DE102021209868B3 - Method for determining a density distribution of printing material within a green body and device for producing a green body by means of an additive manufacturing process - Google Patents

Abstract

The invention relates to a method for determining a density distribution (14) of printing material (1) within a green body (2) with a plurality of layers (Ni). First, an image (12) is provided for each layer (Ni) of the multiple layers (Ni) of the green body (2). An optical parameter (10, 11) relating to the print material (1) is determined using the image (12) using a filter (9). A local density distribution (13) of the printing material (1) is calculated in the respective layer (Ni) as a function of the determined optical parameters (10, 11). The density distribution (14) of the printing material (1) within the green body (2) is determined on the basis of the respective local density distribution (13) of the respective layer (Ni).

Description

The invention relates to a method for determining a density distribution of printing material within a green body. The invention also relates to a device for producing a green body using an additive manufacturing method. Various parts or components can now be manufactured using an additive manufacturing process. So-called 3D printers are also used in the private sector as an additive manufacturing process. The product immediately after a generative manufacturing process or after the printing process of a 3D printer has been completed can be referred to as a green body or green body. The green body or the green body represents in particular a blank that is usually further processed into an end product. This further processing can be, for example, thermal treatment or sintering.

During this post-treatment, irregular distributions of particles within the green body can have a significant impact on subsequent processing. For example, inhomogeneities in the particle distributions within green compacts play a major role in the production of dimensionally stable and flawless components in a so-called “metal binder jetting” process. Due to their low strength, the green compacts cannot easily be examined by conventional metallographic analyzes to determine the density distribution. On the other hand, examination of many alloys using non-destructive testing, such as computed tomography, may also not be possible.

In particular, there is the task of determining information relating to green bodies in order to open up the possibility of optimizing subsequent further processing.

The pamphlet WO 2020/169286 A1 describes, for example, a method for the non-destructive detection of aging phenomena in a component having regularly recurring structures. In this way, several two-dimensional scan images can be generated, which can be analyzed with regard to a large number of inhomogeneities. Signs of aging can be detected from this.

The DE 10 2011 009 624 A1 describes a method for monitoring a generative manufacturing process in real time, with a component being recorded at least optically and the installation space thermally when the layer is applied, and a device for carrying out such a method.

The EP 2 942 130 A1 relates to a device for the additive manufacturing of at least one component area of a component, in particular a component of a turbine or a compressor. The device comprises at least one powder feed for applying at least one powder layer to a build-up and joining zone of a lowerable component platform and at least one radiation source for generating at least one high-energy beam, by means of which the powder layer in the build-up and joining zone can be locally and/or fused to form a component layer can be sintered. Furthermore, the device has a camera system, by means of which at least one stereoscopic recording can be generated for the three-dimensional detection of at least one region of the component layer.

The EP 3 878 583 A1 describes an additive manufacturing apparatus comprising an energy source for melting material to form a component in an additive manufacturing process, a camera aligned with the energy source to obtain image data of the melted material during the additive manufacturing process, and a controller to To control energy source during the additive manufacturing process in response to the processing of the image data. The controller adjusts the control of the energy source based on a correction that is determined by: applying an artificial intelligence model to image data captured by the camera during an additive manufacturing process, the image data including an image of a melt pool of the additive manufacturing process ; predicting a failure in the additive manufacturing process using an output of the artificial intelligence model; and compensating for the error by generating a correction to adjust a configuration of the energy source during the additive manufacturing process.

The invention is based on the finding that a density distribution of printing material within the green body can significantly influence further post-processing or post-treatment. For example, severe distortions can occur during tempering or sintering. Distortions can in particular be internal stresses in the green body caused by shrinkage processes. In the worst case, the green body can break apart during post-treatment, similar to ice cubes thrown into warm water. If the density distribution within the green body is known, it could be determined in advance whether the green body in question is the action will withstand. Alternatively, a generative manufacturing process for producing the green body can be improved or optimized with the aid of the density distribution.

