Image processing method and image processing device for determining process interference

By recording and analyzing images in the additive manufacturing process through image processing methods, the problem of difficult to quantify process smoke is solved, component quality and process parameters are improved, and reliable quantification and dynamic monitoring of process smoke is achieved.

CN120303694APending Publication Date: 2025-07-11SIEMENS ENERGY GLOBAL GMBH & CO KG
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202380081829.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2022-11-28
Filing Date
2023-10-25
Publication Date
2025-07-11

AI Technical Summary

Technical Problem

The prior art cannot reliably predict, determine or quantify the amount of process smoke generated during additive manufacturing, making it difficult to ensure the quality of component structure.

Method used

Images in the additive manufacturing process are recorded and analyzed by image processing methods, including image comparison and background correction before and after process events, calculating the amount of interference of process smoke, and optimizing process parameters to improve component quality.

Benefits of technology

Reliable quantification of process smoke is achieved, structural quality and dimensional precision of components are improved, reproducibility and reliability of process parameters are enhanced, and the intensity changes of interference are dynamically monitored.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120303694A_ABST
    Figure CN120303694A_ABST
Patent Text Reader

Abstract

The invention relates to an image processing method for determining process disturbances (S). The method comprises: (i) recording a plurality of images (B) during a run-or manufacturing process, (ii) classifying the images (B) as images recorded before, during and / or after a process event that initiates an interference (S), (iii) identifying the interference (S) by comparing each image record (B) during the process event with other image records before and after the process event. Calculating the background of each image record (B) during the process event, and (iv) calculating the entirety (G) of the interference (S) from the image records (B) during the process event from which the calculated background has been cleared. The invention further relates to an image processing device, to the use thereof, and to a corresponding computer program product.
Need to check novelty before this filing date? Find Prior Art

Description

Field of the Invention

[0001] The present invention relates to a method for image processing for determining process disturbances or artifacts, and a corresponding image processing device. In particular, the method is for image processing for determining the amount of process smoke as a process disturbance in additive manufacturing. In addition, a corresponding monitoring system and a computer program product belonging to the method are part of the present invention. Background Art

[0002] In particular, process disturbances can be artifacts that distort or interfere with measurement results or image detection. In contrast, the corresponding processes generally can relate to an operating method or a manufacturing method, preferably a manufacturing method having the same type of process flow always repeated. More specifically, the manufacturing method can be a layer-by-layer method for manufacturing components (based on a powder bed).

[0003] The design and material properties of high-performance components are the subject of continuous development in order to improve or expand the functionality and / or field of use of the corresponding components in use. In thermal machines, especially in gas turbines, the development generally aims at increasingly higher operating temperatures. In order to meet the challenges of changing industrial requirements, for example, the development particularly strives for an increase in the strength of such component structures, an increased thermo-mechanical load capacity, and a service life.

[0004] Generative or additive production is also becoming increasingly interesting for the serial production of the above-mentioned components, such as turbine blades or burner components, due to technological improvements.

[0005] Additive manufacturing methods (AM: "additive manufacturing"), also popularly known as 3D printing, include, for example, selective laser melting (SLM) or laser sintering (SLS), or electron beam melting (EBM) as powder bed methods. Other additive methods are, for example, "Directed Energy Deposition, directed energy deposition (DED)" methods, especially laser deposition welding, electron beam or plasma powder welding, wire welding, metal powder injection molding, the so-called "sheet lamination" method, or thermal spraying methods (VPS LPPS, GDCS).

[0006] For components designed complexly or finely, such as labyrinth structures, cooling structures, and / or lightweight structures, additive production methods have proven to be particularly advantageous. In particular, additive production is advantageous due to a particularly short process step chain, since the manufacturing or production steps of the component can be carried out to a large extent based on the corresponding CAD file and the selection of corresponding production parameters.

