Method for determining material properties from foam samples

Abstract

To provide methods and control systems for determining material properties from foam samples.SOLUTION: A computer-implemented method for determining a material property of a foam sample comprises: providing a representation of the sample; extracting at least one structural feature from the representation, where the at least one structural feature comprises walls, struts, or nodes; providing the at least one structural feature to a material model suitable for obtaining at least one material property from the structural feature; and outputting the at least one material property received from the material model.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] explanation The present invention is in the field of methods for determining material properties from foam samples, and in particular from images of foam samples. [Background technology]

[0002] Material properties, such as mechanical, thermal, or chemical properties, strongly depend on the structure of the material, both at the macroscopic and microscopic levels, especially for porous materials such as foams. Determining material properties usually requires laborious testing: samples must be prepared, appropriate test equipment must be provided, and test protocols must be strictly followed. This requires well-trained personnel. Therefore, it is desirable to provide a faster method involving fewer personnel and less equipment. Summary of the Invention [Problem to be solved by the invention]

[0003] WO2015 / 080912A1 discloses a method for digitally modeling oil field reservoirs from computed tomography images of core samples. These models can be used to obtain oil field characteristics by performing simulations. However, the models require many specific assumptions and physics to be incorporated, making them largely inapplicable to fields other than oil field analysis.

[0004] WO2018 / 206225A1 discloses a method for modeling an object from an image and comparing it to its desired shape to detect defects, but does not provide material properties.

[0005] US Patent Application Publication No. 2014 / 044315 A1 discloses a method for improving the accuracy of target property values ​​obtained from rock samples, but this method is largely not transferable to foam samples.

[0006] Samuel Pardo Alonso, in his doctoral thesis entitled "X-ray imaging applied to the characterization of the cellular structure and its evolution of polymer foams" from March 1, 2014, discloses a method to generate 3D models from images, but does not provide material properties.

[0007] Therefore, the objective of the present invention was to provide a method for determining material properties with little effort in terms of equipment, personnel, and time. The method would need to be adaptable to a variety of different foam materials and size scales. The method was intended to be fast and reliable. [Means for solving the problem]

[0008] These objectives were achieved by a computer-implemented method for determining material properties of a foam sample, including:

[0009] (a) providing a sample representation; and (b) extracting at least one structural feature from the representation, the at least one structural feature including a wall, a support, or a node; (c) providing the at least one structural feature to a material model suitable for obtaining at least one material property from the structural feature; (d) outputting at least one material property received from the material model. .

[0010] The invention further relates to a non-transitory computer-readable data medium storing a computer program comprising instructions for performing the steps of the method according to any one of the preceding claims.

[0011] The present invention further relates to a production monitoring and / or control system for monitoring and / or controlling material properties of a sample, comprising:

[0012] (a) an input unit configured to receive a representation of the samples; (b) a processing unit configured to extract at least one structural feature from the representation, the at least one structural feature comprising a wall, a support, or a node; and (c) a processing unit configured to provide the at least one structural feature to a material model suitable for obtaining at least one material property from the structural feature; (d) an output unit configured to output the material properties received from the material model.

[0013] Preferred embodiments of the invention can be found in the description and claims. Combinations of the different embodiments are within the scope of the invention. [Brief explanation of the drawings]

[0014] [Figure 1] 1 illustrates a possible implementation of the present invention. [Figure 2a] 1 shows an example of image processing using the method of the present invention. [Figure 2b] 1 shows an example of image processing using the method of the present invention. [Figure 2c] 1 shows an example of image processing using the method of the present invention. [Figure 2d] 1 shows an example of image processing using the method of the present invention. [Figure 2e] 1 shows an example of image processing using the method of the present invention. [Figure 2f] 1 shows an example of image processing using the method of the present invention. [Figure 2g] 1 shows an example of image processing using the method of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0015] The method according to the present invention is useful for determining the material properties of foam samples. The sample can be a small piece of material or a complete workpiece. Foam samples include porous materials such as polymer foams, zeolites, and supports for exhaust catalysts. The internal structure of a sample refers to the distribution of phase boundaries within the sample. For example, the interstitial spaces and interpolymer phase boundaries within a polymer foam. The internal structure features can be on various size scales, such as the microrange (i.e., 0.1-1000 μm) or nanoscale (i.e., 1-100 nm), depending on the sample and image resolution.

