WO2025016952A1

WO2025016952A1 - Method and system for classifying polyurethane objects from a waste stream

Info

Publication number: WO2025016952A1
Application number: PCT/EP2024/069968
Authority: WO
Inventors: Patrick Weiss; Oliver Pikhard; Achim Stammer; Christoph Schnorpfeil; Anant Vikas AGGARWAL; David Josef RUESSMANN; Christine Schmitz
Original assignee: Basf Se; Tomra Feedstock Gmbh
Priority date: 2023-07-14
Filing date: 2024-07-15
Publication date: 2025-01-23

Abstract

Method and classification system for classifying PU objects from a waste stream comprising several PU objects according to at least one predefined property, wherein the PU objects may comprise at least one impurity. The PU objects are irradiated with NIR and/or middle-infrared radiation and a NIR and/or middle-infrared radiation spectrum is recorded by detecting NIR and/or middle-infrared radiation. Based on the recorded a NIR and/or middle-infrared radiation spectrum, it is determined whether the irradiated PU objects exhibit the predefined property while verifying a possible presence of the at least one impurity in the one or more irradiated PU objects. This comprises that the recorded NIR and/or middle-infrared radiation spectrum is compared to one or more reference NIR and/or middle-infrared radiation spectra of reference PU objects exhibiting the predefined property and comprising a known amount of the at least one impurity.

Description

Method and system for classifying polyurethane objects from a waste stream

FIELD OF THE INVENTION

The present invention relates to a classification system and to a method for classifying one or more polyurethane (PU) objects from a waste stream. Moreover, the present invention is directed to a computer program comprising instructions for classifying one or more PU objects from a waste stream and to a non-transitory computer-readable data medium storing the computer program. The present invention generally concerns the sorting of PU objects from a waste stream for reuse and recycling purposes.

BACKGROUND OF THE INVENTION

PUs have become an integral part of the human lifestyle today and are used in different applications such as in varnishes and coatings, adhesives, electrical potting compounds, fibres and PU laminate. However, today, most PU is used in form of foams, e.g., in mattresses or the like. PU foams are formed in a wide range of densities and maybe flexible, semi-rigid or rigid in structure. Various blowing agents, additives, surfactants, catalysts, etc. are added during manufacturing of foams. PU, including foams, is one of the most important groups of materials available today and hence, their recycling is of great economical, as well as environmental interest. However, the recycling of PU and, in particular, of PU foams remains challenging such that a lot of PU that is produced ends up as waste in various waste streams and often is simply disposed on landfills.

Challenges that are typically faced when it comes to the recycling of PU foams can be attributed to the presence of various additives, surfactants, blowing agents, and catalysts. Moreover, PU waste is often mixed with various other materials such as metals, latex, wood BASF SE

220246

or textiles that interfere with a recycling process. Furthermore, PU materials generally have different composition and may include ether, ester, polyols, toluene diisocyanate (TDI) and methylene diphenyl diisocyanate (MDI).

For example, WO 2022/038052 A1 discloses a method for separating waste PU foams, wherein for each PU sample of a supply stream comprising PU samples from waste at least one respective spectrum is recorded. The disclosed method is characterized in that at least one respective spectrum is recorded by near-infrared (NIR) spectroscopy, wherein each PU sample of the supply stream is classified by a classification algorithm, which classification algorithm is based on machine learning, based on the respective at least one spectrum into a respective class of at least two classes. The supply stream comprising PU samples is separated into at least two streams according to the classification into the respective class and wherein each class corresponds to a type of PU.

However, for an efficient recycling and reuse of PU waste, a separation of PU types present in the PU waste with comparatively high accuracy is generally needed. It is therefore still desirable to be able to more accurately classify PU objects in order to further improve the sorting of PU objects.

SUMMARY OF THE INVENTION

The present invention is based on the objective of providing an improved method and an improved classification system for classifying one or more PU objects from a waste stream. Furthermore, the present invention is based on the objective of providing an improved computer program comprising instructions for classifying one or more PU objects from a waste stream and of providing a non-transitory computer-readable data medium storing the computer program.

According to the invention, a method for classifying one or more PU objects from a waste stream that may include several PU objects according to at least one predefined property is proposed. The one or more PU objects may comprise at least one impurity. The method comprises the steps of:

- providing the waste stream including the several PU objects,

- irradiating the one or more PU objects of the several PU objects with NIR and/or middleinfrared radiation (MIR), BASF SE

220246

- recording a NIR and/or MIR spectrum of the one or more irradiated PU objects by detecting NIR and/or MIR reflected from the one or more irradiated PU objects,

- determining based on the recorded a NIR and/or MIR spectrum whether the one or more irradiated PU objects exhibit the at least one predefined property while verifying a possible presence of the at least one impurity in the one or more irradiated PU objects, and

- if it is determined that the one or more irradiated PU objects exhibit the at least one predefined property, classifying the one or more irradiated PU objects into a first group, and else, if it is determined that the one or more irradiated PU objects do not exhibit the at least one predefined property, classifying the one or more irradiated PU objects into a second group, wherein determining whether the one or more irradiated PU objects exhibit the at least one predefined property comprises comparing the recorded NIR and/or MIR spectrum to one or more reference NIR and/or MIR spectra of one or more reference PU objects exhibiting the predefined property and comprising a known amount of the at least one impurity.

The present invention includes the recognition that PU contributes to sustainable outcomes in many ways. One of the most important ways of protecting our natural resources is by reducing waste through reuse and recycling, e.g., by implementing a closed-loop supply chain, in which new products are made entirely from recycled materials. As a highly recyclable substance, PU is playing a major role in this effort. High quality recycling will not only reduce the rate of environmental damage but support in building a sustainable future.

However, the high throughput associated with industrial processes where waste materials are processed in bulk remains problematic with known techniques of separating waste PU. Furthermore, the purity levels of sorted preferred PU fractions is directly influencing the viability of the subsequent chemical recycling process to ensure obtaining a high-quality recycled product.

One way of identifying different PU types is to record a near infrared radiation (NIR) spectrum and/or MIR spectroscopy from the respective PU object that is indicative of a particular PU type. However, the closeness in wavelength distribution in NIR and/or MIR spectra of different PU types, e.g., in a PU can make sorting PU objects according to the NIR and/or MIR spectrum less accurate and the sorting result less reliable. The present invention includes the further recognition that sorting PU objects according to the NIR BASF SE

220246

and/or MIR spectrum can be used for achieving very accurate and reliable sorting results. However, sorting PU objects according to the NIR and/or MIR spectrum can be even more challenging since typically PU objects from a waste stream may contain impurities and may be modified in some way. Such impurities may affect the NIR and/or MIR spectrum and may cause a modification of the NIR and/or MIR spectrum with respect to a NIR and/or MIR spectrum recorded from a pure PU object having no impurities. Such modification of an NIR and/or MIR spectrum may even lead to the result that due to the closeness in wavelength distribution in NIR and/or MIR spectra of different PU types, a recorded NIR and/or MIR spectrum may be falsely interpreted to be associated with a different PU type compared to the actual PU type of the investigated PU object. As a result, the respective PU object is sorted into a wrong group and the sorting process itself will be inaccurate. It will thus be appreciated that any specific impurity such as a contamination of a PU object when analysing such PU objects may detrimentally affect the quality of the purity levels of these sorted preferred PU fractions. Therefore, particular care should be taken in the configuration of the classification system and method that is used in detecting and identifying different types of PU objects from a waste stream.

With the method according to the invention, it is possible to implement a comparatively cost-effective separation process with high-throughput to obtain high purity PU fractions that enable efficient chemical recycling into high-quality recycled products. This can be achieved with the method according to the invention in that recorded NIR and/or MIR spectrum of the one or more irradiated PU objects is analysed with respect to a possible presence of an impurity in the one or more irradiated PU objects. In particular, an improved accuracy can be achieved with the method according to the invention since as part of the analysis it can be determined whether the recorded NIR and/or MIR spectrum of the one or more irradiated PU object includes a modification caused by at least one impurity present in the one or more irradiated PU object. In particular, for determining whether the recorded NIR and/or MIR spectrum of the one or more irradiated PU objects includes a modification, the recorded NIR and/or MIR spectrum is compared to at least one reference NIR and/or MIR spectrum that is associated with a known impurity.

Thus, when analysing the recorded spectrum of the at least one irradiated PU object with respect to the at least one predefined property, a classification may be determined by comparing the recorded NIR and/or MIR spectrum of the one or more irradiated PU objects to reference NIR and/or MIR spectra of one or more reference PU objects. Of the one or more reference PU objects, it is known whether they exhibit the predefined property and that they comprise a known amount of the at least one impurity. Thereby, it is possible to classify upon a favourable comparison the one or more irradiated PU objects into a first BASF SE

220246

group upon a non-favourable comparison to classify the one or more irradiated PU objects into a second group.