Therefore, this invention proposes a method for determining a density distribution of print material within the green body. A first aspect of this invention relates to this method for determining the density distribution within the green body. The green body can be referred to as a green body. The green body preferably has several layers. In particular, the green body can be produced by a generative or additive manufacturing process. The term "generative" means in particular "layered". Such an additive or generative manufacturing method is in particular a manufacturing method which builds up or stacks the green body in layers. Printing the green body using a 3D printer represents a generative method, for example. First, an image can be provided for each layer of the multiple layers of the green body. The images can be captured using an optical capture unit or already exist as digital data. For example, a camera can take an image of each layer or a selection of multiple layers of the green body. Alternatively, the optical acquisition unit can capture multiple images for each slice of the multiple slices.

In a further step, an optical parameter relating to the print material can be determined on the basis of the image using a filter. The optical parameter can be a color distribution, brightness distribution and/or pattern distribution of the printing material. These distributions are preferably determined based on the image. Additionally or alternatively, an intensity distribution of the colors, brightness and/or pattern relating to the print material can be determined for the optical parameter by means of the filter. In particular, the filter can be an optical filter, such as a filter for edge or corner detection. All filters commonly used in image processing can be used in this process. Thus, the filter can recognize the color distribution, brightness distribution and/or pattern distribution of the print material from the image. The optical parameter can be determined by means of a computing unit. The computing unit can thus apply the filter to the image. Alternatively, multiple filters can be used to determine the optical parameter. In particular, each filter can be tuned or designed for a specific optical parameter, such as the color distribution or the pattern distribution. Alternatively, it is possible for the image to be pre-processed before the filter is applied.

In a further step, the local density distribution of the print material in the respective layer can be calculated or determined as a function of the determined optical parameter relating to the print material. The local density distribution of the printing material preferably relates to the respective layer. A separate optical parameter can be determined for each layer. The local density distribution can relate in particular to the respective layer. The density distribution of the printing material within the green body can be understood in particular as an overall or global density distribution of the printing material of the entire green body.

A density distribution over several layers, but not all layers of the green body, can be understood as a regional density distribution. The printing material can in particular contain a binder and metal particles. In particular, the binder can be a liquid adhesive. The metal particles can in particular be present in powder form. The metal particles can include aluminum, copper, steel, stainless steel, iron, gold, silver and/or lead. The metal particles can contain mixtures of the metals mentioned.

In a further step, the density distribution of the printing material within the green body can be determined using the respective local density distribution of the respective layer. In particular, a global density distribution of the printing material within the green body can be determined as a function of a plurality of local density distributions of the respective layers. In particular, the global density distribution can be determined using a number of local density distributions for layers that have already been produced. A separate local density distribution can thus be determined for each layer. The global density distribution of the printing material in the green body can be determined or ascertained from a plurality of local density distributions or from all local density distributions. The terms "calculate", "determine" and "determine" can be considered as synonyms. The density distribution refers to the print material of the green body or the print material in the green body.

In particular, the local density distribution represents a density distribution in a two-dimensional plane, the respective layer. A predetermined layer thickness can be taken as a basis for the local density distribution in order to obtain a value for the local density distribution. In particular, the density distribution of the printing material within the green body represents a three-dimensional density distribution within the green body and can be regarded as the “global density distribution” of the green body. The local density distribution is preferably two-dimensional, i.e. limited to the respective layer. The global density distribution is in particular three-dimensional and preferably includes the entire green body. The regional density distribution is preferably also three-dimensional, but in particular comprises only part of the green body.

This three-dimensional density distribution is calculated in particular as a function of a number of two-dimensional density distributions of the respective layers. This process can be referred to as "mapping to 3D". With this method, the density distribution of the printing material within the green body can already be determined during production of the green body. Thus, an important parameter, namely the density distribution, can already be obtained for further post-processing in a generative manufacturing process. This density distribution can preferably be taken into account in a thermal post-processing or a sintering process. For example, temperature curves or parameters of the sintering process can be adapted to the determined density distribution of the printing material within the green body. As a result, the post-processing of the green body can be designed more efficiently. In particular, the manufacturing process of the green body can be optimized to the extent that the density distribution of the printing material is as homogeneous as possible. This makes it easier to avoid later distortions or fractures during post-processing.

