the support panel and, for each of the pieces in the batch, the acquisition of raw intensity images according to the different projection angles, and the correction, for each projection angle, of the raw intensity image acquired for that projection angle by means of the scattered radiation image determined for that projection angle. Figure for the abstract: Fig. 1;

Description

Title of the invention: Method for non-destructive testing of a batch of industrial parts by X-ray tomography Technical field

[0001] The field of the invention is that of non-destructive testing of industrial parts using the X-ray absorption tomography technique. The invention finds particular application in the testing of aeronautical parts, in particular turbomachine blades. Prior art

[0002] The blades equipping the fan rotors placed at the front of aircraft engines have seen their size increase with the advent of high bypass ratio engines. Thanks to a composite material using carbon fibers woven in a 3D geometry and cemented with an epoxy-type resin, these blades retain a good part of their mechanical resistance to deformation and impacts while becoming lighter.

[0003] Various indications may nevertheless appear during the manufacture of these blades, such as the presence of foreign bodies or porosities in the solidified resin, imperfections in the weaving or even non-conformity of the final shape. These indications, despite their small size (of the order of a hundredth of a millimeter) compared to the dimensions of the blades (of the order of one to several meters), may constitute weak points in the part. Therefore, their control is essential for their exploitation.

[0004] Currently, an industrial non-destructive testing system based on the conical tomography technique ensures the control of most of these indications. The volume information of the internal morphology of the object, or tomography, is more precisely obtained by applying a reconstruction algorithm to successive radiographs (projections) taken on the rotating object. They correspond to the shots at different successive angles of the imaged object.

[0005] However, the reduction of manufacturing costs and the development of new blades with ever larger sizes require the improvement of this system, in particular the reduction of control time.

[0006] One of the challenges to be addressed in the context of this improvement is the presence of artifacts (i.e. inconsistencies in the image resulting from the mathematical reconstruction or the physical acquisition) which degrade the quality of the tomographic image and harm, or even prevent, the ability to detect indications.

[0007] One of the most recurring artifacts in cone geometry tomography is the artifact associated with scattered radiation. This artifact results from the secondary emission of X-rays from the object. Indeed, X-rays coming from the X-ray source (primary) and interacting with the object can induce the emission of X-rays from the object (secondary). These secondary rays reach the X-ray detector and modify the ratio between the X-rays passing through the object and interacting with the object.

[0008] Hardly visible on radiographic images, these scattering artifacts can appear in reconstructed images as lines extending along certain edges of the structure of the object ("streaking"), a diffuse and homogeneous variation (low frequency noise) of the gray levels or even a homogeneous deformation of the gray levels inside the objects.

[0009] Current tomographic NDT systems for 3D woven composite blades use energies below 200 keV. The secondary rays are therefore at low energies and not very visible. However, with the reduction in acquisition time and the consequent increase in noise in the projections, scattered artifacts are increasingly visible. This problem is even more encountered with conical tomographs which use large X-ray detectors in order to have a large field of vision for large blades but which then capture more scattered radiation.

[0010] Known techniques for correcting artifacts due to scattered radiation in cone tomography systems can be classified into two categories, a first called software based on modeling of the scattered radiation signal or numerical techniques to suppress it and a second called hardware based on empirical estimation of the scattered radiation signal or barrier techniques to suppress it.

[0011] Software techniques can be quite accurate but are very expensive in terms of computational time and complex in terms of algorithmic implementation. Hardware-based techniques can be simpler to implement, but require modification of the acquisition protocol and therefore an increase in exposure time. Statement of the invention

[0012] An objective of the invention is to propose a technique for tomographic inspection of industrial parts, in particular blades for aircraft engines such as 3D woven composite blades, which can both be carried out in a reduced inspection time and allow effective elimination of artifacts due to scattered radiation.