Thereby, the following advantages can be achieved. Firstly, no pre-processing step such as drying when the impurity is water, may be required prior to the classification, since the possible presence of an impurity is considered during the classification of the one or more irradiated PU objects. Moreover, a reliability and/or accuracy of the classification can be increased, i.e., PU objects can be classified more precisely even though they include an impurity. Furthermore, a solution is provided to the challenge that spectra of different PU types may look very similar such that the presence of an impurity may lead to a high degree of uncertainty when trying to differentiate different PU types. With the method, it is thus possible to obtain high purity PU fractions by classifying one or more PU objects according to a predefined property. For example, with the method, a purity level of PU objects according to a predefined property of at least 95 % are achievable. A throughput that makes pre-recycling process economically viable may be achievable.

The method can be further used in the context of a pre-recycling process. Moreover, the NIR and/or MIR analysis as part of the method may be combined with other ways of determining whether the one or more irradiated PU objects exhibit the at least one predefined property such as with neural network-based processing of images captured from the one or more irradiated PU objects.

Preferably, the PU objects from the waste stream are flexible PU objects, e.g., made of or comprising PU foam. Generally, PU foams are produced by a reaction between a polyisocyanate component and a polyol component. Typically, further materials, in particular additives, such as flame retardants (e.g. phosphorous-based), polymerization catalysts (e.g. tertiary amines), fillers and surfactants as siloxanes can be added in the production process of the polymers. The properties of a polyurethane foam may be influenced by the chemistry of polyisocyanate and polyol components used and the recipe applied in polymerization. For example, the starting materials may influence the crosslinking density of the polymers in a three-dimensional network. Rigid polyurethane are typically obtained from monomers with a comparably low molecular weight and high functionality creating a highly cross-linked, dense network. Industrially and consequently in large quantities, especially methylene-di(phenylisocyanate) (MDI) or its polymeric forms or tolylene 2,4 and 2,6-diisocyanate (TDI) are used as polyisocyanate components for the production of PU rigid foams and PU flexible foams. For a representative composition of these PU foams, see for example US 9,023,907 B2,WO 2015/121057 and WO 2013/139781. BASF SE

220246

Preferably, NIR comprises a wavelength range of 0.7 pm to 2.5 pm and MIR comprises a wavelength range of 2.5 pm to 25 pm.

The method may further comprise a mechanical sorting step in which PU objects irradiated with NIR and/or MIR are sorted mechanically either in a group of wanted PU objects if the recorded NIR and/or MIR spectrum of the one or more at least one irradiated PU objects is within a pre-set NIR and/or MIR spectral range or, if the recorded NIR and/or MIR spectrum of the one or more at least one irradiated PU objects is not within the pre-set NIR and/or MIR spectral range in a group of unwanted other PU and/or non-objects. The mechanical sorter employed for conducting the mechanical sorting may be a valve block, mechanical fingers, robotic arms or, alternatively, a reverse belt setup.

In the method, the NIR and/or MIR spectrum of the one or more at least one irradiated PU objects may be acquired as multi- or hyperspectral data.

The method may include a pre-sorting preparation such as a pre-sorting step that can be carried out manually, mechanically, and/or sensor-based for removal of non-plastic objects and/or non-PU materials. Additionally, or alternatively, the method may include cleaning, partial drying, shredding and/or grading, e.g., prior to providing the waste stream comprising the several PU objects.

Preferably, in the method, a first data driven model is used for determining based on the recorded a NIR and/or MIR spectrum whether the one or more irradiated PU objects exhibit the at least one predefined property. Additionally, or alternatively, in the method, a first data driven model is used for determining based on the recorded a NIR and/or MIR spectrum an amount of the at least one impurity within the one or more irradiated PU objects.

The first data-driven model may be a first classification model such as a first neural network that is trained for determining whether the one or more irradiated PU objects exhibit the at least one predefined property and/or an amount of the at least one impurity within the one or more irradiated PU objects, a vision transformer that is configured for determining whether the one or more irradiated PU objects exhibit the at least one predefined property and/or an amount of the at least one impurity within the one or more irradiated PU objects or the like. The first trained neural network can be a multi-scale neural network or a recurrent neural network (RNN) such as, but not limited to, a gated recurrent unit (GRU) recurrent neural network or a long short-term memory (LSTM) recurrent neural network. Alternatively, the first neural network may be a convolutional neural network (CNN). BASF SE

220246

The first neural network may be trained using training data, e.g., comprising one or more reference NIR and/or MIR spectra of one or more reference PU objects exhibiting the predefined property and comprising a known amount of the at least one impurity. For example, if the first neural network is a feed-forward neural network such as a CNN, a backpropagation-algorithm may be applied for training the first neural network. In case of a RNN, a gradient descent algorithm or a back-propagation-through-time algorithm may be employed for training purposes. As a result of the training, operating parameters for the first neural network circuitry are generated such that when receiving a recorded NIR and/or MIR spectrum of the one or more at least one irradiated PU objects as input, the trained first neural network outputs a first output result indicative of whether the one or more irradiated PU objects exhibit the at least one predefined property and/or whether the one or more at least one irradiated PU objects are classified into a first group or the second group.

Additionally or alternatively, in the method, a statistical method may be used for determining based on the recorded a NIR and/or MIR spectrum whether the one or more irradiated PU objects exhibit the at least one predefined property. Optionally, in the method, a statistical method may be used for determining based on the recorded a NIR and/or MIR spectrum an amount of the at least one impurity within the one or more irradiated PU objects. A statistical method used in the method may be a method of least squares.

The method may comprise determining based on the recorded a NIR and/or MIR spectrum an amount of the at least one impurity within the one or more irradiated PU objects. If it is determined that the one or more irradiated PU objects exhibit the at least one predefined property, in the method, the one or more irradiated PU objects may be classified into

- the first group, if the determined amount of the eat least one impurity is equal to or below a predefined impurity threshold, and/or

- the second group, if the determined amount of the at least one impurity is above the predefined impurity threshold.

With the method it is thus possible to decide whether the one or more irradiated PU objects comprise an impurity at an amount that is too large for recycling the one or more irradiated PU objects. In this case, although the one or more irradiated PU objects may fulfil the at least one predefined property, in general, they will still be sorted into a group of unwanted objects since they contain the impurity at an amount that is too high for reuse of the one or BASF SE

220246

more irradiated PU objects. The impurity threshold may thus be set in accordance with processing requirement of a planned recycling process for the one or more irradiated PU objects. For example, the impurity threshold may be set to an amount of 10%, 20 %, 30 %, 40 % or 50 % of the total mass of the one or more irradiated PU objects.

Preferably, the method may comprise

- capturing an image showing the one or more irradiated PU objects,

- using a second data driven model for determining based on the captured image whether the one or more irradiated PU objects exhibit the at least one predefined property, and

- for classifying the one or more irradiated PU objects into the first group or the second group, taking into account an output result of the second data driven model indicative of whether the one or more irradiated PU objects exhibit the at least one predefined property.

It is thus preferred that the method comprises an optional sorting step or a step to remove objects comprising a certain amount of the at least one impurity, to further improve the purity of the PU material by type. Impurities such as metals may be removed using an electromagnetic (EM) detector. Furthermore, in addition or as an alternative, sorting may be achieved based on a water content of the one or more PU objects, e.g., employing a loss-of-drying (LOD) sensor. Moreover, improved purity of the PU material may be achieved using laser object detection and/or a visible (VIS) sensor, e.g., a camera, for object recognition to determine size, height, and/or colour of the one or more irradiated PU objects to more effectively sort the various PU types. Accordingly, sorting may be based on a determined size, height, and/or colour of the one or more irradiated PU objects. For analysing the captured image, the second data driven model is employed. The second data-driven model may be a second classification model such as a second neural network that is trained for determining based on the captured image whether the one or more irradiated PU objects exhibit the at least one predefined property, a vision transformer that is configured for determining based on the captured image whether the one or more irradiated PU objects exhibit the at least one predefined property or the like. The second trained neural network can be a multi-scale neural network or a recurrent neural network (RNN) such as, but not limited to, a gated recurrent unit (GRU) recurrent neural network or a long short-term memory (LSTM) recurrent neural network. Alternatively, the second neural network may be a convolutional neural network (CNN). BASF SE

220246

The first neural network may be trained using training data, e.g., comprising reference images showing one or more irradiated reference PU objects. For example, if the second neural network is a feed-forward neural network such as a CNN, a backpropagation- algorithm may be applied for training the second neural network. In case of a RNN, a gradient descent algorithm or a back-propagation-through-time algorithm may be employed fortraining purposes. As a result of the training, operating parameters for the second neural network circuitry are generated such that when receiving a captured image showing the one or more irradiated PU objects as input, the trained second neural network outputs a second output result indicative of whether the one or more irradiated PU objects exhibit the at least one predefined property and/or whether the one or more at least one irradiated PU objects are classified into a first group or the second group.