[0016] The material properties can be mechanical properties such as Young's modulus, elasticity, tear resistance, abrasion resistance, coefficient of friction, etc.; thermal properties such as heat capacity, thermal conductivity, etc.; electrical properties such as electrical conductivity, resistivity, dielectric constant, etc.; or optical properties such as transparency, refractive index, diffusivity, etc.

[0017] A representation of a sample in the context of the present invention is a data structure that contains the internal structure of the sample. The representation associates each location in the sample information with the material or void at that location. The representation can be two-dimensional or three-dimensional, preferably three-dimensional. The representation can already exist from previous work, be received from a remote computer or the cloud, or be generated on the same computer as the steps of the method according to the present invention. Preferably, the representation is generated from an image showing the internal structure of the sample. The representation includes visual data such as pixel or voxel data. Various formats can be used, such as point maps, point clouds, triangulated surface models such as Standard Triangulation Language (STL), etc. The format of the representation must be suitable for including information about materials and / or voids.

[0018] Images of a sample can be provided in a variety of ways. For example, they can be obtained directly from a measuring device or from a database containing previous measurements. Various measurements are available, such as photography, optical or electron microscopy of cut or broken samples, or non-destructive methods such as computed tomography, magnetic resonance imaging, ultrasound, or confocal microscopy. Images are typically acquired from parallel planes within the sample, although planes can also be angled relative to one another. The distance between the planes from which images are acquired typically does not exceed the largest feature of the sample's internal structure. The number of images can vary. Typically, more images yield more accurate results. However, the number of images increases computation time, especially if the image resolution is high. Therefore, it is usually best to use the minimum number of images that provides sufficient accuracy for a particular sample. This number can range from 5 to 1000, e.g., 10 to 50 or 100 to 400. Image resolution must be high enough to clearly identify features, but not too high to avoid excessive computation time. Common resolutions for images are -10x10 to 1024x1024 pixels; images do not need to be quadratic, so 768x1024 or 512x288 pixels can be used as well. Preferably, the image is grayscale or can be converted to grayscale.

[0019] Preferably, before generating a representation from the images, the images are pre-processed to facilitate the detection of phase boundaries. Pre-processing may include adjusting brightness, contrast, noise reduction, applying a threshold, or a combination thereof. Even more preferably, the pre-processing parameters that yield the best results with the method according to the invention are saved and automatically suggested to the user, or applied directly to further images to be pre-processed.

[0020] The generation of representations from images can be achieved in a variety of ways. Most of them involve edge or surface detection or segmentation, which determines the respective phase boundaries. Edge detection involves, for example, assigning a threshold gray value to edge voxels, and converting 3D voxel data into 3D surface data by interpolating between voxel gray values, the derivative of the maximum gray value, intermediate gray values ​​between the light air voxel and dark material voxel levels, or searching for a locally adaptive gray threshold. Noise and artifact reduction, as well as interpolation, are the subject of many publications known to the skilled artisan.

[0021] Preferably, the representation is generated from the image by segmenting the grayscale image by applying a thresholding algorithm, thereby converting the grayscale into an image in which each color represents a phase, i.e., a specific material or void. For example, in a foam, the material may be white and the voids may be black. In a composite material containing three materials, the first material may be white, the second material gray, and the third material black. In some cases, the segmented image may already be sufficient for the representation. However, it is often useful to apply further methods to reliably extract structural features from the representation. Preferably, the segmented image is subjected to a distance function that assigns each pixel or voxel the distance to the nearest pixel or voxel of a different color. Preferably, after applying the distance function, a watershed algorithm is applied to identify objects such as pores, embedded particles, walls, pillars, or nodes. If the watershed algorithm overflows, the centers of walls, pillars, and nodes can also be determined.