Provision is also made for the filter to detect edges and/or corners as the optical parameter in the respective layer and for the density distribution to be derived by means of the detected edges and/or corners. This can affect the local density distribution and/or the global density distribution. In particular, the local density distribution of the printing material in the respective layer can be determined as a function of the detected edges and/or corners. In this case, the filter is designed in particular as an edge filter and/or corner filter. With the help of the recorded edges and/or corners, different areas in the respective layer can be recognized. For example, phase boundaries between metal particles and the binder can be detected using corners or edges. For example, metal particles can be detected within the adhesive. Other particles or impurities can also be detected. Additionally or alternatively, several areas in a layer with different color distribution, brightness distribution and/or pattern distribution of the printing material can be detected. In this case the optical parameter refers to the color distribution, brightness distribution and/or pattern distribution of the print material. In addition or as an alternative to this, the optical parameter can relate to different intensity distributions of the colors, brightness and/or patterns. Thus, with the help of appropriate filters, differences with regard to the named distributions as well as differences in the named intensity distributions can be recorded or detected. In particular, the local density distribution can be calculated as a function of these detected differences. The density distribution of the printing material within the green body can be determined on the basis of several local density distributions. Based on said detected differences, it is possible to recognize and define a clear boundary between even and uneven distribution of the print material. The corresponding filters can use predefined threshold values for edge detection and/or corner detection.

An additional or alternative embodiment makes it possible for each layer of the green body to be produced by an additive manufacturing process and/or for each layer to be imaged using a camera. The camera preferably has a correspondingly high resolution of more than two megapixels, for example. In particular, after a relevant layer has been completed, the camera takes an image of this layer before the next layer is produced by the additive manufacturing method. In this way, the (global) density distribution of the green body that has already been produced can be determined simultaneously during the additive manufacturing process. The generative manufacturing process can preferably be adapted as a function of a density distribution that has already been determined. The generative manufacturing method is preferably adapted in such a way that an inhomogeneous density distribution is homogenized, ie a degree of inhomogeneity in the density distribution is reduced. As a result, mechanical properties of the green body can be improved. They can be sintered more easily in a sintering furnace without fear of distortion or damage. This can be achieved in particular by producing homogeneous green compacts. A largely homogeneous green body can be produced with the aid of the method for determining the density distribution of printing material within the green body and by taking this density distribution into account during the additive manufacturing process. The degree of homogeneity with regard to the density distribution of printing material within the green body influences in particular different component material properties, such as dimensional accuracy, strength and freedom from defects.

An additional or alternative embodiment allows multiple layers to be added are combined in a zone, a regional density distribution of the zone is determined by means of a moving average over the layers of the zone and the density distribution within the green body is determined based on one or more regional density distributions. In the additive manufacturing process, the binder, i.e. the liquid adhesive, can penetrate several layers. The liquid adhesive can, for example, seep into layers that have already been produced. It is also possible that the binder or liquid adhesive penetrates into an overlying layer. This can lead to the fact that the optical parameter with regard to the printing material in the image is not yet constant. In order to avoid an unnecessary delay in the production of the green body, the local density distribution is preferably determined based on the image, even if the optical parameter is not yet constant or is still changing. Any falsification of the optical parameter can be compensated for by averaging in the form of a moving average over several layers of the zone.

The regional density distribution thus represents in particular an averaging of several local density distributions of the respective layers. Any deviations in the density distributions in the respective layers can be compensated for by the moving average, which leads to the regional density distribution. In particular, the moving average may be determined along a print orientation of a 3D printer. For example, the print orientation may be vertically up. In this way, the density distribution can be meaningfully determined even if the binding material has not yet reached its final state. An unnecessary delay in the additive manufacturing process can thus be avoided.