For example, the method may include a training of a second neural network stored on a computer-readable storage medium to classify objects in a bulk flow as described in EP 3 920 082 A1 which is incorporated herein by reference in its entity. Accordingly, a method of training a second neural network stored on a computer-readable storage medium to classify objects in a bulk flow may be included. The method of training a second neural network comprises the steps of:

- providing input image data depicting objects to be classified, which input image data is captured by means of an input imaging sensor of a first sensor technology design;

- providing auxiliary image data, which auxiliary image data is captured by means of an auxiliary imaging sensor of a second sensor technology design, and which auxiliary image data depicts said or similar objects which are classified in accordance with a predetermined classifying scheme;

- by means of a processing unit, train the second neural network stored on the computer- readable storage medium to classify the depicted objects in the input image data based on classifications of depicted objects in the auxiliary image data, wherein the depicted objects in the input image data correspond to objects in a bulk flow, and wherein the second sensor technology design is different from the first sensor technology design.

For instance, the second neural network may be trained to be able to classify objects depicted in input image data captured by means of an RGB-sensor based on classifications of depicted objects in auxiliary image data captured by means of a NIR and/or MIR sensor BASF SE

220246

or an X-ray sensor. Thus, the trained neural network may be configured to classify depicted objects in a satisfactory manner using much cheaper sensors.

For example, the first sensor technology design and the second sensor technology design may be selected from a group of sensor technology designs including NIR and/or MIR infrared sensor, X-ray sensor, CMYK-sensor, RGB-sensor, a volumetric sensor, point measurement system for spectroscopy, visible light spectroscopy, NIR spectroscopy, MIR spectroscopy, X-ray fluorescence sensors, electromagnetic sensors, laser sensor such as line laser triangulation system or a scanned laser for scattered laser, multispectral systems using LED’s, pulsed LED’s or lasers, LIBS (laser induced breakdown spectroscopy), Fluorescence detection, detectors for visible or invisible markers, transmission spectroscopy, transflectance/intreractance spectroscopy, softness measurement, thermal camera, and/or wherein the first sensor technology design and the second sensor technology design may be of the same general sensortechnology design but have different qualitative differences.

Herein, the term “impurity” is to be understood broadly and generally refers to anything that is present in the irradiated PU object next to the intentionally included components of the PU object such as a contaminant, pollutant, or dirt. That is, an “impurity” can be any substance that was unintentionally included into the PU object during the manufacturing process or after the manufacturing process. Generally, a PU object is manufactured according to a certain recipe including various different predefined components at predefined amounts that shall form the PU object. An “impurity” may thus be any substance that is not comprised in the recipe based on which a PU object has been or is to be manufactured. An “impurity” can be added to or can contaminate the PU object already during the manufacturing or after manufacturing. An “impurity” is thus an unwanted substance that is unintentionally added at any point in time to the PU object. In particular, an “impurity” is a compound or element that may lead to a modification in the recorded a NIR and/or MIR spectrum of the irradiated one or more PU objects. The modification of a recorded a NIR and/or MIR spectrum particularly relates to a recorded a NIR and/or MIR spectrum for the “clean” PU object, i.e., the PU object without comprising the impurity. The “clean” PU object thus refers to a PU object that only includes those compounds or elements that are intentionally included into the PU object through the manufacturing process. Preferably, the at least one impurity is one of water, blood, urine, dust, or styreneacrylonitrile copolymers (SAN). For example, an “impurity” may be the water content of a PU foam or a specific contamination such as blood or urine or SAN. An “impurity” may also include a modification of the one or more PU objects caused by external light, e.g., sunlight, or by dust, or the like that can be recognized in recorded a NIR- and/or MIR spectrum . BASF SE

220246

That is, am “impurity” can also be an unwanted chemical modification of a component of the PU object that is created by electromagnetic radiation, e.g., external light.

Preferably, the predefined property of the one or more PU objects is a property of the PU object that is intentionally realised, e.g., as part of or through a manufacturing process of the PU object such as a predefined chemical composition or a predefined chemical or physical behaviour of the PU object. In particular, the predefined property of the one or more PU objects may be a PU type or composition, an added compound type present in the one or more PU objects or a ratio of ingredients of the one or more PU objects. For example, a PU type or composition may be toluene diisocyanate (TDI), hexamethylene diisocyanate (HDI), 1 ,4-Butanediol (BDO), polyalkylene glycols, poly(tetramethylene ether) glycol, methylene diphenyl diisocyanate (MDI). An added compound type may be a blowing agent of a PU foam. A blowing agent may be carbon dioxide, pentane, 1 , 1 ,1 ,2- tetrafluoroethane (HFC-134a) and 1 ,1 ,1 ,3,3-pentafluoropropane (HFC-245fa), HFC - 141 B. Furthermore, an added compound type may be a chain extenders (f = 2) or a cross linker (f > 3). Chain extenders may be ethylene glycol, 1 ,4-butanediol (1 ,4-BDO or BDO), 1 ,6- hexanediol, cyclohexane dimethanol and hydroquinone bis(2-hydroxyethyl) ether (HQEE). Furthermore, an added compound type may be a surfactant such as polydimethylsiloxanepolyoxyalkylene block copolymers, silicone oils, nonylphenol ethoxylates, and other organic compounds. Furthermore, an added compound type may be a catalyst such as alkyl tin carboxylates, oxides and mercaptides oxides. Amine catalysts such as triethylenediamine (TEDA, also called DABCO, 1 ,4-diazabicyclo[2.2.2]octane), dimethylcyclohexylamine (DMCHA), dimethylethanolamine (DMEA), and bis-(2- dimethylaminoethyl)ether, a blowing catalyst also called A-99. A Lewis acidic catalyst such as dibutyltin dilaurate. Ingredients of the one or more irradiated PU objects may be di- and tri-isocyanates and polyols. Common polyols are, e.g., selected from the group consisting of polyether polyols, polyester polyols, polyetherester polyols and mixtures thereof.

Polyetherols are by way of example produced from epoxides, for example propylene oxide and/or ethylene oxide, or from tetrahydrofuran with starter compounds exhibiting hydrogenactivity, for example aliphatic alcohols, phenols, amines, carboxylic acids, water, or compounds based on natural substances, for example sucrose, sorbitol or mannitol, with use of a catalyst. Mention may be made here of basic catalysts and double-metal cyanide catalysts, as described by way of ex-ample in WO 2006/034800, EP 0090444, or WO 2005/090440.

Polyesterols are by way of example produced from aliphatic or aromatic dicarboxylic acids and polyhydric alcohols, polythioether polyols, polyesteramides, hydroxylated polyacetals, BASF SE

220246

and/or hydroxylated aliphatic polycarbonates, preferably in the presence of an esterification catalyst. Other possible polyols are mentioned by way of example in "Kunststoffhandbuch [Plastics hand-book], volume 7, Polyurethane [Polyurethanes]", Carl Hanser Verlag, 3rd edition 1993, chapter 3.1.

Preferably, in the method, for providing the waste stream comprising the several PU objects, a PU block is shredded to obtain the several PU objects.

Preferably, the one or more irradiated PU objects to be classified have a spatial extension in at least one direction of 30 mm to 5mm, preferably, of 30 mm to 300 mm, e.g., of 30 mm to 50 mm. In case the one or more irradiated PU objects are made of or comprise foam, they may also be referred to as foam elements.

The present invention also relates to a classification system for classifying one or more PU objects from a waste stream comprising several PU objects according to at least one predefined property, wherein the one or more PU objects may comprise at least one impurity, the classification system comprising:

- a PU object providing unit that is configured for providing the waste stream comprising the several PU objects,

- a radiation source that is configured for irradiating the one or more PU objects of the several PU objects with NIR and/or MIR,

- a sensor unit that is configured for recording a NIR and/or MIR spectrum of the one or more irradiated PU objects by detecting NIR and/or MIR reflected from the one or more irradiated PU objects,

- an evaluation unit that is configured for determining based on the recorded a NIR and/or MIR spectrum whether the one or more irradiated PU objects exhibit the at least one predefined property while verifying a possible presence of the at least one impurity in the one or more irradiated PU objects, and

- a classification unit that is configured classifying the one or more irradiated PU objects into a first group, if it is determined that the one or more irradiated PU objects exhibit the at least one predefined property and else, for classifying the one or more irradiated PU objects BASF SE

220246

into a second group, if it is determined that the one or more irradiated PU objects do not exhibit the at least one predefined property.