[0007] Manufacturing gas turbine blades by means of the described powder bed-based method (“LPBF”, English “Laser Powder Bed Fusion”) advantageously enables the implementation of new geometries or concepts that can reduce manufacturing costs or construction and production cycles, optimize the manufacturing process, and, for example, improve the thermomechanical design or durability of the component. Components manufactured in a conventional manner, such as by casting techniques, are significantly inferior to additive manufacturing processes, for example, in terms of their shaping freedom, as well as in terms of the required production cycles and the high costs and production technology expenditures associated with the production cycles.

[0008] However, high thermal stresses are inherently generated in the component structure by the powder bed process. In particular, an overly short irradiation path or vector causes strong overheating, which in turn causes warping of the structure. However, strong warping during the construction process also easily causes structural detachment, thermal deformation, or geometric deviations beyond the allowable tolerances.

[0009] In the LPBF method, inherently interfering by-products are also generated during the production process, and process fumes belong to such by-products. The process fumes cause a shielding effect and secondary effects through interaction with the molten beam. This causes irradiation inaccuracies, corresponding process deviations, and, decisively, also damage to the structural quality of the component to be manufactured. The influence of such interferences, such as process fumes, is usually locally related to the position on the construction platform. This makes it difficult to attribute the defects generated in the component to specific influencing factors or irradiation parameters. The amount of the generated fumes is mainly related to the material used, the protective gas used, and the process parameters.

[0010] So far, the amount of the generated fumes cannot be reliably predicted, determined, or quantified by means of currently available measurement methods. Summary of the Invention

[0011] Therefore, the object of the present invention is to provide a mechanism that allows reliable determination and / or quantification of process interferences, such as the fume generation during a layer-by-layer process.

[0012] The above object is achieved by the subject matter of the independent claims. Advantageous designs are the subject matter of the dependent claims.

[0013] One aspect of the present invention relates to an image processing method for determining process interferences or artifacts, such as the generation of the process fumes. The method includes recording a plurality of images during a corresponding operation or manufacturing process.

[0014] According to the present invention, the images can be recorded by means of common measures for image detection and process monitoring, such as, in particular, optical tomography or camera recording, for example, by a CCD camera or other image sensors.

[0015] The method further comprises: classifying the images into images recorded before, during and / or after a process event that triggers the mentioned interference. The process event is preferably a triggering factor for the interference or measurement deviation; in the case of an additive manufacturing method, it is preferably the irradiation of the powder bed, which is the cause of process fume generation.

[0016] The method further comprises: calculating or correcting the image background (including image noise) for each image recording during the process event or each image recording recorded during the process event by comparing each image recording during the process event or each image recording recorded during the process event with other image recordings before or after the process event (images recorded before and after the event).

[0017] The method further comprises: calculating, defining or displaying the overall amount or intensity of the interference based on the image recording during the process event that has been cleared of the calculated background or cleared of the darkening effect (contrast shift) caused by process fumes and / or cleared of image noise.

[0018] Advantageously, a background image is thereby formed during the image processing or process monitoring, which shows the partially irradiated component structure, but does not show the disturbing effects of the melt pool and process fumes.

[0019] Furthermore, by means of such a measurement or monitoring method, the process instructions (see CAM) can be optimized in terms of the interference factors inherently formed in the process, such as the amount of the generated fumes or reduction thereof. In the said manner, the structural quality and dimensional precision of the component to be manufactured can of course also be significantly improved indirectly or through extensive research.

[0020] In particular, for the first time by the present invention, reliable process parameters for the industrial additive manufacturing of components are developed and in particular their reproducibility is improved. Finally, by the solution according to the present invention, the influence of the interference can be reliably quantified, so as to infer the process, and the interference currently may not only relate to process fumes but also to other interference effects. In particular, the amount of the fumes or the total intensity of the interference variables to be determined by the image processing method can be precisely determined, and the interference variables are generated when using the parameter set.

[0021] Indirectly, the component quality can also be advantageously improved by the improved process parameter development, which must be implemented for each new material and each new printer-machine type.

