Method and device for additive manufacturing of workpieces

Abstract

In a method for additive manufacturing of a workpiece (14), a data set (42) is obtained, the data set defining the workpiece (14) with a plurality of workpiece layers (16, 20) arranged one above the other. The plurality of workpiece layers (16, 20) arranged one above the other are produced, and the plurality of workpiece layers (16, 20) form a layer stack (18), which has an uppermost workpiece layer (20) and a plurality of workpiece layers (16) located therebelow at a defined point in time. The layer stack (18) is thermally actuated (46) at a defined point in time, and a plurality of images of the uppermost workpiece layer (20) are captured. The layer stack (18) is inspected using the plurality of images to obtain an inspection result representative of the workpiece. An individual deformation time curve of the uppermost workpiece layer (20) is determined in response to the thermal actuation (46). The individual deformation time curve has a plurality of characteristic features, the plurality of characteristic features comprising an individual deformation increase, an individual deformation maximum, and an individual deformation decrease. The inspection result is determined using at least one of the characteristic features from the plurality of characteristic features.

Description

[0001] The invention relates to a method for additive manufacturing of a workpiece, the method comprising the following steps:

[0002] a) obtaining a data set, the data set defining a workpiece with a plurality of workpiece layers arranged one above the other,

[0003] b) producing a plurality of workpiece layers arranged one above the other using a layer-forming tool controlled in accordance with the data set, wherein the plurality of workpiece layers arranged one above the other form a layer stack which, at a defined point in time, has an uppermost workpiece layer and a plurality of underlying workpiece layers,

[0004] c) thermally actuating the layer stack at a defined point in time,

[0005] d) taking multiple images of the uppermost workpiece layer, and

[0006] e) Inspecting the layer stack using multiple images to obtain inspection results representative of the workpiece.

[0007] The present invention further relates to a device for additive manufacturing of a workpiece, the device comprising: a memory for obtaining a data set, the data set defining the workpiece with multiple workpiece layers arranged above each other; a manufacturing platform; a layer forming tool; a heating tool; a camera, the camera pointing to the manufacturing platform; and an evaluation and control unit, the evaluation and control unit being configured to produce multiple workpiece layers arranged above each other on the manufacturing platform using the layer forming tool and the data set, wherein the multiple workpiece layers arranged above each other form a layer stack having an uppermost workpiece layer and multiple workpiece layers below at a defined point in time, the evaluation and control unit being further configured to use the heating tool to thermally excite the layer stack at a defined point in time, and to use the camera to record multiple images of the uppermost workpiece layer, and finally to use the multiple images to inspect the layer stack to obtain an inspection result representative of the workpiece.

[0008] Such a method and such a device are basically known from DE 10 2014 212 246 B3.

[0009] Additive methods for manufacturing workpieces are sometimes referred to as 3D printing. There are various additive manufacturing methods. In selective laser sintering (SLS) or selective laser melting (SLM), a powder bed called a powder bed made of granular material is used. The granular material is often a metal material. However, there are also methods using granular plastic materials, especially polymers. With the help of a laser beam or an electron beam, selected powder particles on the upper side of the powder bed are locally selectively melted or at least partially melted, and in this way adhere to each other when cooling. Then a new powder layer is spread on the workpiece structure and the remaining unmelted powder, and another workpiece layer is produced with the help of a laser beam or an electron beam. Thus, the workpiece is produced layer by layer in successive steps. Typically, the individual workpiece layers are produced from bottom to top on a manufacturing platform, and after each workpiece layer, the manufacturing platform is lowered to the layer height of the next layer.

[0010] Additive manufacturing of workpieces makes it possible to produce individual workpieces with a high degree of complexity and low material costs. At the same time, however, there are major challenges with regard to the workpiece quality, since each individual material layer may present anomalies, which may lead to defects in the workpiece. Anomalies may lead to defects such as holes, micropores or pores in the layer structure, local layer peeling / delamination, cracks in and / or on the surface, dents, shape deviations and / or material stresses. For this reason, there are many proposals for detecting defects in additively manufactured workpieces, if possible even during the manufacture of the layer sequence. For example, US 2015 / 0061170 A1 discloses an optical measuring sensor with a camera, which can be arranged to enable 3D coordinate measurement on the respective uppermost material layer.

[0011] DE 10 2016 115 241 A1 discloses an additive manufacturing process involving selectively heating a powder layer to form a solid workpiece layer. The produced workpiece layer is excited with ultrasonic energy waves using a wave generating laser. The propagated ultrasonic energy waves are detected and analyzed to determine physical properties of the workpiece layer. Further workpiece layers are produced in response to the information obtained. US 7,278,315 discloses another laser-ultrasonic method for detecting defects in additively manufactured workpieces.

[0012] DE 10 2016 110 266 A1 discloses a method and an apparatus for the additive manufacturing of workpieces, wherein laser ultrasonic measurement, absolute measuring interferometry or laser pulse thermography are proposed for checking the workpiece layers. In the latter case, the thermal radiation emitted from the workpiece surface can be analyzed spectrally. As a checking method, DE 10 2016 110 266 A1 also mentions measuring the geometry and temperature of what is called a melt pool.

[0013] DE 10 2014 212 246 B3 mentioned in the opening paragraph discloses thermal stimulation of an additively manufactured workpiece during the manufacturing process in order to detect defects in the workpiece layers at an early stage by capturing the thermal radiation from the uppermost workpiece layer by thermography and analyzing it.

[0014] DE 10 2016 201 289 A1 discloses another method for additive manufacturing of workpieces, wherein first measurement data are collected during the additive building process using a thermographic material test or using an eddy current material test. After the additive building process, second measurement data are collected by means of a computer tomography scan and compared with the first measurement data. The results of the material test are classified using an algorithm from the field of supervised machine learning (not described in more detail).

[0015] DE 10 2017 124 100 A1 discloses a method and a device for additive manufacturing of workpieces, wherein during manufacturing, a test is performed by means of laser ultrasound. For analysis, the results of the test are compared with the results of a simulation of the test.