[0013] To this end, the invention proposes a method for non-destructive testing of a batch of industrial parts by X-ray tomography, comprising: - for a part of the batch taken as a calibration part: • the acquisition, with a beam blocking grid interposed between an X-ray source and the calibration part, of calibration intensity images according to different projection angles, the beam blocking grid comprising a set of X-ray attenuation elements supported by a support panel; • for each projection angle, the determination of a scattered radiation image from the calibration intensity image acquired according to this projection angle; - the replacement of the beam blocking grid with a replacement panel having an X-ray attenuation identical to that of the support panel and for each of the parts in the batch: • the acquisition, with the replacement panel interposed between the X-ray source and the part, of raw intensity images according to the different projection angles, and • the correction, for each projection angle, of the raw intensity image acquired for this projection angle by means of the scattered radiation image determined for this projection angle.

[0014] Some preferred but non-limiting aspects of this method are as follows: - the determination, for each projection angle, of the scattered radiation image comprises: • converting the calibration intensity image acquired at this projection angle into a calibration attenuation image, the calibration attenuation image comprising areas of high attenuation corresponding to the X-ray attenuation elements of the beam blocking grid; • interpolation of high attenuation areas of the calibration attenuation image; - the interpolation is carried out by using for each of the high attenuation zones an average value of pixels constituting the high attenuation zone; - the interpolation is performed with a radial basis function with a linear kernel; - it comprises the preliminary steps of acquiring, with the beam blocking grid but without a part of the batch interposed between the X-ray source and an X-ray detector, a grid intensity image and locating in the grid intensity image areas of low illumination in the shadow of the X-ray attenuating elements of the beam blocking grid, the location of areas of low illumination being used to locate the areas of strong attenuation; - the location of low-light areas includes: • converting the grid intensity image into a grid attenuation image, the grid attenuation image comprising areas of high attenuation corresponding to areas of high illumination of the grid intensity image; • localization of areas of high attenuation by segmentation of the grid intensity image; - the correction, for each projection angle, of the raw intensity image includes: • the conversion of the raw intensity image acquired according to this projection angle into a raw attenuation image; • the determination of a corrected attenuation image by subtracting the scattered radiation image determined for this projection angle from the raw attenuation image: - it further includes for each projection angle the conversion of the corrected attenuation image determined for this projection angle into a corrected intensity image; - it also includes, for each part of the batch, once the correction of the raw intensity image acquired according to each of the projection angles has been carried out, a volume reconstruction of the part. Brief description of the drawings

[0015] Other aspects, aims, advantages and characteristics of the invention will appear better on reading the following detailed description of preferred embodiments thereof, given by way of non-limiting example, and made with reference to the appended drawings in which:

[0016] - [Fig.l] is a diagram of a non-destructive testing system according to a rea possible use of the invention using a beam blocking grid in order to determine scattered radiation information;

[0017] - [Fig.2] is a diagram detailing a possible implementation of the method according to the invention;

[0018] - [Fig.3] is another diagram illustrating different stages of the process according to the invention;

[0019] - [Fig.4] illustrates a beam blocking grid that can be used in the scope of the invention;

[0020] - [Fig.5] illustrates a replacement panel that can be used in the context of the invention.

[0021] DETAILED DESCRIPTION OF PARTICULAR EMBODIMENTS

[0022] The invention relates to a method for non-destructive testing of a batch of industrial parts by X-ray tomography, in particular by cone beam volumetric tomography. By batch of industrial parts is meant here a set of parts bearing the same reference, i.e. identical parts with the exception of their possible indications that the non-destructive testing seeks to highlight.

[0023] In an exemplary embodiment, the parts are aircraft engine blades, for example blades of low density such as those made from organic matrix composite materials, i.e. composed of a woven textile preform of carbon fiber densified by the hot injection under pressure of a thermosetting type polymer resin such as for example a poly-epoxy resin.

[0024] With reference to [Fig. 1], the invention may make use of a conical geometry tomograph which, in a simplified approach, is composed of an X-ray source 1, a support 2 for the object to be imaged and an X-ray detector 3. The axis connecting the source 1 and the detector 3 constitutes the optical axis of the tomograph. This optical axis intercepts the detector 1 perpendicularly at its center. The source 1 emits a conical shaped beam, the apex of which is the focal point of the tomograph represented by the source 1 itself. The axis of the cone is the optical axis of the tomograph. The support 2 is placed between the source 1 and the detector 3, generally at the isocenter of the tomograph (axis of rotation of the tomograph). The support may be a rotating support allowing shots to be taken at different successive angles of the imaged object, here a blade C.

