Image processing method for determining a process disturbance, and image processing device

Abstract

The invention relates to an image processing method for determining a process disturbance (S). The method comprises, (i), capturing a plurality of images (B) during an operating process or manufacturing process, (ii), classifying the images (B) as those captured before, during and / or after a process event triggering the disturbance (S), (iii), calculating a background for each image capture (B) during the process event by comparing same with other image captures before or after the process event, and, (iv), calculating an entirety (G) of the disturbances (S) from the image captures (B) during the process event, from which image captures the calculated background has been removed. The invention also relates to an image processing device, the use thereof and a corresponding computer program product.

Description

[0001] Description

[0002] Image processing method for determining a process disturbance and image processing device

[0003] The present invention relates to an image processing method for determining a process disturbance or an artifact, as well as a corresponding image processing device. In particular, the method is an image processing method for determining the amount of process smoke as a process disturbance in additive manufacturing. Furthermore, a corresponding monitoring system and a computer program product associated with the method are part of the present invention.

[0004] In particular, the process disturbance may be an artifact that distorts or disrupts the measurement result or image acquisition. In contrast, the corresponding process may generally affect an operating procedure or a manufacturing process, preferably manufacturing processes with recurring, similar process sequences.

[0005] Even more specifically, the manufacturing process can involve layer-by-layer processes for the (powder-bed-based) production of components.

[0006] The design and material properties of high-performance components are subject to continuous development to increase or expand the functionality and / or application areas of the corresponding components. In heat engines, especially gas turbines, development often aims at ever-increasing operating temperatures. To meet the challenges of changing industrial requirements, for example, development efforts particularly aim to increase the strength, increase thermomechanical resilience, and extend the service life of such component structures.

[0007] Due to technical advances, generative or additive manufacturing is becoming increasingly interesting for the series production of the above-mentioned components, such as turbine blades or burner components.

[0008] Additive manufacturing processes (AM: "additive manufacturing") , colloquially also referred to as 3D printing, include, for example, selective laser melting (SLM) or laser sintering (SLS) , or electron beam melting (EBM) . Other additive processes include, for example, "Directed Energy Deposition (DED)" processes, in particular laser material deposition, electron beam or plasma powder welding, wire welding, metallic powder injection molding, so-called "sheet lamination" processes, or thermal spray processes (VPS LPPS, GDCS).

[0009] Additive manufacturing processes have proven particularly advantageous for complex or intricately designed components, such as labyrinth-like structures, cooling structures, and / or lightweight structures. Additive manufacturing is particularly advantageous due to its particularly short chain of process steps, as a manufacturing or production step of a component can be carried out largely based on a corresponding CAD file and the selection of appropriate manufacturing parameters.

[0010] The production of gas turbine blades using the described powder bed-based processes (LPBF, or Laser Powder Bed Fusion) advantageously enables the implementation of new geometries or concepts that can reduce manufacturing costs or assembly and lead times, optimize the manufacturing process, and, for example, improve the thermo-mechanical design or durability of the components. Components manufactured conventionally, such as by casting, are still inferior to the additive manufacturing route, for example, in terms of their design freedom and also in terms of the required lead time and the associated high costs and manufacturing effort. However, the powder bed process inherently creates high thermal stresses in the component structure.In particular, irradiation paths or vectors that are too short lead to severe overheating, which in turn causes structural distortion. Severe distortion during the build process can easily lead to structural delamination, thermal deformation, or geometric deviations outside of acceptable tolerances.

[0011] The LPBF process also inherently produces disruptive by-products during production, including process smoke. This process smoke interacts with the melt beam, causing shadowing and secondary effects. This results in inaccuracies in irradiation, corresponding process deviations, and significantly impairs the structural quality of the component being manufactured. The influence of such a disturbance, such as process smoke, is usually localized and dependent on its position on the build plate. This makes it difficult to attribute the resulting defects in the component to a specific influencing factor or irradiation parameter. The amount of smoke produced depends primarily on the material used, the shielding gas employed, and the process parameters.

[0012] It is not yet possible to reliably predict, determine or quantify the amount of smoke produced using currently available measurement methods.

[0013] It is therefore an object of the present invention to provide means which allow process disturbances, such as the said smoke development, to be reliably determined and / or quantified by means of the layer-by-layer process.

[0014] This object is achieved by the subject matter of the independent patent claims. Advantageous embodiments are the subject matter of the dependent patent claims. One aspect of the present invention relates to an image processing method for determining a process disturbance or an artifact, such as the aforementioned procedural smoke development. The method comprises capturing a plurality of images during the corresponding operating or manufacturing process.