[0016] The publication by Everton et al., “Review of in-situ process monitoring and in-situ metrology for metal additive manufacturing,” Materials and Design 95 (2016), pp. 431-445, outlines different methods for inspecting additively manufactured workpieces using camera images and pyrometry.

[0017] US 8,449,176 B2 describes a method for automatically processing thermographic data which are recorded after thermal stimulation of a measurement object. The variance of the data is compared with the variance of corresponding data associated with a reference object.

[0018] DE 10 2017 108 874 A1 and US 2020 / 158499 A1 with the same priority disclose an optical system for overall material testing by means of illumination from a plurality of different directions. In some variants, the system can be used to determine a height map of a material layer to be tested.

[0019] The surfaces of additive manufacturing and therefore also the surface of each individual workpiece layer are typically very rough (in the range of a few μm rms) and produce strong reflections, at least when using metallic material particles. In addition, topographic reliefs are often produced, such as writing marks due to the laser process (herringbone patterns) or patterns due to scanning strategies (e.g. checkerboard patterns). In addition, defects (e.g. pilling or particle deposits) may appear on the surface during this process. Overall, these changes (which may be in the range of several hundred microns) make the inspection of the workpiece layer, in particular the detection of defects below the surface, very difficult. Due to the rough surface and other surface changes, local temperature changes and deformations that are not caused by anomalies and defects in the workpiece may occur during thermal excitation. The temperature changes and deformations on the rough surface are superimposed on the signal from the underlying workpiece layer. Therefore, a method of the type mentioned in the opening section is desired, in which the signal caused by the surface topography can be more reliably distinguished from the abnormal signal from the workpiece layer.

[0020] Against this background, the object of the present invention is to specify an alternative method and a corresponding device for the additive production of workpieces with high quality. In particular, the object is to monitor the quality of the material layer close to the process in an efficient manner so that emerging or impending layer defects can be corrected at an early stage. Here, the aim is to distinguish anomalies in the workpiece layer as reliably as possible from effects that may be caused by a rough but defect-free surface.

[0021] According to a first aspect of the invention, the object is achieved by a method of the type mentioned in the opening paragraph, wherein a single time deformation curve of the uppermost workpiece layer in response to a thermal stimulus is determined using a plurality of images, wherein the single time deformation curve has a plurality of characteristic features, the plurality of characteristic features comprising a single deformation increase, a single deformation maximum and a single deformation decrease, and wherein an inspection result is determined using at least one of the characteristic features mentioned from the plurality of characteristic features. Preferably, the inspection result is determined using at least two of the mentioned characteristic features. It is also preferred if at least one of the mentioned characteristic features comprises a single deformation maximum.

[0022] According to a further aspect, the object is achieved by a device of the type mentioned in the opening paragraph, wherein the evaluation and control unit is further arranged to determine an individual time deformation curve of the uppermost workpiece layer in response to the thermal excitation using the plurality of images, wherein the individual time deformation curve has a plurality of characteristic features, the plurality of characteristic features comprising an individual deformation increase, an individual deformation maximum and an individual deformation decrease, and wherein the evaluation and control unit determines the inspection result using at least one of the characteristic features mentioned from the plurality of characteristic features.

[0023] The novel method and the novel device utilize thermal excitation of a layer stack in order to detect defects as early as possible in the manufacturing process. However, in contrast to the method and the device mentioned in the opening section, images are now used to determine and analyze the deformation curve of the topmost workpiece layer. Therefore, the novel method and the novel device primarily consider mechanical (dimensional and / or geometric) changes on the workpiece surface in response to thermal excitation. In some exemplary embodiments, a thermal analysis (such as temperature distribution) can be performed on the layer stack in response to thermal excitation as an addition and supplement to the analysis of the deformation curve, and the inspection results of the thermal analysis can be advantageously combined with the inspection results from the deformation curve. Nevertheless, the defect detection in the novel method and the novel device is based on the analysis of the mechanical deformation of the layer stack in response to thermal excitation (which manifests itself in the topmost workpiece layer).

[0024] The novel method and the novel device also take into account the temporal behavior of the layer stack in response to thermal excitation. Therefore, it is not only whether or to what extent the deformation of the layer stack becomes visible on the topmost workpiece layer that is to be analyzed. Instead, the time curve of the deformation is analyzed within a defined time interval at the beginning of and / or after the thermal excitation. Therefore, a plurality of images are advantageously recorded in a time series in a time-interleaved manner. In other words, a plurality of images advantageously include an image stack, and the images are contained in the image stack in chronological order. As described below with reference to a preferred exemplary embodiment, anomalies in the layer stack (including anomalies below the surface of the topmost workpiece layer) can therefore be reliably detected, even if the layer surface exhibits roughness and / or has a written structure and / or causes light reflection.

[0025] As explained in more detail below, the time deformation curve has a number of features that can characterize anomalies in the layer stack. These features are, in particular: a single deformation increase, which can appear as a rising edge in the deformation curve; a single deformation maximum in the deformation curve; and a single decrease in deformation, which can be seen as a falling edge in the deformation curve. In some exemplary embodiments, turning points can be sought in the individual deformation curves and analyzed. The novel method and the novel device advantageously use at least two of the three mentioned features when analyzing the time deformation curve and detecting anomalies, since it has been pointed out that in this way interference signals from rough but defect-free surfaces can be more reliably distinguished. The inspection result is particularly advantageously determined using a single deformation maximum in the deformation curve, known as overshoot, since this feature particularly characterizes voids below the surface of the uppermost workpiece layer, and such voids can be serious defects. Furthermore, in some exemplary embodiments, it is preferred that all three mentioned features are used in combination with each other to determine the inspection result.

[0026] As already indicated, the novel method and the novel device enable reliable detection of anomalies in the layer sequence of an additively manufactured workpiece during the manufacturing process. The above-mentioned objects are therefore fully achieved.

[0027] In a preferred configuration of the invention, each image from the plurality of images has a plurality of image segments and a respective individual time deformation curve in response to the thermal stimulus is determined for each image segment from the plurality of image segments, wherein the examination result is determined using the respective individual time deformation curve.