[0025] The method according to the invention comprises determining, using one part of the batch, a model of the scattered radiation and then applying this model to all the parts of the batch to obtain improved radiographic images free from artifacts due to the scattered radiation. The application to all the parts of the batch of the model of the scattered radiation determined for a single part of the batch is made possible because the parts of the batch have the same morphology.

[0026] Determination of the model of the scattered radiation

[0027] With reference to Figures 1, 2 and 3, this MRD determination comprises the selection of any one of the parts of the batch as calibration part C and the acquisition, with a beam blocking grid 4 interposed between the X-ray source 1 and the calibration part C, of ​​calibration intensity images ICI, ICj, ICn according to n different projection angles. The beam blocking grid comprises a set of X-ray attenuation elements EA supported by a support panel that is quasi-transparent to X-rays. The acquisition of the calibration intensity images can be carried out in simple mode (one shot per projection) or in expanded mode (two shots per projection, each of the shots capturing one half of the part C).

[0028] This MRD determination then comprises, for each projection angle, the determination of a scattered radiation image RDI, RDj, RDn from the calibration intensity image ICI, ICj, ICn acquired according to this projection angle.

[0029] This determination, for each projection angle, of the scattered radiation image RDI, RDj, RDn may comprise the conversion of the calibration intensity image ICI, ICj, ICn acquired according to this projection angle into a calibration attenuation image AC1, ACj, ACn. This conversion, which may be carried out by means of the natural logarithm, makes it possible to convert the projections from the intensity space (grey levels proportional to the intensity of the radiation received by the detector 3) to the attenuation space (grey levels proportional to the attenuation undergone by the beam during its journey between the source 1 and the detector 3). As shown in [Fig.l], the calibration attenuation image comprises areas of high attenuation FA in correspondence with the attenuation elements EA of the X-rays of the beam blocking grid 4.

[0030] As represented by the arrows I in [Fig.l], the high attenuation zones FA of the calibration attenuation image AC1, ACj, ACn are then interpolated to provide the scattered radiation image RDI, RDj, RDn, thus passing from partial information of the scattered radiation localized to the high attenuation zones to global scattered radiation information at the scale of the image. The interpolation can be carried out by exploiting for each of the high attenuation zones a scattered radiation value taken as an average value of pixels constituting the high attenuation zone, for example the average value of the gray levels of a fixed number of pixels. The interpolation can in particular be carried out between the values ​​of the directly neighboring high attenuation zones. The interpolation can for example be carried out using radial basis functions, by Launay triangulation or according to a bilinear method.The interpolation preferably exploits a radial basis function with a linear kernel.

[0031] Alternatively, the determination, for each projection angle, of the scattered radiation image RDI, RDj, RDn can be carried out in the intensity space, by interpolating low illumination zones in the scattered radiation image RDI, RDj, RDn in correspondence with the attenuation elements EA of the X-rays of the beam blocking grid 4. It is however preferable to carry out this determination in the attenuation space in which the variations in the gray levels evolve linearly with the variations in the thicknesses of the part and for which the segmentation (see below) and interpolation processes display fewer errors.

[0032] The beam blocking grid 4 (or “beam stop array” in English) blocks a portion of the primary X-rays before they reach the calibration part C and detector 3, while allowing the illumination of the majority of the calibration part C by the primary X-rays. For this, the cumulative surface area of ​​the attenuation elements EA is significantly smaller than the surface area of ​​the support panel. In expanded mode, the support panel is designed to cover two X-ray detectors jointly placed perpendicular to the optical axis of the tomograph.

[0033] The beam blocking grid 4 is placed on the optical axis between the source 1 and the calibration part C. The optical axis passes through the grid 4 perpendicularly in its center. The grid-part distance is limited by the radius of the volume of the part in revolution, for example between 150 and 300 mm for current and future generation 3D woven composite blades. The position of the grid is therefore determined by the position of the calibration part in the optical axis and by the dimensions of this calibration part.

