Article-based cleanliness definition

Abstract

The invention relates to a computer-assisted method for creating an article-based cleanliness definition of preconcentrates from sorting systems for solid waste materials, the method comprising the following method steps: a) detecting at least two-dimensionally resolved raw sensor data relating to at least one preconcentrate, by means of a sensor; b1) classifying the raw sensor data on the basis of predefined material classes; b2) creating a first false colour image on the basis of the classification of the raw sensor data; c) correcting the false colours of the first false colour image of secondary article components; d) creating a second false colour image on the basis of the change in the false colour of the secondary article component regions; and, e) creating a false colour distribution on the basis of the second false colour image.

Description

[0001] ITEM-BASED PURITY DEFINITION

[0002] Description:

[0003] The present invention relates to the technical field of methods for creating an article-based purity definition of pre-concentrates from sorting plants for solid waste materials.

[0004] When quantifying the amount of contaminants in product fractions from sorting plants for, for example, mixed solid waste materials, heterogeneous input material streams are sorted into pre-concentrates of the highest possible purity using multi-stage sorting processes. These pre-concentrates are then fed to specialized processing plants for further recycling.

[0005] Examples of pre-concentrates from the sorting of, for example, lightweight packaging waste, household waste and commercial waste include polypropylene, polyethylene, polyethylene terephthalate (PET), thermoformed trays made of PET, polystyrene, corresponding films, liquid cartons, paper, cardboard and cartons, which can each be further subdivided into subcategories and color classes, as well as sorting residues.

[0006] Individual waste items that form a pre-concentrate generally consist of a main component and one or more secondary components. Using a PET bottle as an example, the main component would be the PET bottle body, while secondary components could include bottle caps, labels, or even contaminants, residues, moisture, or similar substances.

[0007] The quality of pre-concentrates in sorting plants is assessed using a state-of-the-art, manual quality control process, based on the specific item. For example, with PET bottles, the pre-concentrate is considered 100% pure if it consists exclusively of items with PET bottles as the main component and their associated secondary components. These secondary components can be assessed according to their material or quality-defining properties, such as the type of material and / or color of the waste item's main component. Following the actual sorting process, the individual pre-concentrates or sorted products are typically on separate conveyor belts or systems and temporarily stored in product hoppers. This continues until a suitable quantity of product is available, depending on the specific case, to compress one or more bales of the same pre-concentrate.

[0008] The individual bunkers can be equipped with load cells to record the current quantity of product. A bunker management system controls the individual bunkers, the conveying systems leading away from the bunkers (such as one or more baler feed belts), and the baler(s). In the baler, the pre-concentrates are compressed into individual bales weighing, for example, 200 kg to 1000 kg each. The bales can then be stored and prepared for transport to the respective downstream processing plants.

[0009] Typically, RGB cameras and deep learning methods are used in state-of-the-art technology to determine the quality of pre-concentrates via object recognition. However, this requires material separation, or the recognition performance decreases significantly when materials are overlapping.

[0010] The technical objective of the present invention is to improve the state of the art and to propose alternatives.

[0011] The technical problem of the invention is solved by an object with the technical features according to the independent claims. Advantageous embodiments are the subject of the dependent claims, the description, and the drawings.

[0012] All features described in connection with individual embodiments of the invention can be provided in different combinations in the invention to simultaneously achieve their advantageous effects. The scope of protection of the present invention is defined by the claims and is not limited by the features described in the description or shown in the figures. Although the invention is described in detail by the preferred embodiment, the invention is not limited by the disclosed examples, and other variations can be derived by a person skilled in the art without departing from the scope of protection of the invention. For the purposes of the invention, purity can also be defined, for example, with regard to color, whether the product is food or non-food, or shape, such as the shape of a bottle or a bowl.

[0013] For the purposes of the invention, the term "material classes" can refer to various materials. In particular, the materials can consist of, or comprise: polyethylene terephthalate, polypropylene, polyethylene, high-density polyethylene, linear low-density polyethylene, low-density polyethylene, polystyrene, beverage cartons, paper, non-ferrous metals, ferrous metals, organic materials, and various sorting residues. Material residues and / or contaminants can also be included. Furthermore, the term "material classes" can encompass additional quality characteristics such as color, shape, or whether the material is food-grade or non-food.

[0014] In accordance with the invention, a false-color image can, for example, be interpreted as the sum of individual regions, that is, as areas with the same false color. In one process step, for example, all regions are identified as individual regions based on their false color and proximity. From all regions, those regions that represent article components with a significant probability are selectively filtered out based on one or more regional characteristics. For example, it may be provided that a threshold value is set for the respective regional characteristics, such as a defined projection area or a defined size range of a projection area.