An additional or alternative embodiment allows at least one threshold value for the one or more filters to be defined by a neural network. In particular, a threshold value for the detection of the edges and/or corners can be specified by the neural network. A trained neural network is preferably used here. Additionally or alternatively, the neural network can learn to define the at least one threshold value on the basis of a plurality of images and/or be trained on the basis of the plurality of images in order to define the at least one threshold value. When training the neural network, machine learning algorithms (machine learning) can be used. Using machine learning, the neural network can be optimized with regard to the definition of the respective threshold values. As a result, the determination or detection of the optical parameter can be optimized and/or accelerated. Determining the density distribution of the printing material within the green body can thus be accelerated or improved.

An additional or alternative embodiment allows the local density distribution and/or the density distribution within the green body to be additionally dependent on a metal in the printing material, a particle shape of the metal, a particle size distribution of the metal, a nozzle shape of a 3D printer, a spray system of the 3D -Printer, a speed of a print head of the 3D printer when generating the respective layer and/or a state of the print head is determined. This embodiment preferably relates to an additive manufacturing method, in particular a 3D printer. A printing process of the 3D printer depends in particular on different factors, such as the powder form, the particle size distribution, the spray system, the print head speed, a print head status, etc. These factors mentioned can be taken into account when determining the density distribution. For example, based on the speed of the print head or the shape of the nozzle, a distribution of print material that has not yet solidified in the green body can be predicted. Since the liquid adhesive can still flow a little inside the green body, there can still be slight differences in the density distribution. With the help of the aforementioned parameters, such as the particle size distribution of the metal, the nozzle shape of the 3D printer, the spray system of the 3D printer and/or the speed of the print head of the 3D printer, a flow behavior of the liquid adhesive or the printing material within the green body be forecast. This can be used to determine the extent to which the printing material will still flow in layers that have already been created. In this way, a subsequent change in the density distribution in layers of the green body that have already been produced can be determined and taken into account when determining the density distribution of the green body.

The factors mentioned above, such as the particle shape of the metal or the speed of the print head of the 3D printer, can already be taken into account when calculating the local density distribution of the printing material in the respective layer. The local density distribution of the printing material in the respective layer can be expressed, for example, on the basis of different colors, patterns, brightnesses and/or intensities thereof. Based on these differences, a clear boundary between even and uneven distribution can be defined. With the help of these additional factors, the calculation of the local density distribution can be improved or made more precise.

An additional or alternative embodiment makes it possible for further layers of the green body to be produced as a function of the determined local density distribution of a predetermined number of layers. This embodiment relates to a green body that is in the manufacturing process, ie has not yet been completely manufactured.

As a result, local density distributions that have already been determined can be taken into account in order to adapt subsequent layers in such a way that the most homogeneous possible density distribution of the green body is achieved. Flow processes of the printing material can also be taken into account here. The advantages and examples already mentioned apply mutatis mutandis and analogously to this embodiment.

An additional or alternative embodiment allows the green body to be post-treated, in particular sintered and/or heat-treated, depending on the density distribution to form a component. The post-treatment of the green body is therefore preferably based on the determined density distribution of the printing material within the green body. Additionally or alternatively, a position and a pose of the component can be determined using digital data. The position or pose is preferably also taken into account in the follow-up treatment. A corresponding position of all components of the green body can be known with the aid of CAD data. The post-treatment can be sintering in a sintering furnace and/or thermal post-treatment in a furnace. The green compact can be hardened in a thermal post-treatment. In particular, detected inhomogeneities in the density distribution can be taken into account during the post-treatment. For example, areas with a higher degree of inhomogeneity can be subjected to a shorter or less intensive post-treatment. For example, areas of the green body with a higher degree of inhomogeneity in the density distribution can be exposed to a lower temperature and/or a shorter exposure time. Based on the determined density distribution, information regarding inhomogeneities can be derived, which can be used efficiently in the after-treatment of the green body.