[0028] Each image block can advantageously be an independent image pixel, so that a plurality of individual time deformation curves are determined pixel by pixel using the image. However, it is also possible to determine and analyze individual time deformation curves using pixel groups or image blocks comprising a large number of image pixels. Determining a plurality of individual time deformation curves using a plurality of time-interleaved images allows the lateral extent and / or lateral distribution of detected anomalies to be determined. Thus, with this configuration a quantitative estimate of the size of any defect can be obtained. This makes it easier to make decisions about the quality characteristics of the workpiece and its suitability for the intended purpose. The term "image" can, on the one hand, refer to an intensity image such as one recorded with a camera, in particular an image having a plurality of image pixels, each image pixel representing an intensity value in terms of brightness, such as a grayscale value image. On the other hand, the term "image" can advantageously refer here to amplitude and phase composite images, which can be determined from the intensity image with the aid of image processing.

[0029] In another configuration, thermally actuating the layer stack includes thermally actuating an uppermost workpiece layer in a locally defined first layer region, and the plurality of images show the first layer region and another layer region of the uppermost workpiece layer, wherein the first layer region and the another layer region are disjoint.

[0030] In this configuration, the image shows a thermally excited region of the topmost workpiece layer and another region of the topmost workpiece layer, which is not excited or at least not directly excited and is therefore different from the first region. For example, the thermal excitation can be performed with a laser beam with a diameter of several millimeters. In general, the thermal excitation can be performed on the topmost workpiece layer in a local area with a diameter of up to 20 mm, preferably in an area of ​​up to 10 mm. Since the image shows another layer area that is not excited or not directly excited, this configuration makes it possible to determine one or more individual deformation curves that do not show or at least do not directly show a reaction to the thermal excitation. In the novel method and the novel device, this additional deformation curve can be advantageously determined and used as a reference deformation curve. For example, the reference deformation curve makes it possible to detect any time-varying deformations in the layer stack that do not come from or do not come directly from the thermal excitation, but are the result of vibrations, shocks or another undesirable influence. The novel method and the novel device advantageously use this reference deformation curve in order to normalize one or more time deformation curves and thus eliminate the influence of vibrations. For example, one or more reference deformation curves can be subtracted from one or more time deformation curves in order to maximize the influence of thermal excitation. In addition, this configuration allows for simpler localization of defects in the image. Thus, this configuration contributes to more reliable defect detection.

[0031] In a further configuration, the plurality of images comprises at least one first image of the uppermost workpiece layer recorded before the thermal excitation, wherein the inspection result is determined using the at least one first image.

[0032] In this configuration, the image stack of images recorded in the time series comprises at least one image showing the uppermost workpiece layer without thermal excitation. Such an image can be recorded at the beginning of each new time series and thus before each new excitation. However, this image can also be provided in the memory as a reference image for a plurality of image stacks. This configuration provides a global reference image without thermal excitation, which facilitates the localization of detected anomalies. This configuration can also advantageously be used for vibration compensation. Therefore, this configuration also advantageously contributes to a more reliable defect detection.

[0033] In another configuration, the plurality of images are normalized using at least one reference image before determining the individual time deformation curves.

[0034] This configuration also advantageously contributes to viewing the reaction of the layer stack to thermal excitation with as little interference as possible, since image information that cannot be attributed to thermal excitation can be minimized. For example, by subtracting at least one reference image from the time-interleaved images of the image stack, a corresponding normalized time image can be generated in which the reaction of the layer stack to thermal excitation is more strongly represented. Thus, this configuration advantageously contributes to achieving a higher level of detection reliability.

[0035] In another embodiment, the plurality of images are recorded with a frame rate ≥ 1 kHz.

[0036] Such a frame rate has proven to be advantageous for analyzing the above-mentioned characteristic features in at least one deformation curve with a high level of reliability. Such a frame rate advantageously enables a transient analysis with a time resolution so high that even smaller defects, including porous locations in the layer stack, can be reliably detected.

[0037] In another configuration, the inspection results are determined based on the plurality of images using principal component analysis.

[0038] Principal component analysis (PCA) is a statistical mathematical method known per se. It is advantageously suitable for structuring and simplifying a wide range of data sets by approximating multiple statistical variables using a smaller number of linear combinations that include as much information as possible (the so-called principal components). Principal component analysis can analyze many deformation curves in a very advantageous and efficient manner, and is therefore particularly suitable when analyzing individual time deformation curves in many image blocks or even at the pixel level. In some exemplary embodiments, the characteristic features of each deformation curve (or alternatively, polynomials or rational functions with up to 6 degrees of freedom) can be advantageously used to model the time variation of each deformation curve in logarithmic form. The coefficient images generated in this way can be converted into a smaller number of more compressed PCA coefficient images by means of principal component analysis. Then, a clustering algorithm can be advantageously applied to these compressed PCA coefficient images for the purpose of segmentation. Combined with threshold judgment, the probability of abnormality can then be determined in the corresponding segmented image area in an efficient manner. Therefore, principal component analysis can determine the inspection results very efficiently.

[0039] In another configuration, the layer stack is repeatedly thermally excited and the inspection result is determined based on the repeated excitations. In some preferred variants of this configuration, at least one excitation parameter, in particular the excitation intensity and / or the excitation duration, is varied and the inspection result is determined based on at least one excitation parameter.

[0040] In this configuration, a plurality of measurement series are performed on the uppermost workpiece layer by repeating steps c) and d), which have already been mentioned in the opening section. The intensity of the excitation (e.g. the energy of the laser pulse or the duration of the excitation) can be varied from one measurement series to another. Furthermore, the position or the excitation pattern on the uppermost workpiece layer can also be varied from one measurement series to another. In this way, a plurality of individual deformation curves can be obtained, which can be correlated with one another directly or, for example, after a principal component analysis, in order to more reliably detect anomalies below the surface of the uppermost workpiece layer. However, even without changing the excitation parameters, repeated measurements can be advantageous for increasing the reliability of the measurement, for example by averaging.