[0015] Furthermore, for the purposes of the invention, the term "waste materials" can also be understood to include waste streams and / or mixed waste streams and / or mixed post-consumer waste streams. These waste streams can, in particular, comprise at least one of the following components: household waste, dry recyclable mixtures, commercial waste, lightweight packaging, plastic bottles, films, paper, cardboard, electronic waste, mixed construction materials, bulky waste, wood, scrap metal, and metal-containing waste, but are not limited to these.

[0016] Furthermore, for the purposes of the invention, a sensor can also be understood to mean a sensor unit, wherein the sensor or sensor unit can also be designed as a soft sensor. Additionally, for the purposes of the invention, a material class can be understood to mean a classification of materials based on their material properties, such as shape, color, or the like.

[0017] According to one aspect, the technical problem of the invention is solved by a computer-aided method for creating an article-based purity definition of pre-concentrates from sorting plants for solid waste materials, the method comprising the following process steps: a) Acquiring at least two-dimensionally resolved sensor raw data relating to at least one pre-concentrate using a sensor; b1) Classifying the sensor raw data based on predefined material classes; b2) Creating a first false-color image based on the classification of the sensor raw data; c) Correcting the false colors of the first false-color image of article components; d) Creating a second false-color image based on the change in the false color of the article component regions; e) Creating a false-color distribution based on the second false-color image.

[0018] Advantageously, multiple material streams can be measured batchwise and / or sequentially with a single sensor. This results in significantly lower investment costs, as, for example, instead of one sensor for each material stream, only one sensor is needed for multiple streams. A further advantage is that this allows for item-based quality assessment of the production batch, even in cases of missing singulation or material overlap, without requiring object recognition. Since the product component is specifically considered independently of the object, and individual false-color regions, which may belong to one or more waste items, are also considered independently of the object (i.e., without object or item recognition), monitoring upstream of a baling press in cases of material overlap becomes possible.

[0019] In a technically advantageous embodiment, it is provided that in process step b1) the classification of the sensor raw data is carried out pixel by pixel.

[0020] This advantageously ensures the highest possible precision and resolution of the data. In a technically advantageous embodiment, the classification of the raw sensor data and / or the creation of the first false-color image is performed on the sensor itself and / or on an external processing unit, with the sensor being designed, in particular, as a soft sensor.

[0021] This can advantageously speed up the process and save on additional components required to carry out the process.

[0022] In a technically advantageous embodiment, it is provided that process step c) comprises the following process steps: c1) identification of individual regions of the first false-color image based on the false color and a neighborhood of the respective region; c2) determination of article component regions of the first false-color image whose false color represents article components with a significant probability; c3) changing the false color of the article component regions to the false color that represents an article main component belonging to the article component;

[0023] Based on information from the neighborhood of the article sub-component, the false color of the article sub-component is corrected to match the false color of the associated article main-component. The result is a new false-color image whose false-color distribution corresponds to an article-based purity definition.

[0024] In a technically advantageous embodiment, it is provided that at least one of the process steps c) and / or c1) and / or c2) and / or c3) and / or c4) and / or c5) is carried out using image processing methods, in particular morphological operations, and / or using machine learning and / or using deep learning models.

[0025] In a technically advantageous embodiment, it is provided that the process step bl ) and / or b2) and / or c) and / or d) is carried out using Generative-Adversarial-Network models.

[0026] In a technically advantageous embodiment, it is provided that process step bl) and / or c) includes different material classes and / or different probabilities and / or settings and / or methods for identifying the article components in each process run. It can also be provided that the process steps are executed multiple times. In this way, a correction and definition of the article components can advantageously be configured depending on the task and depending on the process run.

[0027] In a technically advantageous embodiment, the method includes the following process step: f) Determining areas and / or area fractions of the first false-color image and / or the second false-color image for each false color.

[0028] From the corrected false-color image, the corrected areas for each material class can then be determined in order to calculate an area-based composition of the corrected false-color image with an article-based purity definition.

[0029] In a technically advantageous embodiment, the method includes the following process step: g) Correction of demixing effects, in particular by weighting the false colors of the first false color image and / or the second false color image.

[0030] This can be advantageously used to correct systematic effects such as residual filling, hydrophobicity, hydrophilicity, fraction-specific enrichment mechanisms in the bulk material, or material-specific wall thicknesses, which offers additional benefits, for example, in the case of material overlap. If material overlap occurs on the conveying unit at the location of the sensor unit, segregation effects can occur, for example, due to the Brazil nut effect. The Brazil nut effect, also known as the muesli effect, occurs in mixed granular media. After repeatedly shaking a muesli package containing particles of various sizes, the largest particles rise to the top.