A second aspect of this invention relates to a device for a green body using an additive manufacturing method. This device can be a 3D printer, for example. In particular, the device has printing material in the form of a binder with metal particles. The printing material thus preferably includes the binder and a plurality of metal particles. The binder is preferably designed as a liquid adhesive. The device has a powder bed, on which the green body can be produced in layers. In the manufacturing process, the green body can be produced layer by layer on the powder bed. For this purpose, the printing material can be layered on the powder bed. The device has an optical acquisition unit to acquire an image for each layer of the green body. The optical detection unit is preferably a camera. In addition, the device has a computing unit. The computing unit is designed in particular to execute each described method and each described embodiment. In particular, the computing unit can determine an optical parameter of the print material based on the image using a filter. The filter can be part of the computing unit. The computing unit can calculate or determine a local density distribution of the printing material in the respective layer as a function of the determined optical parameter. Furthermore, the computing unit can determine the (global) density distribution of the printing material within the green body based on the respective local density distribution of the respective layer.

The features, embodiments and advantages presented in connection with the method for determining the density distribution according to the first aspect of the invention apply correspondingly to the device according to the second aspect of the invention and vice versa. Device features can thus be interpreted as process features and vice versa.

The device can contain the computing unit. The processing unit can have a processor device that is set up to carry out an embodiment of a method. For this purpose, the processor device can have at least one microprocessor, at least one microcontroller, at least one FPGA (Field Programmable Gate Array), at least one DSP (Digital Signal Processor) and/or a neural network. Furthermore, the processor device can have program code which is set up to carry out the embodiment of the method when executed by the processor device. The program code can be stored in a data memory of the processor device. The computing unit can include an internal or external storage unit. The external storage unit can be in the form of a cloud unit.

The device may comprise a computer program [product] containing instructions that cause each embodiment of the method to be carried out. The computer program product can be stored on a computer-readable medium.

The invention also includes developments of the method according to the invention, which have features as have already been described in connection with the developments of the motor vehicle according to the invention. For this reason, the corresponding developments of the method according to the invention are not described again here.

The invention also includes the combinations of features of the described embodiments.

• 1: eine schematische Darstellung einer Vorrichtung zum Erzeugen eines Grünkörpers oder Grünlings in Form eines 3D-Druckers;
• 2: eine schematische Darstellung eines beispielhaften Verfahrens;
• 3: eine schematische Darstellung eines Beispiels zum Bestimmen der Dichteverteilung anhand zweier ausgewählter Schichten des Grünlings.

The invention will now be explained in more detail using the attached drawings as examples. The drawings indicate possible exemplary embodiments for the invention without restricting the scope of the invention. All of the features described in the exemplary embodiments and embodiments can be extracted separately and correspondingly combined with other embodiments and examples. It shows:

• 1 1: a schematic representation of a device for producing a green body or green body in the form of a 3D printer;
• 2 : a schematic representation of an exemplary method;
• 3 : a schematic representation of an example for determining the density distribution using two selected layers of the green body.

The exemplary embodiments explained below are preferred exemplary embodiments of the invention. In the exemplary embodiments, the described components each represent individual features of the invention that are to be considered independently of one another, which also develop the invention independently of one another and are therefore also to be regarded as part of the invention individually or in a combination other than that shown. Furthermore, the exemplary embodiments described can also be supplemented by further features of the invention already described. Elements with the same function are each provided with the same reference symbols in the figures.

1 10 shows a device 100 for producing a green body 2 or green body 2. In ceramics, in metalworking and/or in the production of sintered workpieces, the green body 2 is often referred to as an unfired or unsintered blank. The green body 2 can usually be easily further processed. The device 100 can preferably produce the green body 2 by means of a generative manufacturing method or an additive manufacturing method. In the case of 1 the device 100 is designed as a 3D printer 100 . The 3D printer has a camera 3 as an optical detection unit 3 , a powder bed 4 , a print head 6 , a nozzle 7 of the print head 6 , a computing unit 5 and a control panel 8 . The green compact 2 is manufactured using printing material 1 . The printing material 1 can have binding material and metal particles. Liquid adhesive, for example, can be used as the binding material. The metal particles can be in powder form. Examples of possible metals are copper, stainless steel, aluminum, lithium, iron and mixtures of these metals.