[0041] In another configuration, the inspection result is determined using a plurality of images of a plurality of respective uppermost workpiece layers.

[0042] In this configuration, a plurality of measurement series are performed on a plurality of respective uppermost workpiece layers by performing steps c) and d) mentioned in the opening part on workpiece layers arranged one above the other (and typically different from one another), wherein the respective inspection result is determined from the plurality of measurement series. A weighted time average can be performed particularly advantageously on the deformation curves of the respective uppermost workpiece layers that are temporally successive. Due to the different roughness and topography of the different successive layers, the random influence of the individual layers can be reduced by this averaging. On the other hand, anomalies below the respective uppermost workpiece layer will affect each measurement series and can therefore be detected more reliably.

[0043] In another configuration, the individual time deformation curves are determined using at least one of the following measurement methods: speckle interferometry, digital holography, shear speckle; laser Doppler vibrometry, Fabry-Perot interferometry, Sagnac interferometry, interferometry using nonlinear optics.

[0044] Surface measurements with interferometric accuracy can be performed in speckle interferometers using coherent light, for example with electronic speckle pattern interferometry (ESPI). ESPI is particularly advantageous for technical surfaces with a roughness in the range of a few μm rms. Furthermore, ESPI is able to measure deformations orthogonal to the surface (z-direction, “out-of-plane”) but also in the plane of the surface (x / y-direction, “in-plane”). For fast area-based measurements in a process-adapted manner in the kHz range, it is advantageous to use an ESPI system with a spatial (rather than temporal) phase shift. If, for application reasons, the measured variables exceed the unambiguous range of the interferometer (due to heating parameters, frame rate, combination of materials), phase unwrapping algorithms can be applied to the measured values ​​or two or more wavelengths can be used in the interferometer to increase the unambiguous range. The use of two different wavelengths or angles (observation or illumination direction) in speckle interferometry allows the measurement of surface shape / topography.

[0045] In the shearing speckle method, a shearing element (e.g. a wedge plate or a tilting mirror) is used in the beam path, as a result of which the surface to be measured is imaged directly on the camera sensor and simultaneously laterally offset. The variable measured here is the deformation gradient in the direction of the lateral image offset. As a result, the sensitivity is essentially given along the preferred lateral direction, which is why it is advantageous to perform further measurements in another lateral direction (preferably orthogonal to the first lateral direction) in order to capture the overall deformation.

[0046] The advantage of the speckle interferometry and sheared speckle method is that the measurement results have a high lateral and axial resolution. Therefore, these methods are advantageous when the aim is to inspect workpieces of large areas. In contrast, it is advantageous to combine laser Doppler vibrometery, Fabry-Perot interferometry, Sagnac interferometry or interferometry with nonlinear optics with scanning of the workpiece surface in the lateral direction (i.e., in order to scan the workpiece surface). The methods mentioned have a high axial resolution and are therefore able to reliably detect anomalies that are in the depth of the layer stack.

[0047] Vibrometers are often used for vibration analysis using the optical Doppler effect to measure the velocity and / or displacement of a surface. Scanning systems (3D scanning vibrometers) or multi-point vibrometers are suitable for spatially resolved measurements. Multi-point vibrometers are advantageous for fast, process-adaptive area-based measurements in the kHz range in order to obtain a time-synchronized record of all measuring points on the surface.

[0048] All of the above methods enable the use of recorded images to determine the thermally stimulated deformation curve in great detail.

[0049] In a further configuration, the inspection result is determined also using thermal transient curves and / or using ultrasonic excitation and / or using simulated deformation curves and / or using melt pool properties and / or using angle selective illumination of the uppermost workpiece layer.

[0050] In this configuration, the inspection of deformation transients based on thermal stimulation is combined with other inspection methods, which, taken individually, have already been proposed in the prior art mentioned in the opening part. Since the information density is increased again, this configuration enables defects below the surface of the uppermost workpiece layer to be detected more reliably. The inspection methods known per se become even more efficient in combination with novel methods and novel devices. The combination of inspection based on deformation transients and inspection of a powder bed with embedded workpiece layers using angle-selective illumination is particularly advantageous, as described in DE 10 2017 108 874 A1 mentioned in the opening part and in US 2020 / 158499 A1 having the same priority.

[0051] It goes without saying that the features mentioned above and also those yet to be explained below can be used not only in the combination specified in each case but also in other combinations or alone, without departing from the scope of the present invention.

[0052] Several exemplary embodiments of the invention are shown in the drawings and are explained in more detail in the following description. In the drawings:

[0053] Figure 1 A schematic diagram showing an exemplary embodiment of the novel device is shown,

[0054] Figure 2 Several exemplary deformation curves are shown to illustrate exemplary embodiments of the novel method.

[0055] Figure 3 An exemplary image of the uppermost workpiece layer illustrating an exemplary embodiment of the novel method is shown,

[0056] Figure 4 Another exemplary deformation curve illustrating an exemplary embodiment of the novel method is shown.

[0057] Figure 5 A flow chart illustrating an exemplary embodiment of the novel method is shown, and

[0058] Figure 6 A flow chart is shown explaining the inspection of the uppermost workpiece layer in an exemplary embodiment of the novel method.

[0059] exist Figure 1In the drawings, an exemplary embodiment of the novel device is designated as a whole by reference numeral 10. The device 10 has a production platform 12, on which a workpiece 14 is additively produced according to an exemplary embodiment of the novel method. The workpiece 14 is built up layer by layer from bottom to top in steps that are sequential in time, that is, workpiece layers 16 are arranged one above the other. The workpiece layers 16 form a layer stack 18, each of which has a respective uppermost workpiece layer 20.

[0060] In the exemplary embodiment shown here, the workpiece layers 16, 20 are each produced in a so-called powder bed from granular material 22, in particular metallic material and / or plastic material. The granular material 22 is taken from a reservoir 24 and spread on the existing layer stack 18 by means of a scraper 26 that can be moved in the direction of arrow 28. For this purpose, the production platform 12 is typically lowered in the direction of arrow 30 by the height of the next workpiece layer and / or the reservoir 24 is raised relative to the production platform 12.