[0022] The method according to the invention comprises (b) extracting at least one structural feature from the representation, all of which are features that are directly linked to the internal structure of the sample. A structural feature can be related to walls, i.e., the material between two particles of different materials or between two pores, e.g., their thickness, curvature, planar expansion, or moment. A structural feature can also be related to struts, i.e., the material between three particles of different materials or between three pores, e.g., their cross-section, length, curvature, or moment. A structural feature can also be related to nodes, i.e., the material between four particles of different materials or between four pores, e.g., their volume or moment. A structural feature can also be related to pores or particles, e.g., their volume, sphericity, or moment. A structural feature can also be related to cells, e.g., their local density, or moment. A structural feature can also be related to a graph, such as a foam graph, e.g., the connections between pores, walls, cells, struts, and nodes.

[0023] By evaluating these features, more complex structural signatures can be obtained. An example is the separate evaluation of absolute values ​​and their spatial distribution, such as determining the strut length gradient of a sample. Another example is the evaluation of structural features, which are often related to physical models, such as the strut length related to the size of the adjacent pore. Another example is the evaluation of foam graphs, such as the number of walls per pore or the number of struts per wall.

[0024] According to the present invention, at least one structural feature includes a wall, a support, or a node. If only one structural feature is extracted, it must be either a wall, a support, or a node. Typically, multiple structural features are extracted. In this case, at least one of the structural features is a wall, a support, or a node, and the others may be one or more of the remaining walls, support, or nodes, or other structural features as described above. Preferably, the structural features include at least two of the walls, support, or nodes, and in particular, the structural features include all of them, i.e., walls, support, and nodes.

[0025] Extracting structural features can be achieved in a variety of ways. Suitable algorithms include watershed, distance estimation, component analysis, local voxel estimation, or a preferred combination of at least two of these. These methods are readily available in image processing libraries, such as SciKit-image, OpenCV, SimpleCV, NumPy, SciPy, PIL / Pillow, Mahotas, ITK, GraphicsMagick, or Cairo. Extracting structural features from representations offers several advantages over other methods, such as measurement methods. They offer greater flexibility, as very different features can be extracted that are otherwise only accessible through different measurement methods. They are often more accurate, as they tend to be less susceptible to artifacts and disturbances.

[0026] The method according to the invention comprises (c) providing the at least one structural feature to a material model suitable for obtaining at least one material property from the structural feature, where it is possible to determine only one material property or, preferably, one or more material properties, such as at least two or at least three.

[0027] A material model generally refers to a model that accepts structural features as inputs and outputs related to material properties. Material models include physical models and data-driven models. Physical models use natural laws such as thermodynamics and classical mechanics to convert structural features into material properties. A significant number of physical models for materials with cellular structure are summarized, for example, in L. Gibson and M. Ashby's Cellular Solids, Cambridge University Press, ISBN 978-0-521-49911-8. Physical models can be verified by experimental data.

[0028] A data-driven model is a trained mathematical model parameterized according to training data, which inputs structural features and outputs associated material properties without requiring knowledge of physical laws. The data-driven model is preferably a data-driven machine learning model. The model can be a linear or polynomial regression, a random forest model, a Bayesian network, or a neural network. Preferably, the data-driven model is trained on historical data and then simplified by taking into account only the structural features that significantly affect the material properties of interest. In this way, the amount of historical data required is reduced while still obtaining a robust model with high accuracy.

[0029] Historical data in the context of the present invention refers to a data set containing at least one structural feature and at least one material property. Such data is typically obtained by measuring past samples, usually by a suitable method for directly or indirectly obtaining the respective material property, e.g., mechanical testing such as indentation. The relevant structural features can be obtained by analyzing images of the sample similar to the methods described above, or by analytical methods specific to each structural feature.

[0030] The material model can run on the same system as the other steps in the process, or on a remote system such as a server or the cloud. In this case, the structural features are sent to the remote system that runs the material model, and the results are received from the remote system. This is typically achieved through a communications interface.

[0031] Typically, the material type, i.e., the chemical composition of each phase of a sample, influences the relationship between structural and material characteristics. When samples have different material types, it is desirable to consider information about the material type of each phase of the sample. In this way, more accurate material characteristics are typically achieved. Material types can be specified as general material classes, such as ceramics, resins, viscoelastic polymers such as rubber, or metals. Material types can also be more specifically indicated by reference to their chemical composition, such as polystyrene, zeolite, melamine-formaldehyde resin, oak wood, borosilicate glass, etc. Therefore, the material type of each phase of a sample is preferably provided.