[0015] According to the invention, the images can be recorded by common image acquisition, process monitoring measures, such as in particular optical tomography or camera recordings, for example CCD cameras or other image sensors.

[0016] The method further comprises dividing the images into those taken before, during, and / or after a process event that triggered the said disturbance. The process event is preferably the trigger of the disturbance or measurement deviation; in the case of an additive manufacturing process, preferably the irradiation of the powder bed, which is the cause of the development of the process smoke.

[0017] The method further comprises calculating or correcting an image background (including image noise) for each image captured during the process event or each image captured during the process event by comparing said image capture with further image captures before or after the process event (images captured before and after the event).

[0018] The method further comprises calculating, defining, or displaying the totality, quantity, or intensity of the disturbances from the images acquired during the process event, adjusted for the calculated background or for a darkening effect (contrast shift) caused by the process smoke and / or for image noise. This advantageously creates a background image through image processing or process monitoring that shows the partially irradiated component structures, but excludes the disruptive influence of the melt pool and process smoke.

[0019] Furthermore, such a measurement or monitoring method can be used to optimize process instructions (see CAM) with regard to the amount or reduction of inherently occurring disturbances, such as smoke. Indirectly or through further investigations, the structural quality and dimensional accuracy of the component being manufactured can also be significantly improved in this way.

[0020] In particular, the present invention makes it possible for the first time to develop reliable process parameters for the industrialized additive manufacturing of components and, in particular, to improve their reproducibility. Finally, the influence of the disturbance, which in this case can relate not only to process smoke but also to other disruptive effects, can be reliably quantified using the inventive approach, and conclusions about the process can be drawn from this. In particular, the amount of smoke or the overall intensity of the disturbance variable to be determined via the image processing method, which arises when using a set of parameters, can be precisely determined.

[0021] Indirectly, component quality can also be advantageously improved through improved process parameter development, which must be carried out for each new material and each new printer machine type.

[0022] Furthermore, the present invention allows the temporal development of the intensity or totality of the disturbance to be determined and recorded dynamically and / or by means of process monitoring of the underlying process. This makes it possible to determine further time-dependent findings, such as deviations in smoke development over time.

[0023] In one embodiment, the images are additionally divided into individual recurring process steps during the underlying process.

[0024] In one embodiment, the recurring process steps relate to the production of individual layers during an additive manufacturing process of the component.

[0025] In one embodiment, the background is calculated pixel by pixel for each image captured during the process event (i.e., each image captured during the process event). This embodiment achieves the most accurate resolution and background correction possible during the process.

[0026] In one embodiment, the smallest difference between a measured value of the respective image taken during the process event and the image taken before and after the process event is adopted or used as a basis for calculating the background. This advantageously creates a background image that is corrected for interference effects such as process smoke and any overexposure.

[0027] In one embodiment, the majority of the images are recorded using image sensors, preferably via a camera or via optical tomography, and corresponding contrasts or gray levels of the images are recorded or calculated as a measured variable.

[0028] In one embodiment, image division involves calculating a temporal average of a measured variable, such as an optical image parameter, for images taken before and after the process event. This averaging enables particularly accurate background calculation while excluding interference effects in image processing in a particularly effective manner.

[0029] In one embodiment, the method comprises calculating or displaying a temporal progression of the entirety of the disturbances during the process event. This embodiment achieves the advantages mentioned above. In particular, a temporal progression or a temporal progression of the entirety of the disturbing artifacts can be determined and, through further processing of such data, correlated with the recording of key process parameters such as the laser or irradiation power, a scanning speed, and / or a hatch or hatch distance in order to draw further conclusions and, in particular, to develop a type of digital twin or model of the overall process.With the help of this model, it is possible to predict and provide a direct relationship between the resulting process smoke and the component quality, for example via a target-actual comparison, or a corresponding process tolerance for quality assurance.

[0030] In one embodiment, the totality of disturbances during the process event is calculated using a histogram, which also represents image noise with its relative frequency over the recorded measured value. This embodiment allows the totality of disturbances, such as the totality of process smoke, to be evaluated and presented in a particularly clear manner.

[0031] In one embodiment, the method is a method for determining the amount of process smoke as a disturbance or falsification of an image recording in powder bed-based additive manufacturing.