[0061] Reference numeral 32 denotes a tool with which the granular material 22 can be selectively solidified on the layer stack 18. In some exemplary embodiments, the tool 32 includes a laser beam 34 and moves it relative to the manufacturing platform 12 along a trajectory 36 to produce the workpiece layer 18 from the granular material 22. The material particles can be selectively melted and / or partially melted with the laser beam 34 so that they adhere to each other and solidify when cooled. This manufacturing principle is called selective laser melting (SLM) or selective laser sintering (SLS).

[0062] In other exemplary embodiments, the layer forming tool 32 can generate an electron beam to produce a workpiece layer on the manufacturing platform 12. In addition, the device 10 can include more than one layer forming tool 32, that is, the device can use, for example, two or more laser beams and / or electron beams to produce a workpiece layer. However, the novel method and the novel device are not limited to this manufacturing principle. Alternatively or additionally, the workpiece layer can be produced using other additive methods, for example using so-called stereolithography or by selectively supplying and / or depositing materials.

[0063] For simplicity, the layer forming tool 32 (hereinafter referred to as writing laser) is connected to an evaluation and control unit (hereinafter referred to as controller 38 for short), which controls the movement of the laser beam 34. The controller 38 has an interface 40, via which a data set 42 can be read, which defines the workpiece 14 to be produced in a plurality of layers arranged one above the other. The controller 38 controls the movement of the laser beam 34 relative to the layer stack 18 according to the data set 42, wherein the laser beam 30 describes a trajectory 36 in each workpiece layer 16, 18 to be produced, which trajectory is generated in each case by the data set 42. In some exemplary embodiments, the controller 38 is implemented by means of one or more commercially available personal computers, which run an operating system (such as Microsoft Windows, MacOS or Linux) and one or more control programs (whereby exemplary embodiments of the novel method are implemented). In some exemplary embodiments, the controller 38 can be implemented as a soft PLC on a commercially available PC. Alternatively or additionally, the controller 38 can be implemented by means of dedicated control hardware with one or more ASICs, FPGAs, microcontrollers, microprocessors or comparable logic circuits.

[0064] The apparatus 10 further has a measuring device which is arranged to inspect the workpiece layers 16, 20. In some advantageous exemplary embodiments, the measuring device is also arranged to inspect the respective uppermost material layer made of granular material 22 on the layer stack 18 before the granular material 22 is selectively solidified to form a new workpiece layer.

[0065] Here, the measuring device comprises a camera 44 and a heating tool 46, each of which is connected to the controller 38 (or to a separate controller for the measuring device, not shown here). The camera 44 is arranged to record a plurality of images of the respective uppermost workpiece layer 18 of the layer stack 18. The heating tool 46 is arranged to thermally stimulate the layer stack 18 at defined points in time. In some exemplary embodiments, the heating tool 46 comprises a further laser beam 48, which irradiates the respective uppermost material layer 18 and locally heats the layer stack 18. Alternatively or additionally, the heating tool 46 may comprise an electron beam and / or inductively thermally stimulate the layer stack 18 with energy pulses.

[0066] The thermal excitation increases the temperature on the surface of the layer stack 18. Due to the temperature gradient, heat diffuses from the surface into the volume of the layer stack. In the process, the material expands. This expansion leads to local deformations in the layer stack and on its surface, the spatial and temporal curves of which are captured with a measuring device. The measuring device can advantageously capture the deformations by means of a camera 44 and interferometry. Accordingly, the camera 44 can be a component of an interferometric measurement system, in particular a speckle interferometer. Alternatively or additionally, the measuring device can implement shear speckle measurement, laser Doppler vibrometer, Fabry-Perot interferometry, Sagnac interferometry, and / or interferometry using nonlinear optics.

[0067] The deformation in response to thermal excitation depends on the material properties and also on the individual layer structure. The surface roughness and the trajectory 36 of the writing beam 34 can influence the individual layer structure. The measuring device with a camera 44 and a heating tool 46 is arranged to capture local deformations in the layer stack in response to thermal excitation with both temporal and spatial resolution. Advantageously, the evaluation and control unit 38 is further arranged to analyze transients in the captured deformations. If the local variation behavior of a dimensionally measurable variable is determined on the surface of the layer stack, conclusions can be drawn about the material properties and in particular defects (anomalies) in the layer stack can be determined. Examples of such defects are blow holes, pores, unmelted particles, delamination, etc. For example, as the porosity changes, the heat conduction changes. In the case of individual defects (such as blow holes with a range of several hundred micrometers in all three dimensions), the heat accumulation together with the mechanical properties gives rise to, for example, characteristic time deformation curves, as described below with reference to Figure 2 and Figure 3 Further explained in more detail.

[0068] In some exemplary embodiments, the measuring device may include an illumination device having a plurality of illumination modules 50a, 50b arranged at different positions relative to the manufacturing platform 12 so as to illuminate the surface of the layer stack from a plurality of different directions. In combination with the camera 44, the illumination device can advantageously be used to perform an additional inspection of the surface of the layer stack using the method described in DE 102017 108 874 A1 mentioned in the opening section and US 2020 / 158499 A1 with the same priority (both incorporated herein by reference). In a particularly advantageous manner, the surface of the powder bed can be inspected with the aid of the illumination device before the particles are selectively solidified, so as to detect the occurrence of anomalies at an early stage and avoid them as much as possible.

[0069] Figure 1Indicates how the laser beam 48 irradiates a locally defined layer region 52 on the uppermost workpiece layer 20. For example, the laser beam 48 can have a diameter in the range of 1 mm to 10 mm and have a Gaussian or preferably flat-topped intensity curve. A further layer region 54 is indicated here which is not directly thermally excited around the thermally excited layer region 52. The further layer region 54 is shown here as an example which is approximately concentric with the layer region 52. However, the further layer region can also be arranged at a spatial distance from the layer region 52 and / or next to it. As already further indicated above, one or more individual time deformation curves can be determined in the further layer region 54 and can be used as a reference when analyzing the individual time deformation curves from the layer region 52.