In 1 the green compact 2 is shown in dashed lines. The dashed lines are intended to represent different layers N _i . In this case, the index i identifies in particular a code number of the respective layer. For example, the green body 2 can have more than 1000 layers. In this context, N500 would mean the 500th shift. The multiple layers N _i are preferably deposited one on top of the other. The green body 2 is thus preferably produced in layers. The powder bed 4 can be preheated or heated to facilitate the production of the green body 2 . The computing unit 5 can control the camera 3 and the print head 6 and thus record images of the layers N _i and influence a speed of the print head 6 . The most up-to-date, uppermost layer N _i is preferably recorded optically in each case. In particular, each layer N _i can be recorded with the camera 3 . The computing unit 5 can also contain a filter 9 which can analyze the images 12 of the layers N _i captured by the camera 3 with regard to an optical parameter 10 , 11 . The camera 3 can capture images 12 of a surface of the layers N _i .

The 3D printer 100 can be used for a metal binder jetting method. The printing material 1 can be printed onto the powder bed 4 in order to connect the metal particles in the printing material 1 in layers. The green body 2 can be produced in this way. The green body 2 can have high porosity and poor material properties if it is not subsequently treated further. For this purpose, the green compacts 2 can be sintered in a sintering furnace. These mechanical properties of the green body 2 can be improved by this post-treatment. In particular, the (global) density distribution 14 or an inhomogeneity distribution of the density can impair the mechanical properties of the green body 2 .

In 2 a possible method for determining or determining the density distribution within the green body 2 is shown as an example. In a first step S1, an image 12 for each layer N _i of the green body 2 can be provided. To this end, in an optional second step S2 object deposited on the powder bed 4 can be detected. It can thus be ensured that the camera 3 records an image of the green body 2 or an image of a layer N _i of the green body 2 . In a third step, an optical parameter 10, 11 relating to the print material 1 can be determined using the image 12 by means of the filter 9. For this purpose, image filtering is preferably carried out in the third step S3. The filter 9 can preferably detect corners 10 and/or edges 11. As part of this filtering, threshold values can be used, which can be defined by a neural network. The arithmetic unit 5 can thus have a neural network or itself be designed as a neural network. As a result, edges 11 and/or corners 10 can be detected in the image 12 in a fourth step S4. The image 12 is preferably converted into a modified image 12′ by the image filtering in the third step S3 and the detection of the corners 10 and/or edges 11 in the fourth step S4. The modified image 12′ preferably shows the corners 10 and/or edges 11 detected by the filter 9 in order to determine a local density distribution 13 therefrom.

In a fifth step S5, the multiple local density distributions 13 can be combined. In this case, the respective layers N _i can be assigned to respective regions of the green body 2 . Each layer N _i preferably represents a corresponding extension of the green body 2 in the respective plane of the layer N _i . The layers N _i are preferably in the form of planar cross sections. These cross sections can be aligned horizontally and/or parallel to a base area of the powder bed 4 . In a sixth step S6, a global density distribution 14 of the printing material 1 within the green body 2 can be determined by averaging over all local density distributions 13. A global density distribution 14 can thus be determined.

In 3 shows an example of how two different local density distributions 13 can be determined using two images 12 . In the upper left area of 3 a plan view of the top layer N _i is shown. A circled area is shown enlarged below this illustration. This enlarged view shows the cuboid section of the green body 2. Two diamond-shaped layers N _i are indicated within this cuboid section. This is the 85th layer N85 and the 210th layer N210. The camera 3 preferably takes at least one image 12 of this layer N i during or after the end of the production method with regard to a respective layer N _i . Accordingly, there is at least one image for layers N85 and N210. The processing unit 5 can subject these images 12 to image pre-processing. For example, the images can be transformed from one camera format into another format that is suitable for the processing unit 5 .

The images 12 can be subjected to further image processing by the computing unit 5 and/or by the neural network. The filter 9 is preferably used here. Several filters 9 can also be used. Each of these filters 9 can be specially designed to detect a specific optical parameter 10, 11. The optical parameter 10, 11 can be a color distribution, a brightness distribution, a pattern distribution and a distribution of corners 10 and/or edges 11. The optical parameter 10, 11 can also include an intensity distribution of colors, brightnesses, patterns, edges, corners and/or lines. The image 12 can be converted into the modified image 12' by the filtering of the image 12 by the processing unit 5 by means of the filter 9. The modified image 12' preferably shows a structure which can be analyzed with regard to the optical parameters 10, 11. The computing unit 5 can determine the local density distribution 13 on the basis of this modified image 12 ′. The first local density distribution 13 shows an outline within the layer N210 in the upper left area, for example. In this area, for example, less printing material 1 can be present than in the remaining area of the relevant layer N210.