[0031] In a technically advantageous embodiment, the method includes the following step: h) Conversion of area-based sensor raw data into mass-based sensor raw data, in particular into mass-based composition and / or mass-based quality parameters. Most imaging sensors capture quality properties based on area, pixels, and / or volume. However, in many applications in the recycling industry, mass-based compositions and / or mass-based quality parameters are relevant. In these cases, further data processing can be performed in which the area-based quality parameters are converted into mass-based quality parameters using, for example, basis weights, densities, tables of values ​​from experience, or machine learning methods.

[0032] According to another aspect, the technical problem of the invention is solved by a data processing device comprising means for carrying out the method according to one of the preceding claims.

[0033] According to another aspect, the technical problem of the invention is solved by a computer program product on a medium, in particular a computer-readable, preferably non-volatile, medium for carrying out the method according to one of the preceding claims.

[0034] According to another aspect, the technical problem of the invention is solved by a computer-readable, in particular non-volatile, medium comprising instructions for carrying out the method according to one of the preceding claims.

[0035] According to a further aspect, the technical problem of the invention is solved by a device for creating an article-based purity definition of pre-concentrates in sorting plants for solid waste materials, comprising a bunker, a conveying system, a baling press, at least one sensor, wherein the device comprises means for carrying out the method according to one of the preceding claims and is in particular configured to carry out the method according to one of the preceding claims.

[0036] A conveying system can also refer to a conveying technology, a conveying unit, or a conveying technology assembly. For example, one or more conveying systems may be required to transport pre-concentrates between the hopper and the baler. This can involve one or more conveying systems or combinations thereof. For instance, one or more chain belt conveyors, one or more belt conveyors, one or more chutes, one or more screw conveyors, one or more vibratory conveyors, or other conveying devices may be used. Special conveying technology can be employed as part of the conveying system to improve the presentation of the material flow on the conveying unit to the sensor. This can be achieved, for example, using one or more acceleration belts or one or more chutes.For example, one or more measurements in free fall can also be provided for this purpose, whereby several and / or different pre-concentrates are transported one after the other via one or more conveyor systems, allowing several different pre-concentrates to be monitored at once with only one sensor.

[0037] In a technically advantageous embodiment, the sensor is designed as an imaging sensor, wherein the sensor is in particular designed as a sensor system comprising several sensors.

[0038] The sensor may be designed for material detection. It may also be designed to function as a sensor system in conjunction with other sensors. For example, at least one sensor may be configured as an imaging sensor. This sensor may be configured, for example, as a hyperspectral or multispectral near-infrared sensor, a hyperspectral or multispectral near-infrared camera, a hyperspectral or multispectral mid-infrared sensor, a hyperspectral or multispectral mid-infrared camera, an RGB color line camera, a 3D laser triangulation sensor, an X-ray transmission sensor, an X-ray fluorescence sensor, a laser-induced breakdown spectroscopy sensor, or a point sensor.

[0039] In a technically advantageous embodiment, the device further comprises at least one illumination source and / or irradiation source suitable for at least one sensor.

[0040] For example, one or more light sources, or one or more irradiation sources, can be provided, which are adequately configured for one or more sensors. For example, halogen lighting can be used simultaneously for an RGB color line sensor and a near-infrared hyperspectral sensor. For example, halogen spotlights and / or light sources, especially LED light sources, can be provided as the light source and / or radiation source. The measurement can be performed, for example, via reflection, transmission, or transflection of the radiation.In one embodiment using a sliding surface, for example a slide, the measurement can be carried out not only by reflecting the radiation from the object onto the sensor, but also transmissively, through the surface of the sliding surface, for example a glass slide, or transflectively by reflection at a sliding surface surface, for example a metal slide surface.

[0041] In a technically advantageous embodiment, the sensor system is arranged on the conveyor system itself and / or via a bypass, in particular via an automated sampling unit.

[0042] In a technically advantageous embodiment, the bunker includes a load cell, in particular for recording a possible bale weight, and / or is designed to be emptied onto at least one conveying system, particularly on one of its bunker sides.

[0043] The load cells can advantageously measure the bale weight. Furthermore, it is advantageous that two conveying units can be fed simultaneously.

[0044] According to another aspect, the technical problem of the invention is solved by a device with a computer-readable medium for the method according to one of the preceding claims, wherein the device further comprises a control unit and a communication interface, wherein the communication interface is coupled to the control unit in such a way as to transmit the false-color images for display on a device.