[0070] Figure 2 As an example, a plurality of individual time deformation curves 56a, 56b, 56c, 56d are shown, which are determined here at selected pixels 58a, 58b, 58c, 58d of an image stack recorded with the camera 44. The image stack comprises a plurality of images 60, one of which is Figure 3 . The images 60 in the image stack each show a deformation on the surface of the uppermost workpiece layer after thermal stimulation thereof with a heating tool. In a preferred exemplary embodiment, the images 60 are recorded at a frame rate of 1 kHz or higher. Accordingly, the deformation curves 56a, 56b, 56c, 56d each have a time resolution of 1 ms or less. Figure 2 The time t is given in ms on the horizontal axis of , wherein the thermal excitation takes place here with a heating pulse which lasts for a few milliseconds (approximately 5 ms) and ends here at t=0. In other words, Figure 3 Different individual deformation curves 56a, 56b, 56c, 56d are shown starting from the moment when the thermal actuation is switched off at time t = 0. The dimension z in the axial direction (i.e. perpendicular to the surface of the uppermost workpiece layer 20) is given in nm on the vertical axis. The dimension z shows the deformation on the surface of the layer stack 18 perpendicular to the surface of the uppermost workpiece layer 20.

[0071] In this case, the deformation curve 56a is an example of a workpiece area (or a pixel 58a imaging this workpiece area) that does not include hidden anomalies or interfering surface roughness. The deformation curve 56a shows here a continuously decreasing curve corresponding to a deformation that continuously decreases after the heating pulse is turned off. In contrast, the deformation curve 56b initially has an individual deformation increase 62 until an individual deformation maximum 64. Only after the individual deformation maximum 64, the deformation curve 56b will fall with an individual deformation decrease 66. The so-called overshoot 68 (the difference between the individual deformation maximum 64 and the maximum of the deformation curve 56a here) is a characteristic feature of a cavity hidden under the workpiece surface and is therefore an anomaly because heat is initially accumulated above the cavity. Deformation curves 56c and 56d are examples of workpiece areas without hidden anomalies but with a roughness signal from the workpiece surface. A certain overshoot can also be seen here, but it is not as obvious as in the case of the deformation curve 56b. Furthermore, in each case the deformation reduction is flatter than in the case of the deformation curve 56 b , as can be seen from the tangent lines 70 a , 70 b , 70 d drawn in dashed lines.

[0072] Figure 4As an example, three individual time deformation curves 56e, 56f, 56g are shown, which represent the time period of the thermal actuation itself (starting from time t=0) and the time period after the thermal actuation is switched off at about t=4 ms. The deformation curve 56e is here again an example of a workpiece region (or a pixel imaging this workpiece region) which does not include hidden anomalies nor interfering surface roughness. Therefore, this deformation curve can be regarded as a reference curve. The deformation curve 56e has a moderate increase 62 in the local deformation until a maximum value 64, which is reached when the thermal actuation is switched off (here reference numeral 65), and a moderate decrease in the local deformation starting at this time. In contrast, the deformation curve 56f has a steeper increase in the local deformation during the thermal actuation than the deformation curve 56e, and also has the aforementioned overshoot 68. The individual deformation maximum 64 of the deformation curve 56f occurs only after the thermal actuation is switched off (indicated by reference numeral 65). The deformation curve 56f is typical for the case with anomalies hidden below the workpiece surface. Compared to the deformation curve 56e, the reduction 66 of deformation is also steeper. Here, the deformation curve 56g is also an example of a workpiece area without hidden anomalies, but with a roughness signal from the workpiece surface. It can be seen that both the increase in deformation and the reduction in deformation are not as steep as in the case of deformation curves 56e and 56f. Compared to the deformation curve 56e, the deformation curve 56g has an overshoot, but the overshoot is smaller than in the case of the deformation curve 56f. In addition, the absolute magnitude of the various signals depends on the corresponding local heating power and (if present) the size and depth of the defect, and also on the surface structure. Therefore, deep defects can in principle cause smaller signals than the roughness on the surface, which is why in some exemplary embodiments, the deformation curves are analyzed relative to the reference curve.

[0073] Accordingly, a number of characteristic features of the time deformation curve indicate anomalies that contrast with the roughness effects of the surface. Characteristic features 62, 64, 66 enable the detection of material anomalies and even the determination of their depth:

[0074] a) On the one hand, heat requires a short period of time to penetrate deeper into the anomaly, generate heat accumulation, and cause associated measurable surface deformation. In this initial time window "start time" during thermal excitation, the effects seen are primarily those reflected in the steepness of the increase in deformation. A characteristic feature of the anomaly is that the slope of deformation curve 56f is greater than the slopes of deformation curves 56e and 56g.

[0075] b) Another distinguishing feature between the roughness signal and the anomaly to be detected below the surface becomes visible at the moment of switching off the thermal excitation and afterwards. The surrounding area outside the anomaly cools much faster than the area above the anomaly, which leads to an elastic deformation, i.e. an additional deformation just above the anomaly. This additional deformation ("overshoot") after switching off the thermal excitation source can advantageously be used as a necessary criterion for an anomaly located below.

[0076] c) Due to the heat accumulation above the anomaly and the associated stronger overall deformation, a stronger elastic relaxation also occurs above the anomaly after switching off the thermal excitation and after the impact described in b) ("drop").

[0077] The following will add reference Figure 5 and Figure 6 Exemplary embodiments of the novel method are explained, which can be explained by means of Figure 1 According to step 80, a data set 42 is read into the controller 38, which data set defines the workpiece 14 with a plurality of workpiece layers 16, 20 arranged one above the other. As an alternative or in addition to this, the controller 38 can first receive a data set via the interface 40, which data set defines the workpiece to be manufactured "as a whole", such as a CAD data set, and based on this, determine a plurality of workpiece layers 16, 20 arranged one above the other. In this case, the controller 38 also ultimately receives a data set, which data set defines the workpiece 14 with a plurality of workpiece layers 16, 20 arranged one above the other. According to step 82, a material layer made of granular material 22 is produced on the layer stack 18 with the scraper 26.