For example, the lower local density distribution 13 shows that a trapezoidal area extending from a lower edge of the layer N85 contains no printing material 1. The local density distributions 13 shown here are examples of possible local density distributions 13. Other local density distributions 13 are correspondingly conceivable. The computing unit 5 can determine the corresponding local density distribution 13 for each individual layer N _i of the green body 2 . For this purpose, the respective images 12 are converted into the modified images 12' by means of the filters 9. The modified images 12' can be determined by the computing unit 5 using specially adapted filters 9, the respective optical parameters 10, 11 and derive the local density distribution 13 therefrom.

In particular, the arithmetic unit 5 can skip the generation of the modified image 12 ′ and detect the optical parameters 10 , 11 directly on the basis of the image 12 with the aid of the filters 9 and determine the local density distribution 13 therefrom. In particular, the computing unit 5 can recognize corners 10 and edges 11 between a uniform and a non-uniform distribution on the basis of different colors, patterns, brightnesses and intensities. The arithmetic unit 5 can thus use different zones within the respective layer N _i recognize. For this purpose, the processing unit 5 or the neural network can apply and/or calibrate threshold values for the detection of the edges or corners. The computing unit 5 or the neural network can use predefined images 12 to learn to define the threshold values. Machine learning algorithms, i.e. machine learning, for example through the neural network, can be used for this. The global density distribution 14 can be determined on the basis of the plurality of local density distributions 13 . For this purpose, the multiple layers N _i can be aligned with one another and the multiple local density distributions 13 can be converted into the global density distribution 14 with the aid of averaging. A subset of several layers N _i can be defined for a regional zone. In addition, a regional density distribution for this regional zone can be determined. This can make it easier to determine the local density distribution 13 .

The global density distribution 14 can be used to post-treat the green compact 2 more effectively. Curing in a heating oven and/or sintering in a sintering oven can thus be better adapted based on the global density distribution 14 . In this way, mechanical stresses within the green compact 2 and/or damage to the green compact 2 can be better avoided. In this way, better quality assurance with regard to the green compacts 2 can be achieved at the same time. The computing unit 5 can also take into account other parameters such as the type of metal, the particle shape of the metal, a particle size distribution of the metal, the shape of the nozzle 7, a speed of the print head 6 and a spray system of the 3D printer 100 when determining the density distribution. The computing unit 5 can use these parameters to determine or predict a flow behavior of the printing material 1 within layers N _i that have already been produced. Based on the determination of this flow behavior, the generation of further layers N _i can be adapted accordingly to the determined flow behavior of the printing material 1 in order to achieve a density distribution that is as homogeneous as possible.

The computing unit 5 can carry out a correlation analysis between the optical parameters and the local or global density distribution 13, 14. For example, a correlation between the speed of the print head 6 and a degree of homogeneity with respect to the global density distribution 14 can be determined. The computing unit 5 can thus set up or determine a large number of correlations between one or more parameters and the local or global density distribution. By means of these correlation analyses, the green compact 2 can be produced with the highest possible degree of homogeneity, that is to say the lowest possible degree of inhomogeneity.

In particular, the global density distribution 14 or the local density distribution 13 can be determined during additive manufacturing. The density distribution can thus be determined online during the printing process and at the same time taken into account during the printing process in order to produce the green body 2 as homogeneously as possible. The well-known disadvantages of after-treatment, such as mechanical stresses or fractures, can be reduced or at least reduced in this way.