[0022] Furthermore, the present invention allows the determination and recording of the development of the intensity or overall amount of the interference over time dynamically and / or during the process monitoring of the underlying process. By this, it is feasible to determine further time-related knowledge, such as the deviation of fume generation with respect to time.

[0023] In one design, during the underlying process, the classification of the images is additionally carried out in each repeated process step.

[0024] In one design, the repeated process involves manufacturing the individual layers during the additive manufacturing process of the component.

[0025] In one design, the background of each image record (i.e., each image recorded during the process event) during the process event is calculated pixel by pixel. By means of this design, as precise a resolution and background correction as possible are achieved during the course of the method.

[0026] In one design, for calculating the background, the minimum difference between the measured variables of the corresponding image recorded during the process event and the recordings before and after the process event is respectively used or based on, for calculating the background. Thereby, a background image is advantageously formed, which background image eliminates interference effects such as process smoke and possible overexposure.

[0027] In one design, a plurality of images are recorded in the manner of an image sensor, preferably via a camera or via optical tomography, and the corresponding contrast or gray level of the images is detected or calculated as the measured variable.

[0028] In one design, the image classification respectively includes forming the time average of the measured variables, such as optical image parameters, of such images recorded before and after the process event. By means of this averaging, a particularly precise background calculation can be achieved in a particularly purposeful manner, especially in the case of excluding interference effects in image processing.

[0029] In one design, the method includes: calculating or displaying the temporal variation curve of the overall interference during the process event. The advantages mentioned above are achieved by means of this design. In particular, the temporal variation curve or the overall temporal variation curve of the interference artifacts can be determined, and in the further processing of such data, the temporal variation curve is correlated with the detection of the main process parameters, such as laser or irradiation power, scanning speed and / or grid spacing or fill spacing, in order to draw further conclusions, and in particular to develop a digital twin or model of the overall process. By means of the model, it is again possible, for example, via a desired-actual comparison, to predict and provide the direct correlation between the formed process smoke and the component quality, or to predict and provide the corresponding process tolerances for quality assurance.

[0030] In one design, the overall interference during a process event is calculated via a histogram, which additionally also represents image noise with a relative frequency with respect to the detected measurement variable. Via this design, the overall interference, such as the overall process smoke, can be evaluated particularly clearly, well, and clearly shown.

[0031] In one design, the method is for determining the amount of process smoke that is an interference or distortion in an image recording in powder bed-based additive manufacturing.

[0032] Another aspect of the present invention relates to an image processing device configured or adapted to perform a method of image processing for determining process interference (as described above).

[0033] Another aspect of the present invention relates to the use of an image processing device for (independently) quantifying process smoke in the additive manufacturing of a component, especially based on a powder bed.

[0034] Another aspect of the present invention relates to a monitoring system, for example as part of a monitoring solution in a conventional additive manufacturing facility, the monitoring system including the described image processing device.

[0035] Another aspect of the present invention relates to a computer program or computer program product, the computer program or computer program product including instructions that, when the program is executed by a computer, for example for controlling and / or monitoring irradiation in an additive manufacturing facility, cause the computer to determine interference as described above.

[0036] A CAD file or computer program product can for example be provided or exist as a (volatile or non-volatile) storage or reproduction medium, such as a memory card, USB stick, CD-ROM, or DVD, or also in the form of a file downloadable from a server and / or in a network. Additionally, for example, it can also be provided in a wireless communication network by transmitting a corresponding file with the computer program product. A computer program product can generally contain program code, machine code, or digital control instructions, such as G-code, and / or other executable program instructions.

[0037] In one design, the computer program product relates to manufacturing instructions according to which, for example, an additive manufacturing facility is controlled via a corresponding computer program by a CAM mechanism for manufacturing a component.

[0038] The computer program product can also contain geometric data and / or structural data in the form of a data set or data format, such as a 3D format or as CAD data, or include a program or program code for providing the data.

[0039] A design solution, feature, and / or advantage of a method or computer program (product) related to image processing also relates to an image processing device and a monitoring system, and vice versa.