[0034] The signal captured by the detector 3 in the non-illuminated areas corresponding to the attenuation elements EA corresponds to the secondary X-rays coming from the calibration part. As shown in [Fig.4], the attenuation elements EA can consist of several solid cylinders made of a dense material which is very attenuating for X-rays.

[0035] In a possible embodiment, the center of mass of each cylinder EA is arranged at a node of a uniform 2D grid. The spacing between the cylinders EA (or the nodes of the grid) can be uniform in both axes of the grid. The cylinders EA, arranged according to the grid, are arranged on a flat support panel 5 made of a rigid material with low attenuation for X-rays, for example a polymer. The support panel 5 constitutes the plane of the grid 4 and has a thickness adapted to accommodate the cylinders EA. Each cylinder EA is oriented towards a common focal point PF which is outside the plane of the grid 4. In use, the axis of each cylinder is aligned with the generatrix of the emission cone whose apex is the source 1. The alignment of the cylinders is therefore determined by the placement of the grid 4 in the optical axis.

[0036] The height and composition of the cylinders depend on the voltage of source 1, which itself depends on the composition and morphology of the calibration part. For blades with a density of less than 2 g / cm3, it is thus possible to use lead or tungsten cylinders with a height greater than 12 mm.

[0037] As indicated previously, the cumulative surface area of ​​the EA cylinders represents only a fraction of the total projection of the calibration part, for example less than 10% of the illuminated surface in the detector 3. By way of example, the diameter of the base of the cylinders can thus be less than 4 mm.

[0038] The number of cylinders and the extent of their arrangement on the support panel (size of the grid) allow for complete spatial sampling of the calibration part on each projection. These values ​​are related to the dimensions and morphology of the calibration part. Grid 4 covers all projections of the calibration part on detector 3 (field of view) as uniformly as possible. Thus, the grid nodes cover and exceed the entire field of view in detector 3.

[0039] The projection of the beam blocking grid 4 gives rise to two types of areas in the projection images: shadow areas which correspond to the projection of the attenuation elements and illuminated areas which correspond to the projection of the grid outside the attenuation elements. When the attenuation elements are arranged to form a uniform 2D rectangular mesh, the shadow areas form a uniform 2D rectangular mesh on the projections captured by the detector X (i.e. the calibration intensity images).

[0040] In a possible embodiment illustrated in [Fig. 3], the method comprises the preliminary steps of acquiring, with the beam blocking grid 4 but without a part of the batch interposed between the source 1 and the detector 3, a grid intensity image IG and of locating in the grid intensity image areas of low illumination in the shadow of the X-ray attenuation elements of the beam blocking grid (i.e. the shadow areas). The location of areas of low illumination provides a reference mask MR (for example a uniform 2D rectangular mesh) which is then used to locate the areas of high attenuation in the calibration attenuation image AC1, ACj, ACn before interpolating them.

[0041] Locating the low illumination areas may include converting the grid intensity image IG into a grid attenuation image AG, the grid attenuation image AG including high attenuation areas corresponding to the low illumination areas of the grid intensity image IG. Locating the high attenuation areas to provide the MR marker mask may be accomplished by segmenting the grid intensity image AG.

[0042] Tomographic reconstruction of the batch parts, without scattering artifacts

[0043] With reference to figures 2, 3 and 5, this CRD reconstruction comprises the substitution of the beam blocking grid 4 by a replacement panel 6 having an X-ray attenuation identical to that of the support panel 5 of the beam blocking grid 4 and, for each of the parts of the batch, the acquisition, with the replacement panel 6 interposed between the source 1 and the part A, of raw intensity images IB1, IBj, IBn according to the different projection angles. This reconstruction continues with the correction, for each projection angle, of the raw intensity image IB1, IBj, IBn acquired for this projection angle by means of the scattered radiation image RDI, RDj, RDn determined for this projection angle.