[0045] According to a further aspect, the technical problem of the invention is solved by a device comprising a computer-readable medium for the method according to one of the preceding claims, wherein the device further comprises a recording medium on which data, in particular according to a specific pattern, preferably including functional data, are stored, most preferably according to a method according to one of the preceding claims.

[0046] According to a further aspect, the technical problem of the invention is solved by a computer-aided method according to any one of claims 1 to 10, in particular for implementation on a device according to any one of claims 11 to 15, wherein the method comprises the process step: i) controlling the feed of the pre-concentrates from the hoppers to the baler. In a technically advantageous embodiment, it is provided that the computer-aided method according to claim 16 comprises the process step: j) feeding the hoppers from the sorting processes upstream of the method.

[0047] In a technically advantageous embodiment, the computer-aided method according to one of claims 16 or 17 comprises the method step: k) automatic emptying of the bunker onto the at least one conveying system.

[0048] In a technically advantageous embodiment, the computer-aided method includes the process step: l) displaying at least one false-color image, in particular of the article-based cleanliness, on a device.

[0049] Exemplary embodiments of the invention are shown in the figures and are described in more detail below.

[0050] They show:

[0051] Fig. 1 is a schematic representation illustrating a method for creating an article-based purity definition; and

[0052] Fig. 2 shows a schematic view of a device for creating an article-based cleanliness definition.

[0053] A combined view of Figures 1 and 2 shows a schematic representation illustrating a computer-aided method for creating an article-based purity definition of pre-concentrates 9 from sorting plants for solid waste materials, the method comprising the following process steps: a) Acquisition of at least two-dimensionally resolved sensor raw data relating to at least one pre-concentrate 9 using a sensor 5; b1) Classification of the sensor raw data based on predefined material classes; b2) Creation of a first false-color image 10 based on the classification of the sensor raw data; c) Correction of the false colors of the first false-color image of article components 12; d) Creation of a second false-color image based on the change in the false color of the article component regions 13; e) Creation of a false-color distribution based on the second false-color image 11.

[0054] This advantageously allows for the quality assessment of a production batch, for example, a bale 16, on an item-by-item basis, even in cases of missing singulation or material overlap of the pre-concentrates 9, without requiring object recognition. The item component can thus be considered specifically and independently of the object. Individual false-color regions, which may belong to one or more waste items, can be examined without object or item recognition, enabling monitoring upstream of the baling press.

[0055] Furthermore, a combined view of Figures 1 and 2 shows that process step c) comprises the following steps: c1) identification of individual regions of the first false-color image 10 based on the false color and a neighborhood of the respective region; c2) determination of article component regions 13 of the first false-color image 10 whose false color represents article components 13 with significant probability; c3) changing the false color of the article component regions 13 to the false color that represents a main article component 14 belonging to the article component;

[0056] Here, the article component region only encompasses the region of the article component, not the entire framed area. Therefore, the representation of the area (rectangle) is only indicative.

[0057] In a combined view of Figures 1 and 2, the following process steps are further illustrated: i) control of the feed of the pre-concentrates 9 from the hoppers 2 to the baler 4; j) feeding the hoppers 2 from the sorting processes upstream of the process (the sorting process is not shown); k) automatic emptying of the hopper 2 onto the at least one conveying system 3; l) display of at least one false-color image on a device 8. Figure 2 shows a schematic view of a device 1 for creating an article-based purity definition, comprising: a hopper 2, various conveying systems 3, a baler 4, and at least one sensor 5, wherein the device includes means for carrying out the method according to one of the preceding claims and is particularly configured to carry out the method according to one of the preceding claims.The sensor 5 is part of a sensor system 6, wherein the device 1 further comprises at least one illumination or irradiation source 7 suitable for at least one sensor 5, which is also shown as part of the sensor system 6, which is arranged on the conveyor system 3.

[0058] Reference symbol list:

[0059] 1 Device

[0060] 2 bunkers, 3 conveyor systems

[0061] 4 balers

[0062] 5 Sensor

[0063] 6 Sensor system

[0064] 7 Lighting or irradiation source 8 Device

[0065] 9 Pre-concentrate

[0066] 10 first false color image

[0067] 11 second false-color image

[0068] 12 article sub-components 13 article sub-components region

[0069] 14 Main Article Component

[0070] 15 external computing units

[0071] 16 bales

Claims

Claims:

1. Computer-aided method for generating an article-based purity definition of pre-concentrates (9) from sorting plants for solid waste materials, the method comprising the following process steps: a) Acquisition of at least two-dimensionally resolved sensor raw data relating to at least one pre-concentrate (9) using a sensor (5); b1) Classification of the sensor raw data based on predefined material classes; b2) Generation of a first false-color image (10) based on the classification of the sensor raw data; c) Correction of the false colors of the first false-color image (10) of article components (12); d) Generation of a second false-color image (11) based on the change in the false color of the article component regions (13); e) Generation of a false-color distribution based on the second false-color image (11).