[0040] As used herein, the expressions "and / or" or "or" when used in a sequence of two or more elements mean that each of the listed elements can be used alone or any combination of two or more of the listed elements can be used. Description of the Drawings

[0041] Other details of the present invention will be described below with reference to the drawings.

[0042] Figure 1 A schematic perspective view showing an additive manufacturing facility including an image processing device according to the present invention is shown.

[0043] Figure 2 A simplified schematic flowchart showing the method steps according to the present invention is shown.

[0044] Figures 3 to 6 Respectively, according to the histogram, an image processing-based, creative quantitative determination of process smoke in an additive manufacturing method is shown.

[0045] Figure 7 An exemplary time-varying curve of the determined amount of process smoke is also shown. Detailed Description of the Invention

[0046] In the embodiments and the drawings, the same or functionally equivalent elements may be provided with the same reference numerals, respectively. The sizes of the elements shown and their relationships to each other are not to be considered as being to scale in principle. Rather, for better visibility and / or for better understanding, individual elements may be shown exaggerated in size.

[0047] Figure 1 An additive manufacturing facility 100 is shown. The manufacturing facility 100 is preferably designed as an LPBF facility and is used for additively constructing a component 10 or a part from a powder bed. The facility 100 may particularly also be a facility for electron beam melting.

[0048] Thus, the facility has a construction platform on which the component geometry is manufactured, and thus welded. The component 10 is manufactured layer by layer from powder or a powder bed 1. For this purpose, the powder is distributed layer by layer on the construction platform via a cladding device not further specified. After each powder layer is coated, the areas of the layer are selectively melted and then solidified by an irradiation device or radiation source 2 using an energy beam, such as a laser or an electron beam 3, according to a preset geometry of the component 10.

[0049] After each layer, the construction platform preferably lowers by an amount corresponding to the layer thickness. This thickness is typically only between 20 μm and 40 μm, such that the entire process can easily include selective irradiation of thousands to tens of thousands of layers. Here, due to the very locally acting energy input, high temperature gradients of, for example, 10 6 K / s or greater can occur. Of course, during and after construction, the stress state of the component is correspondingly high as well, which makes the additive manufacturing process quite complex.

[0050] The geometry of the component is typically specified by a CAD file ("Computer-Aided-Design"). After such a file is read into the manufacturing facility 100, the additive process then first needs to determine a suitable irradiation strategy, for example with the aid of CAM ("Computer-Aided-Manufacturing"), whereby the component geometry can also be divided into individual layers.

[0051] As further elaborated in the description according to Figure 2 (see further below), the method for image processing according to the invention preferably relates to determining process disturbances in powder bed-based additive manufacturing, wherein the image B is preferably recorded layer by layer during the manufacturing process by a recording device 20, for example as part of an image processing device 30, in particular in order to determine and / or quantify the generation of process fumes S as a process disturbance.

[0052] The mentioned recording device 20 can record the image B, for example, via suitable measures for image detection or process monitoring, such as in particular camera recordings, optical tomography, CCD cameras or other image sensors.

[0053] For example, the image processing device 30 can be further configured to determine the process disturbance S via or by or using a computer program or a computer program product. Furthermore, the determined process disturbance (see below) or the overall disturbance or even the time course of the mentioned overall disturbance can be part or all of a computer program product generated by a corresponding computer program.

[0054] Figure 1 A monitoring system 40 for additive manufacturing is also shown, which monitoring system includes an image processing device 30.

[0055] Figure 2 The main method steps according to the invention are shown in a schematically simplified flow chart. Thus, the invention relates to a method for image processing for determining a process disturbance S, the method comprising: (i) recording a plurality of images B during the operation or manufacturing process.

[0056] In addition, the method includes: (ii) classifying image B as an image recorded before, during, and / or after a process event that causes interference S. The process event should be understood as an event that causes an actual interference; in an additive manufacturing method, the event thus involves irradiation by means of a laser or an electron beam along a trajectory and vectors preset, for example, by CAM, to selectively solidify the structure layer by layer according to a preset CAD component geometry.