[0078] According to step 84, the surface of the material layer is advantageously (but not absolutely necessarily) checked by means of the camera 44 and the lighting modules 50a, 50b in order to detect any anomalies, such as, in particular, grooves, holes, depressions, undulations, accumulations of material, density variations and / or grain inhomogeneities (e.g., lumps) in the material layer. If the surface of the material layer meets all the desired criteria, the method branches according to step 86 to step 88, according to which the uppermost workpiece layer 20 is produced by means of the writing laser 32. The writing laser 32 selectively melts the material grains along the defined trajectory 36 and in this way bonds the melted or partially melted grains to one another.

[0079] If the surface of the new material layer does not meet or does not meet the desired standard sufficiently, the method can advantageously return to step 82 to rework or completely recreate the surface of the material layer. According to step 90, the uppermost workpiece layer 20 that has been produced is inspected with the aid of the camera 44 and the heating tool 46, wherein the inspection based on the novel method can also detect anomalies in the depth of the layer stack 18. Anomalies can also be formed later, for example due to stress cracks or later delamination between the individual workpiece layers 16. According to step 92, steps 82 to 90 are repeated until the workpiece 14 is completed according to the data set 42. Subsequent workpiece layers can be subsequently modified if necessary in order to correct deviations in shape or size. According to step 94, the manufactured workpiece can be released for the intended use based on the inspection results of the repeated steps 82 and / or 90.

[0080] Figure 6 Shown according to Figure 4 In step 90 of the exemplary embodiment of the workpiece layer 20, a first image I0 of the uppermost workpiece layer 20 is recorded, and then thermal stimulation is performed in step 98. According to step 100, when closed (refer to Figure 2 ) and / or at the start of thermal actuation (reference Figure 4 ), recording a plurality of temporally successive (interlaced) images I N According to step 102 , it is decided whether a further image sequence F+1 is to be recorded, wherein the thermal excitation then preferably takes place in step 98 with a different intensity, a different duration and / or a different excitation position and / or excitation pattern.

[0081] The fact that characteristic transient features scale differently due to the introduced heating energy can be exploited, depending on whether they are caused by anomaly-related heat accumulation or by surface roughness. The corresponding changes or scaling behavior provide an additional distinguishing feature between pure surface effects and anomaly features when comparing the coefficient images described below for different heating settings.

[0082] According to optional step 104, advantageously all images I of the image sequence F are N Normalization. For example, we can NThe image content of the first image I0 is subtracted from the image content of the first image I0 to eliminate the image background that is not caused by thermal excitation. In order to correct vibrations, in particular if there is little surrounding material or at the edge, it can be advantageous to apply a zero-order or high-order Legendre fit (or other polynomial fit) subtraction to each image of the image stack. In addition, the Legendre fit subtraction (or other polynomial) can be advantageously used to compensate for the influence of spatially varying heating curves and / or to increase anomaly contrast. In addition, local frequency filters or Legendre fit subtraction can advantageously contribute to better distinguishing defects, because defects show different deformation behavior from their surrounding areas. Therefore, the influence of a spatially slowly varying heating curve can be distinguished from the local influence of the defect itself.

[0083] According to step 106, a plurality of individual deformation curves D are then determined for a plurality of pixels of the image sequence. I (x, y). According to step 108, using the individual deformation curve D I The coefficient image K(x,y) is determined from the image K(x,y). In one variant, the slope during the thermal excitation, the maximum height of the overshoot and / or its time point and / or the drop deformation and / or any turning point in the deformation curve can be determined pixel by pixel as coefficients. In another variant, the corresponding time variation in the deformation curve can be determined pixel by pixel in linear or logarithmic form by a polynomial or by a rational function with several degrees of freedom, preferably with 6 degrees of freedom. The coefficients of the polynomial or rational function then form the coefficients of the coefficient image K(x,y).

[0084] All information of the time curve with the above-mentioned influences is then compressed in several coefficient images K(x,y), which is advantageous in terms of storage requirements and data transmission. Using principal component analysis according to step 110, these coefficient images can be converted into a smaller number of more compressed PCA coefficient images. The clustering algorithm for segmentation purposes according to step 112 is advantageously applied to this form of compression. In combination with the threshold judgment, the probability of anomaly can then be determined in the corresponding segmented image area according to step 114. In order to obtain information about the depth of the anomaly, the time point when the defect feature first appears, i.e. the "start time", or the time point of the maximum overshoot can also be determined. Both features provide information about the relative depth of the anomaly. For example, a relatively early overshoot maximum in the time period up to 10ms after turning off the thermal excitation indicates an anomaly, while an overshoot maximum 20ms or even later after turning off the thermal excitation indicates that the deformation curve is recorded at the periphery of the workpiece layer.

[0085] Another (optional) method for distinguishing between the influence of surface roughness and the influence caused by anomalies below the surface is to combine and calculate a plurality of measurement signals recorded at the same location but in successive layers according to step 116. The corresponding layer surfaces of different layers are different and often uncorrelated, while the anomalies below the surface persist and only decrease in signal intensity due to increasing depth. If a weighted average is formed from N successive layer measurements at the same location, the surface signal is reduced by a factor of ˜(1 / sqrt(N)) compared to the anomaly ratio.