[0032] Typically, the material type is taken into account by selecting the appropriate model. It is also possible to set up separate models for each material type. This makes sense when only a few different material types are of interest, for example, in a factory producing a few different products. Alternatively, the material type of each phase can be used as an additional input parameter in the model. Obviously, for data-driven models, the historical data used to train the model must be properly labeled with the material type of each phase in the sample. For physical models, bulk material properties are usually chosen because they can be easily obtained from a database. Errors arising from the fact that the properties of materials in small structures deviate from the bulk material are acceptable for most applications.

[0033] The method according to the present invention includes (d) outputting at least one material property received from the material model, where outputting means writing the material property to a non-transitory data storage medium, displaying it on a user interface, or transmitting it to another program on a local or remote system, and preferably the at least one material property is output to a user interface.

[0034] The method for determining the material properties of a sample preferably comprises: (a1) providing an image of a foam sample; (a2) providing the material type of each phase of the sample; (a3) converting an image into a representation; (b) extracting at least one structural feature from the representation, the at least one structural feature including a wall, a support, or a node; (c) training based on physical models or historical data including structural features and material types; providing at least one structural feature to a material model, the material model being a data-driven model; (d) outputting the at least one material property received from the material model.

[0035] An example of how the present invention can be implemented is shown in Figure 1. Samples may be manufactured at a factory 10. They are subjected to a microscope device 11, which generates images of the samples. These images are converted into a representation by a processing unit 12. The representation is provided to a processing unit 13, which extracts at least one structural feature from the representation. The at least one structural feature is provided to a processing unit 14, which provides it to a model. The model has been trained with historical data obtained from a data storage device 15. The model obtains material properties, which are provided to an output device 16. The output device 16 can output the material properties to the factory 10, for example, to adjust production parameters.

[0036] The present invention further relates to a non-transitory computer-readable data medium storing a computer program comprising instructions for executing the steps of the method according to the present invention. The computer-readable data medium includes, for example, a hard drive on a server, a USB storage device, a CD, a DVD, or a Blu-ray disc. The computer program may contain all the functions and data necessary for executing the method according to the present invention, or may provide an interface for having parts of the method processed on a remote system, for example a cloud system.

[0037] The present invention further relates to a production monitoring and / or control system for monitoring and / or controlling material properties of foam samples. Unless explicitly stated differently below, descriptions of methods, including preferred embodiments, also apply to systems. The system can be a computing device, such as a computer, tablet, or smartphone. Often, the computing device has a network connection for communicating with other computing devices, such as a server or cloud network. Production can refer to mass production in a factory or the production of several samples in the context of a research program. Monitoring is typically performed in the context of quality control to ensure that products always fall within set ranges of certain material properties or to classify products based on different specifications (e.g., high-quality products versus average-quality products). Control can refer to the process of selecting the best samples to facilitate and speed up the research and development process.

[0038] According to the present invention, the system comprises (a) an input unit configured to receive images showing the internal structure of the sample. Preferably, the input unit comprises a user interface that allows a user to select the images to be processed, for example from a local or remote storage medium or directly from a measurement device that analyzes the sample. Preferably, the input unit is configured to receive the material type of each phase of the sample. The input unit can be implemented as a web service or a standalone software package. The input unit may form a presentation layer or an application layer. Preferably, the input unit comprises the user interface.

[0039] According to the invention, the system comprises (b) a processing unit configured to extract at least one structural feature from the representation. The processing unit may be a local processing unit and include a central processing unit (CPU) and / or a graphics processing unit (GPU) and / or an application specific integrated circuit (ASIC) and / or a tensor processing unit (TPU) and / or a field programmable gate array (FPGA). The processing unit may also interface to a remote computer system, such as a cloud service. It can be a source.

[0040] According to the present invention, the system comprises (c) a processing unit configured to provide at least one structural feature to a material model suitable for obtaining at least one material property from the structural feature. The processing unit may be the same as (b) or different. For example, the processing unit of (b) is on a local machine, while the processing unit of (c) is an interface to a cloud service.