The evaluation unit is further configured for determining whether the one or more irradiated PU objects exhibit the at least one predefined property comprises comparing the recorded NIR and/or MIR spectrum to one or more reference NIR and/or MIR radiation spectra of one or more reference PU objects exhibiting the predefined property and comprising a known amount of the at least one impurity.

With the classification system flexible PU objects such as PU foams may be reliably sorted according to the at least one predefined property. Thereby, the classification system may achieve a throughput of 0.5 to 30 tons per hour/m at an accuracy level of higher than 90 % with a single stage throughput, and a purity level of higher than 95% utilizing a cascade of the system. It will be appreciated that the purity level achieved utilizing a cascade or sorting steps may be directly impacted by the ratio of impurities to PU objects in the mixed waste stream. For example, with the method it is possible to achieve with a 50:50 ratio of impurities:PU objects, at a throughput of 0.5 to 30 tons per hour/m at an accuracy level higher than 90 % with a single stage throughput.

The classification system according to the invention may be used for conducting the method for classifying one or more PU objects from a waste stream comprising several PU objects according to at least one predefined property disclosed herein. In particular, the irradiation of the one or more PU objects of the several PU objects with NIR- and/or MIR, and the recording of a NIR and/or MIR spectrum of the one or more irradiated PU objects by detecting NIR- and/or MIR reflected from the one or more irradiated PU objects may be carried as disclosed in WO 2021/249698 A1 , which is incorporated herein by reference in its entirety.

The radiation source and the sensor unit of the classification may form a spectroscopy system as described in WO 2021/249698 A1. The spectroscopy system may detect the one or more PU objects in a detection area. For example, the one or more PU objects can be provided on a conveyor belt and the detection area is a predefined area on the conveyor belt.

Accordingly, the radiation source may include a first light source adapted to emit the first set of light beams of NIR- and/or MIR and a second light source adapted to emit a second set of light beams of NIR and/or MIR. By this arrangement, a more intense illumination may BASF SE

220246

be provided when irradiating the one or more PU objects. Further, the illumination of the one or more PU objects may easily be tailored by using different types of light sources having different characteristics as the first and second light sources. Furthermore, a more robust classification system may be achieved. The classification system may not need to be taken out of operation if one of the first and second light sources fails and may consequently still be operated during exchange of one of the light sources.

The classification system may further comprise in addition to a spectroscopy system a laser object detection system such as a lasertriangulation system. The triangulation system may detect PU objects of the several PU objects of the waste stream in a detection zone. The detection zone may be a predefined area on a conveyor belt and may overlap with the detection area or may not overlap with the detection area.

The laser triangulation system may include a laser arrangement adapted to emit a line of laser light towards a detection zone through which the waste stream comprising the several PU objects is provided. The laser arrangement typically includes one or more laser light sources and optionally optical elements for forming emitted laser light into a line of laser light.

The laser triangulation system may include a camera-based sensor arrangement configured to receive and analyse light, which is reflected and/or scattered by matter in the detection zone. The received light of the camera-based sensor arrangement may be originating or predominantly originating from the line of laser light. Hence, a limited amount of ambient light may still reach the camera-based sensor arrangement. The camera-based sensor arrangement may thus be adapted such that it views the detection zone in order to receive and analyse light, which is reflected and/or scattered by the waste stream in the detection zone. Like in any laser triangulation system, the reflected light of the line of laser light will move on the sensor element of the camera-based sensor arrangement in response to a height variation of the waste stream in the detection zone. A detected shift owing form the angle difference between the field of view of the camera of the camera-based sensor arrangement and the line of laser light. Various properties of the waste stream in the detection zone may be determined based on measurements carried out by the camerabased sensor arrangement. The sensor element of the camera-based sensor arrangement is typically an imaging sensor element including an array of light sensitive sensor pixels.

The classification system may further comprise a focusing arrangement, wherein the focusing arrangement is adapted to direct and focus the first set of light beams and the second set of light beams on a scanning element. The scanning element may be adapted BASF SE

220246

to redirect the first and second sets of light beams towards the first detection zone, whereby the first and second set of light beams converge at the first detection zone. This arrangement provides the advantage that the one or more PU objects may be illuminated by different sets of light beams entering the one or more PU objects at different angles. The one or more PU objects may thus be efficiently illuminated by the first set of light beams and the second set of light beams converging at the one or more PU objects.

The classification system may further include a first optical filter arranged between the radiation source and a detection area in which the one or more PU objects are to be irradiated. The first optical filter may be configured for counteracting light originating from the first set of light beams and the second set of light beams from reaching the camerabased sensor arrangement. This arrangement of the first optical filter may counteract undesired light that otherwise would risk disturbing the camera-based sensor system form reaching the same. The provision of the first optical filter may be advantageous when the detection area and the detection zone overlap.

The classification system may further comprise an ejection arrangement coupled to a processing unit, wherein the ejection arrangement is adapted to eject and sort PU objects into the first group orthe second group in response to receiving a signal form the processing unit based on whether the one or more irradiated PU objects have been classified into the first group or the second group. The ejection arrangement being adapted to eject and sort said matter by means of at least one of a jet of compressed air, a jet of pressurized water, a mechanical finger, a bar of jets of compressed air, a bar of jets of pressurized water, a bar of mechanical fingers, a robotic arm and a mechanical diverter. By the provision of an ejection arrangement coupled to the processing unit, the classification system may eject and thus sort the several PU objects from the waste stream into a plurality of fractions. Hence, the waste stream may be sorted based on analysis performed by the evaluation unit. The ejection arrangement may also include a mechanical sorter that could be a valve block, alternatively a reverse belt setup.

The classification system may include pre-sorting preparation unit for a manual, mechanical and/or sensor-based removal of non-plastic objects and/or non-PU materials. The presorting preparation unit may be configured for cleaning, partial drying, shredding and grading of PU objects.

The PU object providing unit may comprise a conveyor belt and/or chute feeding means for transporting the PU objects. BASF SE

220246

The sensor unit may be or may comprise an NIR and/or MIR scanner. The NIR and/or MIR scanner may acquire the NIR and/or MIR spectra as multi- or hyperspectral data.

The classification unit may provide the classification result for the one or more irradiated PU objects to a valve block, mechanical fingers, robotic arms or a reverse belt setup for sorting the one or more irradiated PU objects into the first group or the second group. For example, the classification unit may provide the classification result to a classification system’s ejection means for sorting the one or more irradiated PU objects.

With the classification system, it is possible to implement a pre-recycling process, using a sensor-based sorting step in the pre-recycling process to obtain a high conversion yield and high product quality process for recycling of PU foams. In particular, the pre-recycling process for recycling of a PU material may include:

- one or more pre-sorting steps to produce a first pre-sorted fraction;

- at least one shredding step whereby the first pre-sorted fraction is shredded into smaller pieces to produce a second pre-sorted fraction;

- at least one separation step wherein the second pre-sorted fractions is separated at into target size fractions;

- at least one vision-based sorting stage wherein at least one desired PU fraction is recovered from the target size fractions, the vision-based sorting stage characterized therein that:

- the target size fraction is fed as an input stream to at least one machine vision device;

- the one or more machine vision devices being equipped with a conveyer belt and/or chute feeding means allowing feeding the input stream to the machine vision device, e.g., at a rate of at least 0.5 to 30 tons per hour/m;

- the one or more machine vision devices further equipped with at least one light source capable of homogeneous distribution of light to the input stream; BASF SE

220246

- the one or more machine vision devices further equipped with at least one first NIR sensor and at least one camera sensor and optionally at least one VIS spectroscopy sensor and/or at least one laser sensor;

The one or more machine vision devices may optionally equipped with at least one high- sensitivity electromagnetic (EM) sensor. The one or more machine vision devices may further equipped with at least one processing unit configured to identify at least one desired PU type according to a classifier algorithm whereby a material type spectral analysis and/or a colour spectral analysis and/or an image analysis in combination with a continuous machine learning analysis or any combination thereof may be used to signal an instruction to an ejection means in operable connection to the machine vision device. With the ejection means, based on the received instructions, effecting sorting the one or more irradiated PU objects from a non-desired fraction with a purity level of at least 95% and a throughput of 0.5 to 30 tons per hour/m may be achieved.