[0086] According to step 118, additional information from other measurement methods can optionally be used to further improve accuracy and reliability, in particular to improve the detection of anomalies and the separation of anomalies and surface effects, and / or to increase the speed of the measurement method by preselecting regions (ROIs), and / or to better determine the spatial position (in particular the depth) and / or to classify in terms of size and / or shape. In particular, further data from other measurement methods can be included in the analysis. These other measurement methods for multimodal analysis of measuring local or covering the entire surface include: using illumination modules 50a, 50b and DE 10 2017 108 874 A1 and US 2020 / 158499 with the same priority. The method described in A1 measures the topography and determines surface defects; measures spatially resolved vibration distribution (vibrometric method); determines surface gradients and surface shape / topography by means of the shear speckle method; measures structure-borne sound on the base plate of a workpiece to be built up in multiple layers (e.g., pulse-echo method with ultrasonic transducers and / or contactlessly with EMAT), in particular for defect classification; measures the properties of the temporary melt pool (e.g., the average temperature radiation from the melt pool) using a pyrometer or by spatially resolved imaging of the melt pool in the VIS or IR spectrum with the aid of a camera; white light interferometry (WLI) for determining statistical dimensions of the surface (e.g., roughness, power spectral density (PSD)); measurement methods for determining the temperature dependence of material constants (e.g., heat capacity, thermal expansion, thermal conductivity, elastic modulus) in the process-relevant range (room temperature to melting temperature).

Claims

1. A method for additively manufacturing a workpiece (14), the method comprising the following steps: a) obtaining a data set (42) which defines the workpiece (14) with a plurality of workpiece layers (16, 20) arranged one above the other, b) producing the plurality of workpiece layers (16, 20) arranged one above the other using a layer forming tool (32) controlled according to the data set (42), wherein: The plurality of workpiece layers (16, 20) arranged one above the other form a layer stack (18) which, at a defined point in time, has an uppermost workpiece layer (20) and a plurality of underlying workpiece layers (16), c) thermally actuating (98) the layer stack (18) at the defined point in time, d) recording a plurality of images (60) of the uppermost workpiece layer (20), and e) inspecting the layer stack (18) using the plurality of images to obtain inspection results representative of the workpiece, Characterized in that the multiple images (60) of the uppermost workpiece layer (20) are recorded in a time series in a time-interleaved manner, so that the multiple images include an image stack, and the images are contained in the image stack in a time sequence, wherein the multiple images (60) are used to determine a separate time deformation curve (56) of the uppermost workpiece layer (20) in response to the thermal excitation (98), wherein the separate time deformation curve (56) has a plurality of characteristic features, and the plurality of characteristic features include a separate deformation increase (62), a separate deformation maximum (64) and a separate deformation decrease (66), wherein the separate time deformation curve (56) is analyzed within a defined time interval at the beginning of and / or after the thermal excitation, and wherein the inspection result is determined using at least two of the characteristic features from the plurality of characteristic features of the separate time deformation curve (56).

2. The method according to claim 1, characterized in that Each image (60) from the plurality of images has a plurality of image patches; and a corresponding individual time deformation curve (56) in response to the thermal stimulus (98) is determined for each image patch from the plurality of image patches, wherein the inspection result is determined using the corresponding individual time deformation curve (56).

3. The method according to claim 1 or 2, characterized in that Thermally actuating the layer stack (18) includes thermally actuating the topmost workpiece layer (20) in a locally defined first layer region (52); and the plurality of images show the first layer region (52) and another layer region (54) of the topmost workpiece layer (20), wherein the first layer region (52) and the another layer region (54) are non-intersecting.

4. The method according to claim 1 or 2, characterized in that: The plurality of images comprises at least one first image of the uppermost workpiece layer (20) recorded before thermal actuation, wherein the inspection result is determined using the at least one first image.

5. The method according to claim 1 or 2, characterized in that: Prior to determining the individual time deformation curve, the plurality of images are normalized using at least one reference image.

6. The method according to claim 1 or 2, characterized in that: The multiple images were recorded with a frame rate ≥ 1 kHz.

7. The method according to claim 1 or 2, characterized in that: The inspection result is determined based on the plurality of images using principal component analysis.

8. The method according to claim 1 or 2, characterized in that: The layer stack (18) is subjected to repeated thermal excitations, wherein at least one excitation parameter is varied, and wherein the inspection result is determined based on the repeated thermal excitations.

9. The method according to claim 8, characterized in that The at least one excitation parameter is an excitation intensity and / or an excitation duration.

10. The method according to claim 1 or 2, characterized in that: The inspection result is determined using a plurality of images of a plurality of corresponding uppermost workpiece layers (20).

11. The method according to claim 1 or 2, characterized in that: The individual time deformation curves are determined using at least one of the following measurement methods: speckle interferometry, digital holography, shear speckle method; laser Doppler vibrometry, Fabry-Perot interferometry, Sagnac interferometry, interferometry using nonlinear optics.

12. The method according to claim 1 or 2, characterized in that: The inspection result is also determined using thermal transient curves and / or using ultrasonic excitation and / or using simulated deformation curves and / or using melt pool characteristics and / or using angle selective illumination of the uppermost workpiece layer.

13. A device for additively manufacturing a workpiece (14), the device comprising a memory for obtaining a data set (42) which defines the workpiece (14) with a plurality of workpiece layers (16, 20) arranged one above the other, - a manufacturing platform (12), - layer forming means (32), - a heating means (46), - a camera (44) directed toward the manufacturing platform (12), and an evaluation and control unit (38) configured to produce a plurality of workpiece layers (16, 20) arranged one above the other on the manufacturing platform (12) using the layer forming tool (32) and the data set (42), wherein The multiple workpiece layers (16, 20) arranged one above the other form a layer stack (18), which has an uppermost workpiece layer (20) and multiple workpiece layers (16) below at a defined time point, and the evaluation and control unit is also configured to use the heating tool (46) to thermally excite the layer stack (18) at the defined time point, and to use the camera (44) to record multiple images of the uppermost workpiece layer (20), and finally to inspect the layer stack (18) using the multiple images to obtain an inspection result representing the workpiece (14), characterized in that the evaluation and control unit (38) is also configured to use the multiple images to determine a separate time deformation curve of the uppermost workpiece layer in response to the thermal excitation, wherein the separate time deformation curve has multiple characteristic features, the multiple characteristic features including a separate deformation increase, a separate deformation maximum value and a separate deformation decrease, and wherein the evaluation and control unit (38) uses at least two of the characteristic features mentioned from the multiple characteristic features of the separate time deformation curve (56) to determine the inspection result, and The device is configured to perform the method as claimed in any one of claims 1 to 12.

Images