2. Computer-aided method according to claim 1 characterized in that in method step bl ) the classification of the sensor raw data is carried out pixel by pixel.

3. Computer-aided method according to one of the preceding claims, characterized in that the classification of the sensor raw data and / or the creation of the first false color image (10) takes place on the sensor (5) itself and / or on an external computing unit (15), wherein the sensor (5) is in particular designed as a soft sensor.

4. Computer-aided method according to one of the preceding claims, characterized in that method step c) comprises the following method steps: c1) identification of individual regions of the first false-color image (10) based on the false color and a neighborhood of the respective region; c2) determination of article component regions (13) of the first false-color image (10), whose false color represents article components (12) with significant probability; c3) Change of the false color of the article subsidiary component regions (13) to the false color that represents an article principal component (14) belonging to the article subsidiary component (12); 5. Computer-aided method according to one of the preceding claims, characterized in that at least one of the method steps c) and / or the method step c1) and / or the method step c2) and / or the method step c3) and / or the method step c4) and / or the method step c5) is carried out by means of image processing methods, in particular morphological operations, and / or by means of machine learning and / or by means of deep learning models.

6. Computer-aided method according to one of the preceding claims, characterized in that the method step b1) and / or the method step b2) and / or the method step c) and / or the method step d) is / are performed by means of an image processing program and / or by means of a machine learning model and / or a deep learning model such as a generative adversarial network model, preferably in one step.

7. Computer-aided method according to one of the preceding claims, characterized in that the method step b1) and / or the method step c) has different material classes and / or different probabilities and / or settings and / or methods for identifying the article components (12) in each method run.

8. Computer-aided method according to one of the preceding claims, characterized in that the method comprises the following process step: f) Determination of areas and / or area fractions of the first false color image (10) and / or the second false color image (11) for each false color.

9. Computer-aided method according to one of the preceding claims, characterized in that the method comprises the following process step: g) correction of demixing effects, wherein in particular a weighting of the false colors of the first false color image (10) and / or the second false color image (11) is carried out.

10. Computer-aided method according to one of the preceding claims, characterized in that the method comprises the process step: h) conversion of area-based sensor raw data or false-color data or material flow properties or compositions or quality parameters derived therefrom into mass-based sensor raw data or false-color data or material flow properties or compositions or quality parameters derived therefrom, in particular into mass-based composition and / or mass-based quality parameters.

11. Device (1) for creating an article-based purity definition of pre-concentrates in sorting plants for solid waste materials, comprising at least one bunker (2), at least one conveying system (3), at least one baling press (4), at least one sensor (5), characterized in that the device comprises means for carrying out the method according to one of the preceding claims and is in particular designed to carry out the method according to one of the preceding claims.

12. Device (1) according to claim 11, characterized in that the sensor (5) is configured as an imaging sensor, wherein the sensor is in particular configured as a sensor system (6) comprising several sensors (5).

13. Device (1) according to one of the preceding claims, characterized in that the device (1) further comprises at least one illumination source (7) and / or irradiation source (7) suitable for at least one sensor (5).

14. Device (1) according to one of the preceding claims, characterized in that the sensor system (6) is arranged on the conveying system (3) itself and / or via a bypass, in particular via an automated sampling unit.

15. Device (1) according to one of the preceding claims, characterized in that the bunker (2) comprises a load cell, in particular for recording a possible bale weight and / or is designed to be emptied onto at least one conveying system (3), in particular on one of its bunker sides.

16. Computer-aided method according to one of claims 1 to 10, in particular for carrying out on a device (1) according to one of claims 11 to 15, characterized in that the method comprises the method step: i) Control of the supply of the pre-concentrates (9) from the hoppers (2) to the baler (4).

17. Computer-aided method according to claim 16, characterized in that the method comprises the process step: j) feeding the bunkers (2) from the sorting processes upstream of the method.

18. Computer-aided method according to one of the preceding claims 16 or 17, characterized in that the method comprises the process step: k) automatic emptying of the bunker (2) onto the at least one conveying system (3).

19. Computer-aided method according to one of the preceding claims, characterized in that the method comprises the process step: l) displaying at least one false-color image, in particular of the article-based purity, on a device (8).

Images