An optional ejection means may comprise a mechanical sorter configured to sort the one or more irradiated PU objects received by the classification system, where the mechanical sorter in operation is arranged to target the one or more irradiated PU objects with a preset NIR spectral scan data and an optionally pre-set water content value to be sorted as a set of wanted object type and the one or more irradiated PU objects not within the pre-set NIR spectral scan data and an optionally pre-set water content value to be sorted or as a set of unwanted object type.

Preferably, the classification system’s evaluation unit comprises a first data driven model and/or a statistical model that is configured for determining based on the recorded a NIR and/or MIR spectrum

- whether the one or more irradiated PU objects exhibit the at least one predefined property, and/or

- an amount of the at least one impurity within the one or more irradiated PU objects. It is thus preferred that the scan data may be analysed by applying machine learning and/or artificial intelligence to in real time optimise the sorting of the one or more irradiated PU objects, into at least one set of wanted objects and at least one set of unwanted objects. For example, the first data driven model may be classification model such as a first neural network. For example, the evaluation unit may comprise a processing circuit configured to analyse the reflected NIR and/or MIR radiation spectrum of the one or more irradiated PU BASF SE

220246

objects by inputting the multi- or hyperspectral data into a convolutional neural network with at least two convolutional layers in order to either detect and classify the one or more irradiated PU objects according to the at least one predefined property in the multi- or hyperspectral data and/or semantically segment the multi- or hyperspectral data. The way of neural network-based classification may be implemented as disclosed in WO 2021/089602 A1 , which is incorporated herein by reference in its entirety.

Preferably, the classification system’s sensor unit comprises a first and a second NIR and/or MIR radiation sensor that are configured for detecting a first and a second NIR and/or MIR spectrum, respectively, the first and a second NIR and/or MIR sensors having a different spectral absorption range.

Additionally or alternatively, the classification system’s sensor unit further comprises a LOD sensor that is configured for determining a water content of the one or more PU objects. From the recorded NIR and/or MIR spectra it is thus possible to also determine a water content can be determined. For example, the classification system may comprise a NIR and/or MIR scanner to acquire the reflected NIR and/or MIR radiation as multi- or hyperspectral data in combination with a LOD scanner to determine the water content of the one or more irradiated PU objects. Thereby, the classifying can be further improved.

Preferably, the classification system’s sensor unit further comprises a VIS spectroscopy sensor that is configured for capturing an image from the one or more PU objects and wherein the evaluation unit comprises a second data driven model that is configured for determining based on the captured image whether the one or more irradiated PU objects exhibit the at least one predefined property and for providing output result indicative of whether the one or more irradiated PU objects exhibit the at least one predefined property. For example, a camera for determining the size, height and/or colour of the at least one PU object may be employed as the VIS spectroscopy sensor or as part of the VIS spectroscopy sensor. For example, the VIS spectroscopy sensor can be used for pre-classifying to provide the number of PU object wherein the number of PU object have comparable dimensions. Thereby, the classification into the first group or the second group can be further improved. Furthermore, a light source can be employed for irradiating the at least one PU object with visible light, preferably, with homogeneous light distribution. The VIS spectroscopy sensor may be employed in an additional sorting steps to improve the purity of the PU material by type via object recognition to determine size and/or height, as well as for colour detection to more effectively sort the various PU types. BASF SE

220246

Preferably, the classification system’s PU object providing unit comprises a shredding unit that is configured for shredding a PU block to obtain the several PU objects. A PU block made of or comprising foam may also be referred to as a “foam piece” or a “foam article”, e.g., a mattress, an arm chair, or a seat such as a car seat. Foam pieces may referto input foam-containing material collected at collection and/or pre-sorting facilities. A foam piece may be shredded to form PU objects, such as a foam elements. Shredding the foam pieces may result in foam elements with a maximum dimension typically in the range of 100 mm to 500 mm.

The present invention also relates to a computer program comprising instructions for classifying one or more PU objects from a waste stream comprising several PU objects according to at least one predefined property. The one or more PU objects may comprise at least one impurity. The computer program is configured for:

- receiving a recorded NIR and/or MIR spectrum of the one or more PU objects that have been irradiated with NIR and/or MIR radiation,

- if it is determined that the one or more irradiated PU objects exhibit the at least one predefined property, classifying the one or more irradiated PU objects into a first group, and else, if it is determined that the one or more irradiated PU objects do not exhibit the at least one predefined property, classifying the one or more irradiated PU objects into a second group, wherein determining whether the one or more irradiated PU objects exhibit the at least one predefined property comprises comparing the recorded NIR and/or MIR spectrum to one or more reference NIR and/or MIR spectra of one or more reference PU objects exhibiting the predefined property and comprising a known amount of the at least one impurity, and

- providing an output result indicative of whether one or more irradiated PU objects have been classified into the first group or the second group.

The present invention relates further to a non-transitory computer-readable data medium storing the computer program according described herein. BASF SE

220246

It shall be understood that the aspects described above, and specifically the method of claim 1 , the classification system of claim 8 and the computer program of claim 14, have similar and/or identical preferred embodiments, in particular as defined in the dependent claims.

It shall be understood that a preferred embodiment of the present invention can also be any combination of the dependent claims or above embodiments with the respective independent claim.

These and other aspects of the present invention will be apparent from and elucidated with reference to the embodiments described hereinafter.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 schematically and exemplarily shows a classification system that is configured for classifying PU objects provided in a waste stream;

Fig. 2 schematically and exemplary shows a classification system that further comprises a shredding unit and an ejection arrangement;

Fig. 3 schematically and exemplary shows a classification system that comprises a VIS spectroscopy sensor and a loss-on-drying sensor;

Fig. 4 shows a flow diagram representing a method for classifying one or more

PU objects from a waste stream comprising several PU objects according to at least one predefined property, the one or more PU objects may comprise at least one impurity;

Fig. 5 shows a flowchart diagram representing a method of sorting PU foams objects;

Fig. 6 shows three curves representing a change in the water content in wt.-% for three different PU foam sample 722 EF, 724 EF and 732 EF, respectively, as a function of time and drying at a constant temperature of 80 °C;

Fig. 7 shows the normalized reflection intensity of six recorded NIR spectra recorded for the sample 722EF as a function of a first wavelength interval; BASF SE

220246

Fig. 8 shows the normalized reflection intensity of six recorded NIR spectra recorded for the sample 724EF as a function of a first wavelength interval;

Fig. 9 shows the normalized reflection intensity of six recorded NIR spectra recorded for the sample 732EF as a function of a first wavelength interval;

Fig. 10 shows the normalized reflection intensity of three recorded NIR spectra recorded for the samples 722 EF, 724 EF and 732 EF, respectively, having different SAN contents and a water content of zero;

Fig. 11 shows the normalized reflection intensity of six recorded NIR spectra recorded for the sample 722 EF as a function of a second wavelength range that is different to the first wavelength interval;

Fig. 12 shows the normalized reflection intensity of six recorded NIR spectra recorded for the sample 724 EF as a function of a second wavelength range that is different to the first wavelength interval; and

Fig. 13 shows the normalized reflection intensity of six recorded NIR spectra recorded for the sample 732 EF as a function of a second wavelength interval that is different to the first wavelength interval.

DETAILED DESCRIPTION OF EMBODIMENTS

Figure 1 schematically and exemplarily shows a classification system 100 that is configured for classifying PU objects 102 provided in a waste stream 104 according to at least one predefined property. The at least one predefined property may be a PU type such as TDI or MDI, a PU composition such as a ratio of TDI and MDI, or a presence of an added compound such as a certain surfactant.

The PU objects 102 may comprise one or more of ether, polyols, TDI and MDI. Preferably, tolylene 2,4- and/or 2,6-diisocynate (TDI) or a mixture thereof, monomeric diphenylmethane diisocyanates, and/or diphenylmethane diisocyanate homologs having a larger number of rings (polymer MDI), and mixtures of these. Other possible isocyanates are mentioned by way of example in "Kunststoffhandbuch [Plastics handbook], volume 7, Polyurethane [Polyure-thanes]", Carl Hanser Verlag, 3rd edition 1993, chapter 3.2 and 3.3.2. BASF SE

220246

Moreover, the PU objects 102 may comprise an impurity such as water, blood, urine, SAN, dust or the like.