Reference List

1

Printing material (binder, metal powder)

2

Green compact, green body

3

optical detection unit, camera

4

powder bed

5

unit of account

6

printhead

7

nozzle shape

8th

control panel

9

filter

S1

first step

S2

second step

S3

Third step

S4

fourth step

S5

fifth step

S6

sixth step

10

Corners, optical parameter

11

edges, optical parameter

12

Picture

12'

modified image

13

local density distribution

14

Density distribution, global density distribution

no

respective layer(s)

N210

210th layer

N85

85th layer

Claims (9)

Method for determining a density distribution (14) of printing material (1) within a green body (2) with a plurality of layers (N _i ) , characterized by the steps: - providing an image (12) for each layer (N _i ) of the plurality of layers (N _i ) of the green body (2), - determining an optical parameter (10, 11) relating to the printing material (1) using the image (12). a filter (9), - calculating a local density distribution (13) of the printing material (1) in the respective layer (N _i ) as a function of the determined optical parameters (10, 11) relating to the printing material (1), - determining the density distribution (14) of the printing material (1) within the green body (2) based on the respective local density distribution (13) of the respective layer, - whereby in the respective layer (N _i ) edges (11) and/or corners (10) are detected as the optical parameter (10, 11) and the density distribution (14) is derived by means of the detected edges (11) and/or corners (10), and/or several areas in a layer (N _i ) with different color distribution, Brightness distribution and/or pattern distribution of the Dr uckmaterials (1) and/or different intensity distributions of the colors, brightness and/or patterns are detected as the optical parameters (10, 11) and the local density distribution (13) is calculated as a function of these detected differences. procedure after claim 1 , wherein each layer (N _i ) of the green body (2) is produced by an additive manufacturing process and/or an image (12) is captured by a camera (3) of each layer (N _i ). Method according to one of the preceding claims, wherein several layers (N _i ) are combined into one zone, a regional density distribution of the zone is determined by means of a moving average over the layers (N _i ) of the zone and the density distribution is based on one or more regional density distributions (14) is determined within the green body (2). Method according to one of the preceding claims, wherein at least one threshold value for the one or more filters (9) is defined by a neural network, and in particular a threshold value for the detection of the edges (11) and/or corners (10) by the neural network is determined. procedure after claim 4 , the neural network learning to define the at least one threshold value using a plurality of images (12) and/or being trained to define the at least one threshold value using a plurality of images (12). Method according to one of the preceding claims, wherein the local density distribution (13) and/or the density distribution (14) within the green body also depends on a metal in the printing material (1), a particle shape of the metal, a particle size distribution of the metal, a nozzle shape (7) a 3D printer, a spray system of the 3D printer, a speed of a print head (6) of the 3D printer (100) when generating the respective layer (N _i ) and/or a state of the print head (6). . Method according to one of the preceding claims, further layers (N _i ) of the green body (2) being produced as a function of the determined local density distribution (13) of a predetermined number of layers (N _i ). Method according to one of the preceding claims, in which the green body (2) is post-treated, in particular sintered, as a function of the density distribution (14) to form a component, and/or a position of the component is determined using digital data, the position also being used in the post-treatment is taken into account. Device (100) for producing a green body (2) by means of an additive manufacturing process, comprising: - a printing material (1), preferably with a binder and with metal particles, - a powder bed (4), on which the green body (2) can be produced in layers, - an optical acquisition unit (3) for acquiring an image (12) for each layer (N _i ) of the green body (2), - a computing unit (5) which is set up o an optical parameter (10, 11) of the printing material (1 ) to be determined based on the image (12) using a filter (9), o to calculate a local density distribution (13) of the printing material (1) in the respective layer (N _i ) as a function of the determined optical parameters (10, 11). or to determine o the density distribution (14) of the printing material (1) within the green body (2) using the respective local density distribution (13) of the respective layer (N _i ) to determine o in the respective layer (N _i ) through the filter edges (11) and/or corners (10) as the optical parameter (10, 11) and to derive the density distribution (14) by means of the detected edges (11) and/or corners (10), and/or several areas in a layer (N _i ) with different color distribution, brightness distribution and/or pattern distribution of the printing material (1), and/or different intensity distributions of the colors, brightness and/or pattern as the optical parameter (10, 11) and as a function of these detected differences to calculate the local density distribution (13).

Images