[0044] The support panel 5 of the beam blocking grid 4 is likely to slightly modify the rendering of the gray levels on the detector 3. This is particularly the case when the energy of the X-rays emitted by the source 1 is relatively low (typically less than 250 keV), for example when imaging parts with a low density (typically less than 3 g / cm3) such as composite blades. The replacement panel 6 is then used to reproduce the attenuation of the support panel 5 and to make the acquisitions carried out during the MRD determination and the CRD reconstruction homogeneous for the purpose of their algorithmic processing. The replacement panel 6 is homogeneous and uniform, it has the same dimensions as the beam blocking grid 4 and is made of the same material as the support panel 5.

[0045] In a possible embodiment, the correction, for each projection angle, of the raw intensity image IB1, IBj, IBn comprises the conversion, for example via the natural logarithm, of the raw intensity image acquired according to this projection angle into a raw attenuation image AB1, ABj, ABn and the determination of a corrected attenuation image AB le, ABjc, AB ne by subtracting the scattered radiation image determined for this projection angle RDI, RDj, RDn from the raw attenuation image AB1, ABj, ABn.

[0046] According to the previously mentioned alternative, this correction can also be carried out directly in the intensity space.

[0047] The subtraction may comprise the application of a regularization operation to the raw attenuation image AB1, ABj, ABn (alternatively the raw intensity image IB1, IBj, IBn) aimed at reducing the inconsistencies (low gray level values ​​close to the values ​​of the scattered radiation image) due to the lack of penetration of X-rays into the part, to the thick zones, to the projection angle. These inconsistencies are only present on a few projections, and the regularization parameters are therefore adapted projection by projection.

[0048] The method can continue with, for each projection angle, the conversion of the corrected attenuation image ABlc, ABjc, ABnc determined for this projection angle into a corrected intensity image IB le, IBjc, IBnc. This conversion, which makes it possible to move from the attenuation space to the intensity space, can be carried out via the application of an exponential on the values ​​of the corrected attenuation images ABlc, ABjc, ABnc.

[0049] The method may further comprise, for each part of the batch, once the correction of the raw intensity image acquired according to each of the projection angles has been carried out, a volume reconstruction of the part using the different corrected intensity images. The volume image V resulting from this reconstruction is free of artifacts due to scattered radiation. This volume image then the subject of an analysis to detect possible indications in the imaged part (anomaly detection or characterization of the material).

[0050] The method according to the invention requires little additional time to eliminate the scattered artifacts. Once the scattered radiation images have been determined once and for all with the calibration piece, this additional time is limited to the subtraction of these scattered radiation images from the different acquired projections.

[0051] The beam blocking grid is also very handy while its construction and use can be carried out at low cost.

[0052] Finally, this solution proves to be compatible with any type of tomograph, in particular those designed to image large parts. In one possible embodiment, the tomograph used (low magnification for imaging large parts, low energies, imaging of a part in two parts) generates projections in a tiff format containing information on the geometry specially adapted to the volume reconstruction algorithm. This information may be lost during the generation of the corrected projection (corrected raw intensity image). Also, the method according to the invention may comprise the recovery of the header of the original file generated by the tomograph (associated with a raw intensity image) and the association of this header with the file corresponding to the corrected projection.

[0053] The invention is not limited to the method as previously described, but also extends to a system for non-destructive testing of a batch of industrial parts by X-ray tomography, comprising a processor configured to implement the steps of the method as previously described, in particular the steps of determining a scattered radiation image RDI, RDj, RDn for each projection angle and correcting, for each projection angle, the raw intensity image IB1, IBj, IBn acquired for this projection angle by means of the scattered radiation image RDI, RDj, RDn determined for this projection angle.The invention further extends to a computer program product comprising instructions which, when the program is executed by a computer, cause the latter to implement the steps of the method as previously described, in particular the steps of determining a scattered radiation image RDI, RDj, RDn for each projection angle and of correcting, for each projection angle, the raw intensity image IB1, IBj, IBn acquired for this projection angle by means of the scattered radiation image RDI, RDj, RDn determined for this projection angle.