The PU objects 102 can be provided with a conveyor belt 106 that is or is part of a PU object providing unit. With the conveyor belt 106, the PU objects 102 can be transported to a detection area in which the PU objects 102 can be irradiated with NIR and/or MIR radiation 108 emitted by a radiation source 110. NIR and/MIR 1 12 reflected from the irradiated PU objects 1 14 can be detected with a sensor unit 116. These sensor unit 116 may comprise an NIR scanner to acquire a NIR and/or MIR spectrum 1 18 as multi- or hyperspectral data representing the detected NIR and/or MIR 112 reflected from the irradiated PU objects 114.

The sensor unit 1 16 is connected to an evaluation unit 120 for example via a cable or wirelessly for transmitting the recorded NIR and/or MIR spectrum 1 18 to the evaluation unit 120. The evaluation unit 120 is configured for determining based on the recorded NIR and/or MIR spectrum 118 whether the irradiated PU objects 114 exhibit at least one predefined property. Furthermore, the evaluation unit 120 is configured for checking whetherthe irradiated PU objects 114 comprise at least one impurity. Thereby, it is possible to determine whetherthe irradiated PU objects 114 really exhibit the at least one predefined property. That is, due to the presence of an impurity in the irradiated PU objects 114, a NIR and/or MIR 118 recorded from NIR and/or MIR radiation reflected from the irradiated PU objects 114 may include a modification, i.e., leading to a modified NIR and/or MIR spectrum.

Considering a situation in which the irradiated PU objects 114 indeed exhibit the at least one predefined property, based on the modified recorded NIR and/or MIR spectrum from these PU objects 114, it may be not possible to reliably classify the irradiated PU objects 114 to indeed exhibit the at least one predefined property. For improving the classification process, it is therefore beneficial to take into account the possibility that the irradiated PU objects 114 may contain the at least one impurity. Taking into account a possible presence of the at least one impurity in the irradiated PU objects 114 can be achieved in various ways. In particular, the recorded NIR and/or MIR spectrum 118 may be compared to a plurality of different reference NIR and/or MIR spectra that each are associated with the at least one predefined property and different amounts of the at least one impurity, respectively.

By matching the recorded NIR and/or MIR spectrum 118 to the reference NIR and/or MIR spectra, it can be determined whetherthe recorded NIR and/or MIR radiation spectrum 1 18 BASF SE

220246

is associated with a particular reference NIR and/or MIR spectrum from the database. If a match is found, it is possible to classify the object to indeed exhibit the at least one predefined property and, in addition, it is possible to determine whether the irradiated PU objects 114 also comprise the at least one impurity. It may even be possible to determine at which amount the irradiated PU objects 1 14 comprise the at least one impurity in case the amount of the at least one impurity associated with the respective reference NIR and/or MIR spectrum is known.

The comparison of the recorded NIR and/or MIR spectrum 1 18 to reference NIR and/or MIR spectra can also be achieved using a first data driven model that is part of the evaluation unit 120. The data driven model may be a neural network that is trained for receiving the recorded NIR and/or MIR spectrum 118 as input and for determining whether the irradiated PU objects 114 exhibit the at least one predefined property and whether the irradiated PU objects 1 14 also comprise the at least one impurity. For example, the first data driven model may be trained with a plurality of reference NIR and/or MIR spectra of one or more reference PU objects exhibiting the predefined property and comprising a known amount of the at least one impurity.

The classification system 100 further comprises a classification unit 122 that is connected to the evaluation unit 120, e.g., via a cable or wirelessly. The classification unit 122 can receive from the evaluation unit 120 whether the irradiated PU objects 114 exhibit the at least one predefined property and also whether the irradiated PU objects 114 comprise the at least one impurity. It is also possible that the classification unit 122 receives from the evaluation unit 120 the amount of the at least one impurity present in the irradiated PU objects 114.

In case the evaluation unit 120 has determined that the irradiated PU objects 114 exhibit the at least one predefined property, the classification unit 122 may classify the irradiated PU objects 114 into a first group. Thereby, the classification unit 122 may take into account an amount of the at least one impurity detected to be present in the irradiated PU objects 114. For example, the classification unit 122 may consider an impurity threshold and compare the determined amount of the at least one impurity to the impurity threshold. In case the determined amount of the at least one impurity exceeds the impurity threshold, the classification unit 122 may be configured to not to classify the irradiated PU objects 114 into the first group although the evaluation unit 120 has determined that the irradiated PU objects 114 exhibit the at least one predefined property. In this case, the irradiated PU objects 114 may be classified into a second group. Moreover, if the evaluation unit 120 has determined that the irradiated PU objects 114 do not exhibit the at least one predefined BASF SE

220246

property, the classification unit 122 may be configured to classify the irradiated PU objects 114 into the second group.

Figure 2 schematically and exemplary shows a classification system 200 that is configured similar to the classification system 100 as described with respect to figure 1. Accordingly, the classification system 200 also comprises a conveyor belt 206 as part of a PU object providing unit 204, a radiation source 210, a sensor unit 216, an evaluation unit 220 and a classification unit 222. These components may be configured as described forthe conveyor belt 106, the radiation source 110, the sensor unit 216, the evaluation unit 120 and the classification unit 122 described with reference to figure 1 , respectively.

The classification system 200 further comprises a shredding unit 224 that is configured for shredding a PU block, for example, PU foam waste such as a mattress comprising PU foam, into several PU objects 202. Herein, the term “PU foam waste” includes end-of-life PU foams and production rejects of PU foams or waste generated through further processing of PU foams. “Production rejects of polyurethane foams" denotes PU foam waste occurring in production processes of PU foams.

The shredding unit 224 is part of the PU object providing unit 204. In operation, the shredding unit may be shredding a PU block and the resulting PU objects 202 can be provided to the conveyor belt 206 that transports the PU objects 202 to a detection area.

The sensor unit 216 comprises a first NIR and/or MIR sensor 226 and a second NIR and/or MIR sensor 228. The first and second NIR and/or MIR sensors 226, 2028 work at different NIR and/or MIR spectral absorption ranges. Thereby, it is possible to provide at least two distinct NIR and/or MIR spectra from the irradiated PU objects 214. This allows to more accurately determine due to better statistics whether the irradiated PU objects 214 exhibit the at least one predefined property. For example, for each of the recorded first and second NIR and/or MIR spectra, it can be determined with the evaluation unit 220 whether the irradiated PU objects 214 exhibit the at least one predefined property. In case the analysis of both spectra yields the same result, a respective output signal indicative of the result may be provided. Thereby the reliability can be further increased.

The evaluation unit 120 comprises a first data driven model 230 that is configured, for example trained, for determining whether the irradiated PU objects 214 exhibit the at least one predefined property. Additionally, the first data driven model 230 is configured for BASF SE

220246

determining an amount of the at least one impurity that may be present in the irradiated PU objects 214.

The classification unit 222 is connected to an ejection arrangement 232. The ejection arrangement 232 is arranged and configured for sorting the PU objects 202 transported along the conveyor belt 206 into the first group or the second group according to the classification result provided by the classification unit 222. To this end, in operation, the ejection arrangement 232 may be controlled according to the classification performed by the classification unit 222 to sort the classified irradiated PU objects 214 into the first group if they exhibit the at least one predefined property or into the second group if they do not exhibit the at least one predefined property. For example, the ejection arrangement 232 may be configured to sort the irradiated PU objects 214 employing a jet of compressed air, a jet of pressurized water, a mechanical finger, a bar of jets of compressed air, a bar of jets of pressurized water, a bar of mechanical fingers, a robot arm or a mechanical diverter. For example the ejection arrangement 232 may comprise a control unit, e.g., comprising processing circuitry for processing a classification result provided by the classification unit 222 for controlling the sorting of the irradiated PU objects 214, accordingly.

Figure 3 schematically and exemplary shows a classification system 300 that comprises the same components as the classification system 100 described with reference to figure

1. Accordingly, the classification system 300 comprises a conveyor belt 306 as part of a PU object providing unit 304, a radiation source 310, a sensor unit 316, an evaluation unit 320 and a classification unit 322. In addition, the classification unit 300 comprises a shredding unit 324 and an ejection arrangement 332 that are configured the same way as the shredding unit 224 and the ejection arrangement 232 described with reference to figure

2.