[0057] The image classification can preferably include forming the time average values of measurement variables such as brightness or gray level of image B recorded before and after the process event, respectively.

[0058] First, the recorded images are preferably divided into individual layer images according to the layer sequence during the manufacturing process. In addition - as described - the division is made according to the recording time points. If the recording time point is, for example, before or after the exposure of the corresponding layer or its irradiation vector, the image is used for calculating the background and image noise. Conversely, if the image is recorded during the exposure, the image serves as a basis for determining the amount of interference caused by smoke.

[0059] As described, the background or image noise can preferably be calculated pixel by pixel. For the mentioned calculation, the smallest difference between the measurement variables of the corresponding image B recorded during the process event and the recordings before and after the process event is preferably used for calculating the background.

[0060] In other words, for each image in which the build platform is exposed or imaged, the image background is preferably calculated individually. Here, the background consists of the images recorded before and after the exposure. For this purpose, the image recorded during the exposure is compared with the images recorded before and after the exposure pixel by pixel. If a pixel has a smaller difference from the pixel in the image that is part of the image after the exposure, then the pixel in the background also takes its value. If the difference is smaller than the pixel in the image recorded before the exposure, the pixel in the background with the smaller difference takes the value. Thereby, a background image is advantageously generated, which shows the partially completed exposed component, but does not show the melt pool and process smoke.

[0061] By comparing the exposed image with the images recorded before and after the exposure and by using the correspondingly smaller differences, more purposeful results are provided in principle, and in particular, the light and dark contrast generated by the structure that has been solidified layer by layer is minimized or calculated.

[0062] The method further includes: (iii) calculating the background and / or image noise of each image recording B during the process event by comparing each image recording B during the process event with other image recordings before and after the process event.

[0063] The method further comprises: (iv) calculating an overall G of the interference S based on the image record B from which the calculated background has been cleared during the process event.

[0064] In particular, the amount of smoke can be determined for each image by means of an evaluation algorithm applied to the recorded images. Here, the measured data - calculated or background-corrected by the described background - is advantageously independent of the image noise generated in the corresponding image sensor.

[0065] By means of the described solution, the present invention makes a significant contribution to significantly improving process monitoring and even optimizing the parameter sets to be developed in relation to materials and design.

[0066] Preferably, a plurality of images B are finally recorded in the form of an image sensor, and the corresponding gray level or black-and-white contrast of the images is detected as a measurement variable.

[0067] The brightness value or gray level of the pixels in the recorded image is actually composed of different influences. Most pixels absorb - in this case - the light reflected by the powder bed 1. Once there is process smoke S above the powder bed 1, the reflected light is deflected or scattered, and the brightness value decreases. However, if the powder bed 1 is exposed to the laser beam 3, the brightness value in the molten pool increases. In addition, for each pixel in the image recording, there is an inherent measurement deviation due to image noise.

[0068] The images recorded just before and after the exposure usually show the construction platform or the manufacturing surface for a few seconds (see Figure 1 ). The images basically depict similar or identical scenarios, but may be slightly different due to image noise. However, the noise contrast is weak enough so that it is assumed that the average values of the pixels are approximately constant respectively, and the corresponding noise function is normally distributed accurately enough. Therefore, the time average of the corresponding images is first calculated.

[0069] As already shown above, preferably the difference between the pixels of the average image and other or additional images is calculated, and then the difference is shown in a histogram (see Figures 3 to 6 below). The histogram shown correspondingly also represents or includes the image noise. In this sense, only the overall G of the interference can be calculated and / or displayed during the process event via the histogram.

[0070] Figure 3 The relative frequency of the image noise plotted according to the respective difference levels of the pixel measurement values (such as brightness values, gray values or gray levels) of the image sensor is shown according to a simple histogram.

[0071] Figure 4 - Similar to Figure 3—— showing a histogram of the total signal of an image sensor type, i.e., a histogram of the process noise including image noise that has been recorded.