Claims

Claims

1. Method for non-destructive testing of a batch of industrial parts by X-ray tomography, comprising: - for a part of the batch taken as a calibration part (C): • the acquisition, with a beam blocking grid (4) interposed between an X-ray source (1) and the calibration part (C), of calibration intensity images (ICI, ICj, ICn) according to different projection angles, the beam blocking grid comprising a set of X-ray attenuation elements (EA) supported by a support panel (5); • for each projection angle, the determination of a scattered radiation image (RDI, RDj, RDn) from the calibration intensity image (ICI, ICj, ICn) acquired according to this projection angle; - the substitution of the beam blocking grid (4) by a replacement panel (6) having an X-ray attenuation identical to that of the support panel (5) and for each of the parts of batch (A): • the acquisition, with the replacement panel interposed between the X-ray source and the part (A), of raw intensity images (IB1, IBj, IBn) according to the different projection angles, and • the correction, for each projection angle, of the raw intensity image (IB1, IBj, IBn) acquired for this projection angle by means of the scattered radiation image (RDI, RDj, RDn) determined for this projection angle.

2. Method according to claim 1, in which the determination, for each projection angle, of the scattered radiation image (RDI, RDj, RDn) comprises: - the conversion of the calibration intensity image (ICI, ICj, ICn) acquired according to this projection angle into a calibration attenuation image (AC1, ACj, ACn), the image calibration attenuation comprising high attenuation zones (FA) in correspondence with the attenuation elements (EA) of the X-rays of the beam blocking grid (4); - the interpolation (I) of the high attenuation zones of the calibration attenuation image.

3. Method according to claim 2, in which the interpolation is carried out by exploiting for each of the high attenuation zones an average value of pixels constituting the high attenuation zone.

4. Method according to one of claims 2 and 3, in which the interpolation is carried out with a radial basis function with a linear kernel.

5. Method according to one of claims 2 to 4, comprising the prior steps of acquiring, with the beam blocking grid (4) but without a part of the batch interposed between the X-ray source (1) and an X-ray detector (3), a grid intensity image (IG) and locating in the grid intensity image areas of low illumination in the shadow of the X-ray attenuation elements (EA) of the beam blocking grid, the location of areas of low illumination being used to locate areas of high attenuation.

6. The method of claim 5, wherein locating the low illumination areas comprises: - converting the grid intensity image (IG) into a grid attenuation image (AG), the grid attenuation image comprising high attenuation areas corresponding to the high illumination areas of the grid intensity image - locating the high attenuation areas by segmenting the grid intensity image.

7. Method according to one of claims 1 to 6, in which the correction, for each projection angle, of the raw intensity image (IB1, IBj, IBn) comprises: - the conversion of the raw intensity image acquired according to this projection angle into a raw attenuation image (AB1, ABj, ABn); - the determination of a corrected attenuation image (ABle, ABjc, AB ne) by subtracting the scattered radiation image (RDI, RDj, RDn) determined for this projection angle from the raw attenuation image (AB1, ABj, ABn).

8. Method according to claim 7, further comprising for each projection angle the conversion of the corrected attenuation image (AB le, ABjc, ABnc) determined for this projection angle into a corrected intensity image (IB le, IBjc, IBnc).

9. Method according to one of claims 1 to 8, further comprising, for each part of the batch, once the correction of the raw intensity image acquired according to each of the projection angles has been carried out, a volume reconstruction of the part.

10. System for non-destructive testing of a batch of industrial parts by X-ray tomography, comprising a processor configured to implement the steps of the method according to one of claims 1 to 9 of determining a scattered radiation image (RDI, RDj, RDn) for each projection angle and correcting, for each projection angle, the raw intensity image (IB1, IBj, IBn) acquired for this projection angle by means of the scattered radiation image (RDI, RDj, RDn) determined for this projection angle.

11. Computer program product comprising instructions which, when the program is executed by a computer, cause the latter to implement the steps of the method according to one of claims 1 to 9 of determining a scattered radiation image (RDI, RDj, RDn) for each projection angle and of correcting, for each projection angle, the raw intensity image (IB1, IBj, IBn) acquired for this projection angle by means of the scattered radiation image (RDI, RDj, RDn) determined for this projection angle.

Images