[0032] A further aspect of the present invention relates to an image processing device that is configured or suitable for performing an image processing method for determining the process disturbance (as described above). A further aspect of the present invention relates to the use of an image processing device for the (isolated) quantification of process smoke in the additive manufacturing of components, in particular powder-bed-based.

[0033] A further aspect of the present invention relates to a monitoring system, for example as part of a monitoring solution in conventional additive manufacturing systems, comprising the described image processing device.

[0034] A further aspect of the present invention relates to a computer program or computer program product comprising instructions which, when the program is executed by a computer, for example for controlling and / or monitoring the irradiation in an additive manufacturing system, cause the computer to determine the disturbance as described above.

[0035] A CAD file or a computer program product can be provided, for example, as a (volatile or non-volatile) storage or playback medium, such as a memory card, a USB stick, a CD-ROM or DVD, or in the form of a downloadable file from a server and / or in a network. Provision can also be made, for example, in a wireless communications network by transmitting a corresponding file containing the computer program product. A computer program product can contain program code, machine code or numerical control instructions, such as G-code, and / or other executable program instructions in general.

[0036] In one embodiment, the computer program product relates to manufacturing instructions according to which an additive manufacturing system, for example via CAM means, is controlled by a corresponding computer program to produce the component. The computer program product can further contain geometric data and / or design data in a data set or data format, such as a 3D format or as CAD data, or comprise a program or program code for providing this data.

[0037] Embodiments, features and / or advantages which in the present case relate to the image processing method or the computer program (product) also relate to the image processing device and the monitoring system, and vice versa.

[0038] The term "and / or" or "respectively," when used in a series of two or more elements, means that any one of the listed elements may be used alone, or any combination of two or more of the listed elements may be used.

[0039] Further details of the invention are described below with reference to the figures.

[0040] Figure 1 shows a schematic perspective view of an additive manufacturing system comprising an image processing device according to the invention.

[0041] Figure 2 shows a simplified schematic flow diagram indicating process steps according to the invention.

[0042] Figures 3 to 6 each indicate, using histograms, the image-processing inventive quantitative determination of process smoke in additive manufacturing processes.

[0043] Figure 7 further shows, by way of example, the temporal progression of the determined quantity of the process smoker.

[0044] In the exemplary embodiments and figures, identical or equivalent elements may be provided with the same reference numerals. The illustrated elements and their relative sizes are generally not to scale; rather, individual elements may be exaggeratedly thick or oversized for clarity and / or clarity.

[0045] Figure 1 shows an additive manufacturing system 100. The manufacturing system 100 is preferably designed as an LPBF system and for the additive construction of parts 10 or components from a powder bed. The system 100 can also specifically relate to an electron beam melting system.

[0046] Accordingly, the system has a build platform on which the component geometry is produced, and thus welded. The component 10 is produced layer by layer from a powder or powder bed 1. For this purpose, the powder is distributed layer by layer on a build platform using a coating device (not further identified). After each powder layer has been applied, regions of the layer are selectively melted and then solidified using an energy beam, for example a laser or electron beam 3, from an irradiation device or beam source 2 in accordance with the predetermined geometry of the component 10.

[0047] After each layer, the build platform is preferably lowered by an amount corresponding to the layer thickness. This thickness is usually only between 20 and 40 pm, so that the entire process can easily encompass the selective irradiation of thousands to tens of thousands of layers. Due to the very local energy input, high temperature gradients, for example, of 10 6K / s or more can occur. Naturally, the stress state of the component is correspondingly large during and after construction, which considerably complicates the additive manufacturing processes. The geometry of the component is typically defined by a CAD ("Computer-Aided Design") file. After such a file has been imported into the manufacturing system 100, the additive process then first requires the definition of a suitable irradiation strategy, for example by means of CAM ("Computer-Aided Manufacturing"), which can also be used to divide the component geometry into individual layers.

[0048] As will be further explained with reference to the description of Figure 2 (see below), the image processing method according to the invention preferably relates to the determination of process disturbances in powder bed-based additive manufacturing, wherein images B are recorded by a recording device 20, for example as part of an image processing device 30, during the manufacturing process, preferably layer by layer, in particular in order to determine and / or quantify the formation of process smoke S as a process disturbance.

[0049] The said recording device 20 can record the images B, for example, via appropriate image acquisition or process monitoring measures, such as in particular camera recordings, optical tomography, CCD cameras or other image sensors.