[0041] According to the present invention, the system comprises (d) an output unit configured to output the material properties received from the material model. The output unit can be implemented as a web service or a standalone software package. The output unit can form a presentation layer or an application layer. Preferably, the output unit is a user interface configured to display the material properties of the samples. A user can then perform the necessary action, for example, adjusting production parameters if the samples are out of specification or selecting the highest quality samples in a research project. Alternatively, the output unit can include or have an interface to a device that automatically adjusts production parameters or classifies samples according to their material properties. Example Figures 2a through 2g show an example of how to achieve steps (a) and (b). Figure 2a shows raw data, e.g., obtained from an X-ray tomography system. After applying a filter to prepare for binarization, the image in Figure 2b is obtained. Figure 2c shows the result of applying a threshold to binarize the image. In Figure 2d, a distance filter was applied to both phases: an opposite negative sign for the pore phase and a positive sign for the material phase. Subsequently, local minima were identified, and a watershed algorithm with lines between cells and no masking was applied. The result is shown in Figure 2e. Figure 2f shows the masked labeled cells obtained from the binarized data from the image in Figure 2c, and the labeled pores are obtained. The resulting watershed boundary shown in Figure 2e represents the foam skeleton and is shown in Figure 2g. From there, voxels within the skeleton can be labeled by the number of neighboring cells. That is, a voxel with any adjacent cells represents a wall, a voxel with three adjacent cells represents a pillar, a voxel with four or more cells represents a node, etc. Connected voxels are labeled with the same kind of label as a single wall, pillar, or node.

[0042] From this, the fraction of material located in the cell walls, φ = 0.249, is taken as the foam density (ρ foam ) and the density of the bulk material (ρ bulk The relative density ρ* = 0.531, which represents the ratio of the relative Young's modulus E * =0.344 is the following formula

[0043]

number

[0044] was obtained using

Claims

1. 1. A computer-implemented method for determining material properties of a foam sample, comprising: (a) providing a representation of said sample; (b) extracting at least one structural feature from the representation, the at least one structural feature comprising a wall, a support, or a node; and (c) providing the at least one structural feature to a material model suitable for deriving at least one material property from the structural feature; (d) outputting the at least one material property received from the material model.

2. The computer-implemented method of claim 1 , wherein the representation is generated from an image showing an internal structure of the sample.

3. The computer-implemented method of claim 2 , wherein the representation is generated from the image by segmenting a grayscale image by applying a thresholding algorithm.

4. The computer-implemented method of claim 3 , wherein the segmented image is subjected to a distance function and a watershed algorithm.

5. The computer-implemented method of claim 2 , wherein, prior to generating a representation from the image, the image is pre-processed by automatically applying stored pre-processing parameters.

6. The computer-implemented method of claim 1 , wherein the method further comprises providing a material type of the sample, and wherein the material model is material type specific.

7. The computer-implemented method of claim 1 , wherein the representation is a three-dimensional representation.

8. The computer-implemented method of claim 1 , wherein the material properties are displayed on a user interface.

9. The computer-implemented method of claim 1 , wherein at least two structural features are extracted from the representation and provided to the material model.

10. The computer-implemented method of claim 1 , wherein the sample is a polymer foam.

11. A non-transitory computer readable data medium storing a computer program comprising instructions for performing the steps of the method according to any one of the preceding claims.

12. 1. A production monitoring and / or control system for monitoring and / or controlling material properties of a foam sample, comprising: (a) an input unit configured to receive a representation of said samples; (b) a processing unit configured to extract at least one structural feature from the representation, the at least one structural feature comprising a wall, a support, or a node; and (c) a material model suitable for deriving at least one material property from said structural features; a processing unit configured to provide the at least one structural feature to the (d) an output unit configured to output material properties received from said material model.

13. 13. A production monitoring and / or control system according to claim 12, wherein the input unit is configured to receive the material type of each phase of the sample.

14. 14. A production monitoring and / or control system according to claim 12 or 13, wherein the processing units (b) and / or (c) are interfaces to remote computer systems.

15. 15. A production monitoring and / or control system according to any one of claims 12 to 14, wherein the input unit and / or the output unit is a user interface.

Images