Furthermore, classification system 300 comprises a VIS spectroscopy sensor 334 that is configured for capturing an image from the PU objects 314 that have been transported to a detection area where they are also irradiated with NIR and/or MIR emitted from the radiation source 310. The images captured by the VIS spectroscopy sensor 334, in operation, are transmitted to the evaluation unit 320. The evaluation unit 320 comprises a second data driven model 336 that is configured, for example trained, for determining based on the captured image whether the irradiated PU objects 314 exhibit the at least one predefined property. For example, the VIS spectroscopy sensor 334 may comprise a camera. Based on the captured image a size, height, and/or colour of the irradiated PU objects 314 can be determined. This information can be further used for classifying the irradiated PU objects 314 according to the at least one predefined property. For example, BASF SE

220246

based on the colour of the irradiated PU objects 314, with the second data driven model 336, the PU type or composition may be determined. Moreover, based on the size or height of the irradiated PU objects 314, a classification into the first group of the second group may be achieved. For example, only those irradiated PU objects 314 may be classified into the first group that exhibit the at least one predefined property and also have a size or height within the predefined dimension range. The evaluation unit 336 may thus take into account for the determination of whether the irradiated PU objects 314 exhibit the at least one predefined property the captured image provided by the VIS spectroscopy sensor 334 and analysed by the second data driven model 336 as well as the recorded NIR and/or MIR spectrum analysed by the first data driven model 320. Based on the combined information, the evaluation unit 320 may be configured for providing an output result indicative of whether the irradiated PU objects 314 exhibit the at least one predefined property.

The classification system’s sensor unit 316 further comprises a loss-on-drying sensor 338. The loss-on-drying sensor 338 is configured for determining a water content of the irradiated PU objects 314. Water may be the at least one impurity that may potentially be present in the irradiated PU objects 314. The information provided by the loss-on-drying sensor 338 may be used as further input, e.g., to the first data driven model 320, for determining whether the irradiated PU objects 314 exhibit the at least one predefined property. In addition, based on the loss-on-drying sensor 338 measurement, the evaluation unit 320, for example, with the first data driven model 320, may determine an amount of the at least one impurity, i.e., the water content, of the irradiated PU objects 314.

Figure 4 shows a flow diagram representing a method for classifying one or more PU objects from a waste stream comprising several PU objects according to at least one predefined property, the one or more PU objects may comprise at least one impurity. The method can be conducted using one of the classification system 100, 200, 300 described with reference to figure 1 , 2 and 3, respectively.

In the method, optionally, a PU block is shredded into several PU objects (step S1). The PU block may be a mattress made of or comprising PU foam. The several PU objects are then provided in a waste stream to a detection area (step S2). In the detection area, one or more PU objects of the several PU objects are irradiated with NIR and/or MIR radiation (step S3). From NIR and/or MIR reflected from the one or more irradiated PU objects, a NIR and/or MIR spectrum is recorded (step S4). Based on the recorded a NIR and/or MIR spectrum it is then determined whether the one or more irradiated PU objects exhibit the at least one predefined property. At the same time it is a possible presence of the at least one impurity in the one or more irradiated PU objects verified (step S5). For determining BASF SE

220246

whetherthe irradiated PU objects exhibit the at least one predefined property, the recorded NIR and/or MIR spectrum is compared to one or more reference NIR and/or MIR spectra of one or more reference PU objects exhibiting the predefined property and comprising a known amount of the at least one impurity.

In particular, when sorting of different PU foam types, e.g., high resilient and/or conventional PU foams, it is possible to detect a presence and preferably even an amount of water, SAN or another impurity in the irradiated PU objects. For example, with a first NIR and/or MIR sensor and a NIR and/or MIR second sensor that work at different spectral absorption ranges, it is possible to detect the water content, e.g. of a PU foam, by using the first sensor and to detect an SAN content by using the first or second sensor. As the water content has a significant impact on the spectral absorption of PU foams, the detection may be used to significantly improve a sorting process and to achieve high quality end products without the need to dry the PU foams before the sorting process.

For example, a first data driven model can be used for determining based on the recorded a NIR and/or MIR spectrum whether the one or more irradiated PU objects exhibit the at least one predefined property. Additionally, or alternatively, the first data driven model can be used for determining based on the recorded a NIR and/or MIR spectrum an amount of the at least one impurity within the one or more irradiated PU objects. In addition, or as an alternative to the first data driven model, a statistical method can be used for determining based on the recorded a NIR and/or MIR spectrum whether the irradiated PU objects exhibit the at least one predefined property. Furthermore, the statistical method can be used for determining based on the recorded a NIR and/or MIR spectrum an amount of the at least one impurity within the one or more irradiated PU objects.

After having determined whether the irradiated PU objects exhibit the at least one predefined property, the irradiated PU objects are classified into a first or group or a second group. In particular, if it is determined that the irradiated PU objects exhibit the at least one predefined property they are classified into the first group (step S6). However, if it is determined that the irradiated PU objects do not exhibit the at least one predefined property, they are classified into the second group (step S7).

Optionally, in the method it is possible to capture an image showing the irradiated PU objects, e.g., with a VIS spectroscopy sensor. A second data driven model can be sued for determining based on the captured image whether the irradiated PU objects exhibit the at least one predefined property. This classification can be further improved by using the information about whether the irradiated PU objects exhibit the at least one predefined BASF SE

220246

property gained from the captured image through the second data driven model in addition to the classification based on the recorded NIR and/or MIR of the irradiated PU objects.

Figure 5 shows a flowchart diagram representing a method of sorting PU foams objects. In the method, a PU foam block is provided (step T1). The PU foam block is an old mattress and comprises at least one of conventional PU, high resilient (HR), hypersoft, combustion modified conventional (CME), combustion modified HR (CMHR), high load bearing (HLB) or the like. In addition, the PU foam block may comprise viscoelastic foam and/or latex foam. The PU foam block is shredded into PU foam objects (step T2).

The shredded PU foam objects are then transported using a conveyor belt to a detection area (step T3). The PU foam objects may have impurities, e.g., different humidity levels. The PU foam objects are illuminated with NIR and/or MIR to record a NIR and/or MIR spectrum from the reflected NIR and/or MIR. Using a PU sorting classifier algorithm, that may comprise a first data driven mode such as a first neural network may be used for determining from the record a NIR and/or MIR spectrum whether the PU foam objects exhibit the at least one predefined property (step T4).

Based on whether the PU foam objects exhibit the at least one predefined property, the PU foam objects are classified into a first group “positive fraction” (step T5) or if the PU foam objects do not exhibit the at least one predefined property they are classified into a second group “negative fraction” (step T6). For example, an air-pressure sorting nit can be used for sorting the PU foam objects. For example, the “positive fraction” may comprise conventional PU and humid conventional PU foams whereas the “negative fraction” may comprise high resilient, hypersoft, CME, CMHR and HLB.

Figure 6 shows three curves 600, 602, 604 representing a change in the water content in wt.-% for three different PU foam sample 722 EF, 724 EF and 732 EF, respectively, as a function of time and at a constant temperature of 80 °C. The PU foam sample 722 EF, 724 EF and 732 EF are made of high resilient foam and comprise MDI. Moreover, the PU foam sample 722 EF, 724 EF and 732 EF have an SAN content of 0 %, 5.6 % and 12.9 %, respectively.

It can be overserved that the water content decrease steadily within the given period of 60 minutes. For example, the water content of 722EF decreases from 153 wt.-% to 32 wt.-%, the water content of 724EF decreases from 135 wt.-% to 30 wt.-%, and the water content of 732EF decreases from 212 wt.-% to 71 wt.-%. BASF SE

220246

Figure 7 shows the normalized reflection intensity of six recorded NIR spectra recorded for the sample 722 EF as a function of a first wavelength range. The six recorded NIR spectra were recorded at a different time instance with a time difference of 15 minute with respect to a respective previously recorded NIR spectrum. Accordingly, the NIR spectrum 700 is associated with a time period of 15 minutes, the NIR spectrum 702 is associated with a time period of 30 minutes, the NIR spectrum 704 is associated with a time period of 45 minutes and the NIR spectrum 706 is associated with a time period of 60 minutes. The NIR spectrum 704 is concealed by the NIR spectrum 706. Furthermore, the NIR spectrum 708 represents a measurement on a "dry" sample, i.e., where only with air moisture is present.

Moreover, the NIR spectrum 710 is associated with a wet sample 722 EF, cf. figure 6. From these measurements, it can be clearly observed that the change in water content has a significant impact on the recorded NIR spectrum for the sample 722 EF.

Figure 8 shows the normalized reflection intensity of six recorded NIR spectra recorded for the sample 724 EF as a function of a first wavelength range. The NIR spectrum 800 is associated with a time period of 15 minutes, the NIR spectrum 802 is associated with a time period of 30 minutes, the NIR spectrum 804 is associated with a time period of 45 minutes and the NIR spectrum 806 is associated with a time period of 60 minutes. Furthermore, the NIR spectrum 808 represents a measurement on a "dry" sample.

Moreover, the NIR spectrum 810 is associated with a wet sample 724 EF, cf. figure 6. Also from these measurements, it can be clearly observed that the change in water content has a significant impact on the recorded NIR spectrum for the further sample 724 EF.