[0050] The image processing device 30 can, for example, further be configured to determine the process disturbance S via, by means of, or using a computer program or computer program product. Furthermore, the determined process disturbance (see below) or a disturbance ensemble or even a temporal profile of said disturbance ensemble can be part or integral of the computer program product resulting from a corresponding computer program. Figure 1 further shows a monitoring system 40 for additive manufacturing comprising the image processing device 30.

[0051] Figure 2 illustrates the essential method steps according to the invention in a simplified schematic flow diagram. Accordingly, the invention relates to an image-processing method for determining the process disturbance S, comprising (i) capturing a plurality of images B during the described operating or manufacturing process.

[0052] Furthermore, the method (ii) comprises dividing the images B into those recorded before, during, and / or after a process event triggering the disturbance S. This process event is to be understood as triggering the actual disturbance; in additive manufacturing processes, this event therefore concerns the irradiation by means of a laser or electron beam in the paths and vectors specified, for example, by CAM for the layer-by-layer selective solidification of the structure according to the specified CAD component geometry.

[0053] The image division may preferably comprise forming a temporal average of a measured value of the images B taken before and after the process event, such as brightness or grayscale.

[0054] First, the acquired images are divided into individual slice images, preferably according to the layer sequence during the manufacturing process. Further, as described, they are subdivided according to the time of acquisition. If, for example, the acquisition time is before or after the exposure of a particular layer or its irradiation vectors, the images are used to calculate the background and image noise. If, however, the image was acquired during exposure, it is used to determine the amount of smoke-induced interference.

[0055] As described, the background or image noise can preferably be calculated pixel by pixel. For the aforementioned calculation, the smallest difference between a measured value of the respective image B, which was acquired during the process event, and the image before and after the process event is preferably used to calculate the background.

[0056] In other words, the image background is preferably calculated individually for each image on which the build plate is exposed or imaged. The background is made up of the images taken before and after exposure. To do this, the image taken during exposure is compared pixel by pixel with the images taken before and after exposure. If a pixel has a smaller difference than the pixel in the image that is part of an image after exposure, this pixel in the background also takes on its value. If the difference is smaller than the pixel in the image taken before exposure, the pixel with the smaller difference in the background takes on this value. This advantageously creates a background image that shows the partially fully exposed components, but not the melt pool and process smoke.

[0057] By comparing the exposed image with those taken before and after exposure and adopting the smaller difference in each case, a more useful result is in principle provided and, in particular, a light-dark contrast resulting from the structure that has already solidified in layers is minimized or eliminated.

[0058] The method further comprises (iii) calculating a background and / or image noise for each image B taken during the process event by comparing it with further image recordings before or after the process event.

[0059] The method further comprises, (iv) , calculating a total G of the disturbances S from the image recordings B corrected for the calculated background during the process event .

[0060] In particular, the amount of smoke per image can be determined using an evaluation algorithm applied to the captured images. The measured data are advantageously independent of the image noise generated in the corresponding image sensor thanks to the background calculation or background correction described above.

[0061] With this approach, the present invention makes a significant contribution to significantly improved process monitoring and even to the optimization of the parameter set to be developed depending on the material and design.

[0062] Preferably, the majority of images B are finally recorded using image sensors and the corresponding grayscale or black-and-white contrasts of the images are recorded as a measurement variable.

[0063] These brightness values ​​or grayscale levels of the pixels in the recorded images are actually composed of various influences. Most pixels absorb the light which - in this case - is reflected by the powder bed 1. As soon as process smoke S is above the powder bed 1, the reflected light is deflected or scattered and the brightness value decreases. If the powder bed 1 is illuminated by the laser beam 3, the brightness value of the melt pool increases. In addition, an inherent measurement error due to image noise occurs for each pixel during image acquisition. The images taken shortly before and after exposure usually show the build platform or production surface for a few seconds (see Figure 1). These images basically depict a similar or the same scenario, but may differ slightly due to image noise.However, the noise contrast is weak enough to assume the average values ​​of the pixels to be approximately constant, and the corresponding noise function to be reasonably accurately normally distributed. Therefore, the temporal average of the corresponding images is first calculated.

[0064] As already indicated above, the differences between the pixels of this average image and the other or subsequent images are preferably calculated, which are then subsequently displayed in a histogram (see Figures 3 to 6 below). The histogram thus displayed also represents the image noise, or includes it. In this sense, the totality G of the disturbances can only be calculated and / or displayed via the histogram during the process event.