[0072] Figure 5 The difference between the "overall signal" and the background or background signal H is shown in a similar diagram. In other words, the calculated background image is subtracted from the corresponding image recorded during the exposure. Then a corresponding histogram is formed from the resulting image difference. The histogram consists of the influence. As is well known, the (backscattered) laser only has values at the upper end or the right part of the histogram.

[0073] According to Figure 6 It can also be seen from the diagram that the right part is thus cut off and thus set to zero. The right side of the histogram, thus the positive side, is therefore only affected by the image noise of the background (see above). Now the factor between the two distributions can be determined via a numerical optimizer such that the two histograms are as close to each other as possible in the positive region. In this way, the determined interference or the overall overexposure error thereof is approximately eliminated. Instead of the mentioned numerical optimizer, for example, another suitable fitting algorithm or suitable regression analysis or compensation calculation can be used.

[0074] A distribution is formed by the described subtraction of the individual histograms, and this distribution is advantageously only affected by the deflection of the process smoke.

[0075] In Figures 3 to 6 the high value of the relative frequency at zero difference in the individual histograms reflects the fact that the gray values or pixel luminances that exceed the "limit" of the measurement accuracy of the corresponding image sensor or the recording device of the sensor must be cut off or considered scaled respectively.

[0076] The amount of the interference overall G or the process smoke S can then be formed via the sum of the products of each difference and the corresponding frequency. Thereby, the amount of smoke on the corresponding image is determined independently of the image noise of the background. In the current case of the histogram representation, the determination of the amount of smoke is particularly carried out by the sum of the relative frequencies on each classification or difference bar.

[0077] Figure 7 Only abstractly and exemplarily is the time variation curve G(t) of the determined amount of process smoke S shown with respect to time. Since this step is an optional step that is not necessarily important for the present invention, this step is only provided with a dotted connecting line in the Figure 2 flowchart.

[0078] Thereby, it is also possible to advantageously obtain the time variation curve of the amount of process smoke on the construction platform. In addition, this can generally be used for the comparison of different parameters or process monitoring.

[0079] Without limiting generality, the present imaging method for determining process interference can alternatively relate to other methods, such as laser-based, standardized or automated operating methods.

Claims

1. An image processing method for determining the amount of process smoke as a process disturbance in additive manufacturing, the method comprising: -(i) recording a plurality of images (B) during a run or manufacturing process, -(ii) classifying the images (B) as images recorded before, during, and / or after a process event that causes the disturbance (S), wherein during the run or manufacturing process, the images (B) are additionally classified in each repeated process step, and wherein the repeated process steps involve manufacturing respective layers during the manufacturing process, -(iii) calculating the background of each image recording (B) during the process event by comparing each image recording (B) during the process event with other image recordings before or after the process event, and -(iv) calculating an overall amount (G) of the disturbance (S) based on the image recordings (B) during the process event from which the calculated background has been removed.

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

3. The method according to any one of the preceding claims, wherein in order to calculate the background, the smallest difference between the measured variables of the respective images (B) recorded during the process event and the recordings before and after the process event is respectively used for calculating the background.

4. The method according to any one of the preceding claims, wherein the plurality of images (B) are recorded in the form of an image sensor, and the gray level of the images is detected as the measured variable.

5. The method according to any one of the preceding claims, wherein the image classification respectively includes: Form a time average of the measured variables of the images (B) recorded before and after the process event.

6. The method according to any one of the above claims, the method comprising: Calculate a time variation curve (G(t), v) of the overall amount (G) of the disturbance (S) during the process event.

7. The method according to any one of the preceding claims, wherein the overall amount (G) of the disturbance (S) during the process event is calculated via a histogram.

8. An image processing device (30), the image processing device being configured to perform the image processing method according to any one of the preceding claims to determine a disturbance (S).

9. A monitoring system (40) for additive manufacturing, the monitoring system comprising the image processing device (30) according to the previous claim.

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