[0065] Figure 3 uses a simple histogram to show a relative frequency of image noise plotted against individual difference classes of the pixel measurement values ​​of the image sensor, for example a brightness value, gray value or a gray level.

[0066] Figure 4 shows - analogous to Figure 3 - a histogram of the total image sensor signal, ie the recorded process smoke including image noise.

[0067] In an analogous representation, Figure 5 shows the aforementioned difference between the "total signal" and the background, or background signal H. In other words, this calculated background image is subtracted from the respective image acquired during exposure. The resulting difference image is then used to generate the corresponding histogram. This histogram is composed of the aforementioned influences. The (backscattered) laser light, as is well known, only occupies values ​​at the upper end, or right-hand part, of the histogram.

[0068] From the representation in Figure 6, it is also clear that this right-hand part has been cut out and thus set to zero. The right and therefore positive side of the histogram is therefore only influenced by the background image noise (see above). Using a numerical optimizer, a factor can now be determined between the two distributions so that both histograms lie as close to each other as possible in the positive range. In this way, an overexposure error of the specific disturbance or its entirety is essentially corrected. Instead of the aforementioned numerical optimizer, for example, another suitable fitting algorithm or a suitable regression analysis or adjustment calculation can be used.

[0069] By subtracting the individual histograms as described, a distribution is created that is advantageously only influenced by the deflection of the process smoke.

[0070] The high value of the relative frequency in the individual histograms of Figures 3 to 6 with a difference of zero reflects the fact that a gray value or a pixel brightness must be taken into account in a truncated or scaled manner beyond the "limit" of the measurement accuracy of the corresponding image sensor or recording device of the sensor.

[0071] The quantity of the disturbance ensemble G or of the process smoke S can then be calculated using the sum of the products of each difference value with the respective frequency. In this way, the quantity of smoke in the respective image is determined independently of the background image noise. In the present case of the histogrammatic representation, the quantity of smoke is determined in particular using the sum of the relative frequencies across the individual classes or difference bars. Figure 7 merely indicates, in an abstract and exemplary manner, the temporal progression G(t) of the quantity of the determined process smoke S over time. Since this step is optional and not absolutely essential to the invention, it is only indicated by dashed connecting lines in the flow chart in Figure 2.

[0072] This advantageously allows for a temporal progression of the amount of process smoke above the build plate to be determined. This can be used, among other things, for comparing different parameters or for process monitoring in general.

[0073] Without loss of generality, the present imaging method for determining the process disturbance may alternatively refer to other methods, for example laser-based, standardized or automated operating procedures.

Claims

Patent claims 1. Image processing method for determining the amount of process smoke in additive manufacturing as a process disturbance, comprising: - (i) taking a plurality of images (B) during an operating or manufacturing process, - (ii) subdividing the images (B) into those taken before, during and / or after a process event triggering the disturbance (S), wherein the subdividing of the images (B) is additionally carried out into individual recurring process steps during the operating or manufacturing process, and wherein the recurring process steps relate to the production of individual layers during a manufacturing process, - (iii) calculating a background for each image acquisition (B) during the process event by comparing it with further image acquisitions before or after the process event, and - (iv) calculating a total (G) of the disturbances (S) from the image recordings (B) corrected for the calculated background during the process event.

2. The method according to claim 1, wherein the background is calculated pixel by pixel for each image acquisition (B) during the process event.

3. Method according to one of the preceding claims, wherein for the calculation of the background a smallest difference of a measured value of the respective image (B) which was recorded during the process event with that recording before and after the process event is adopted for the calculation of the background.

4. Method according to one of the preceding claims, wherein the plurality of images (B) are recorded by image sensors and gray levels of the images are recorded as a measured variable.

5. Method according to one of the preceding claims, wherein the image division comprises in each case forming a temporal average of a measured value of such images (B) which were recorded before and after the process event.

6. Method according to one of the preceding claims, comprising calculating a time course (G(t), v) of the totality (G) of the disturbances (S) during the process event.

7. Method according to one of the preceding claims, wherein the totality (G) of the disturbances (S) during the process event is calculated via a histogram.

8. Image processing device (30) which is designed to carry out an image processing method for determining a disturbance (S) according to one of the preceding claims.

9. Monitoring system (40) for additive manufacturing comprising an image processing device (30) according to the preceding claim.

10. Computer program product (CP) comprising instructions which, when the program is executed by a computer, for example for controlling and / or monitoring the irradiation in an additive manufacturing system (100), cause the computer to determine the disturbance (S) according to one of claims 1 to 7.

Images