Figure 9 shows the normalized reflection intensity of six recorded NIR spectra recorded for the sample 732 EF as a function of a first wavelength interval. The NIR spectrum 900 is associated with a time period of 15 minutes, the NIR spectrum 902 is associated with a time period of 30 minutes, the NIR spectrum 904 is associated with a time period of 45 minutes and the NIR spectrum 906 is associated with a time period of 60 minutes. Furthermore, the NIR spectrum 908 represents a measurement on a "dry" sample, only with air moisture present. BASF SE

220246

Moreover, the NIR spectrum 910 is associated with a wet sample 732 EF, cf. figure 6. Also from these measurements, it can be clearly observed that the change in water content has a significant impact on the recorded NIR spectrum for the further sample 732 EF.

Figure 10 shows the normalized reflection intensity of three recorded NIR spectra 1000, 1002, 1004 recorded for the samples 722 EF, 724 EF and 732 EF, respectively. These samples have different SAN contents. That is the sample 722 EF has an SAN content of 0 %, the sample 724 EF has an SAN content of 5.6 % and the sample 732 EF has an SAN content of 12.9 %. All samples 722 EF, 724 EF and 732 EF have the same water content of 0 wt.-%. A water content of 0 wt.-%. means that only air moisture is present in the samples. The difference in the SAN contents results in an individual modification of the respective NIR spectrum that is characteristic for the respective amount of SAN in the respective samples 722 EF, 724 EF and 732 EF.

Figure 11 shows the normalized reflection intensity of six recorded NIR spectra recorded for the sample 722 EF as a function of a second wavelength range that is different to the first wavelength interval, cf. Fig. 7. The NIR spectrum 1100 is associated with a time period of 15 minutes, the NIR spectrum 1102 is associated with a time period of 30 minutes, the NIR spectrum 1104 is associated with a time period of 45 minutes and the NIR spectrum 1106 is associated with a time period of 60 minutes. Furthermore, the NIR spectrum 1 108 represents a measurement on a "dry" sample. Moreover, the NIR spectrum 1110 is associated with a wet sample 722 EF, cf. figure 6. From these measurements, in addition to Fig. 7, it can be clearly observed that the change in water content has a significant impact on the recorded NIR spectrum for the sample 722 EF.

Figure 12 shows the normalized reflection intensity of six recorded NIR spectra recorded for the sample 724 EF as a function of a second wavelength range that is different to the first wavelength interval, cf. Fig. 8. The NIR spectrum 1200 is associated with a time period of 15 minutes, the NIR spectrum 1202 is associated with a time period of 30 minutes, the NIR spectrum 1204 is associated with a time period of 45 minutes and the NIR spectrum 1206 is associated with a time period of 60 minutes. Furthermore, the NIR spectrum 1208 represents a measurement on a "dry" sample. Moreover, the NIR spectrum 1210 is associated with a wet sample 724 EF, cf. figure 6. From these measurements, in addition to Fig. 8, it can be clearly observed that the change in water content has a significant impact on the recorded NIR spectrum for the sample 724 EF.

Figure 13 shows the normalized reflection intensity of six recorded NIR spectra recorded for the sample 732 EF as a function of a second wavelength interval that is different to the BASF SE

220246

first wavelength interval, cf. Fig. 9. The NIR spectrum 1300 is associated with a time period of 15 minutes, the NIR spectrum 1302 is associated with a time period of 30 minutes, the NIR spectrum 1304 is associated with a time period of 45 minutes and the NIR spectrum 1306 is associated with a time period of 60 minutes. Furthermore, the NIR spectrum 1308 represents a measurement on a "dry" sample. Moreover, the NIR spectrum 1310 is associated with a wet sample 732 EF, cf. figure 6. Also from these measurements, in addition to Fig. 9, it can be clearly observed that the change in water content has a significant impact on the recorded NIR spectrum for the further sample 732 EF.

Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims.

For the processes and methods disclosed herein, the operations performed in the processes and methods may be implemented in differing order. Furthermore, the outlined operations are only provided as examples, and some of the operations may be optional, combined into fewer steps and operations, supplemented with further operations, or expanded into additional operations without detracting from the essence of the disclosed embodiments.

In summary, the present invention relates to a method and to a classification system for classifying PU objects from a waste stream comprising several PU objects according to at least one predefined property, wherein the PU objects may comprise at least one impurity. The PU objects are irradiated with NIR and/or middle-infrared radiation and a NIR and/or middle-infrared radiation spectrum is recorded by detecting NIR and/or middle-infrared radiation. Based on the recorded a NIR and/or middle-infrared radiation spectrum, it is determined whetherthe irradiated PU objects exhibit the predefined property while verifying a possible presence of the at least one impurity in the one or more irradiated PU objects. This comprises that the recorded NIR and/or middle-infrared radiation spectrum is compared to one or more reference NIR and/or middle-infrared radiation spectra of reference PU objects exhibiting the predefined property and comprising a known amount of the at least one impurity.

In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. BASF SE

220246

A single unit or device may fulfil the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

Procedures like determining based on the recorded a NIR and/or MIR spectrum whether the one or more irradiated PU objects exhibit the at least one predefined property while verifying a possible presence of the at least one impurity in the one or more irradiated PU objects, classifying the one or more irradiated PU objects into a first group, classifying the one or more irradiated PU objects into a second group etc. performed by one or several units or devices can be performed by any other number of units or devices. These procedures can be implemented as program code means of a computer program and/or as dedicated hardware.

A computer program product may be stored/distributed on a suitable medium, such as an optical storage medium or a solid-state medium, supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems.

Any units described herein may be processing units that are part of a classical computing system. Processing units may include a general-purpose processor and may also include a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), or any other specialized circuit. Any memory may be a physical system memory, which may be volatile, non-volatile, or some combination of the two. The term “memory” may include any computer-readable storage media such as a non-volatile mass storage. If the computing system is distributed, the processing and/or memory capability may be distributed as well. The computing system may include multiple structures as “executable components”. The term “executable component” is a structure well understood in the field of computing as being a structure that can be software, hardware, or a combination thereof. For instance, when implemented in software, one of ordinary skill in the art would understand that the structure of an executable component may include software objects, routines, methods, and so forth, that may be executed on the computing system. This may include both an executable component in the heap of a computing system, or on computer- readable storage media. The structure of the executable component may exist on a computer-readable medium such that, when interpreted by one or more processors of a computing system, e.g., by a processor thread, the computing system is caused to perform a function. Such structure may be computer readable directly by the processors, for instance, as is the case if the executable component were binary, or it may be structured to be interpretable and/or compiled, for instance, whether in a single stage or in multiple BASF SE

220246

stages, so as to generate such binary that is directly interpretable by the processors. In other instances, structures may be hard coded or hard-wired logic gates, that are implemented exclusively or near-exclusively in hardware, such as within a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), or any other specialized circuit. Accordingly, the term “executable component” is a term for a structure that is well understood by those of ordinary skill in the art of computing, whether implemented in software, hardware, or a combination. Any embodiments herein are described with reference to acts that are performed by one or more processing units of the computing system. If such acts are implemented in software, one or more processors direct the operation of the computing system in response to having executed computerexecutable instructions that constitute an executable component. Computing system may also contain communication channels that allow the computing system to communicate with other computing systems over, for example, network. A “network” is defined as one or more data links that enable the transport of electronic data between computing systems and/or modules and/or other electronic devices. When information is transferred or provided over a network or another communications connection, for example, either hardwired, wireless, or a combination of hardwired or wireless, to a computing system, the computing system properly views the connection as a transmission medium. Transmission media can include a network and/or data links which can be used to carry desired program code means in the form of computer-executable instructions or data structures and which can be accessed by a general-purpose or special-purpose computing system or combinations. While not all computing systems require a user interface, in some embodiments, the computing system includes a user interface system for use in interfacing with a user. User interfaces act as input or output mechanism to users for instance via displays.

Those skilled in the art will appreciate that at least parts of the invention may be practiced in network computing environments with many types of computing system configurations, including, personal computers, desktop computers, laptop computers, message processors, hand-held devices, multi-processor systems, microprocessor-based or programmable consumer electronics, network PCs, minicomputers, main-frame computers, mobile telephones, PDAs, pagers, routers, switches, data centres, wearables, such as glasses, and the like. The invention may also be practiced in distributed system environments where local and remote computing system, which are linked, for example, either by hard-wired data links, wireless data links, or by a combination of hardwired and wireless data links, through a network, both perform tasks. In a distributed system environment, program modules may be located in both local and remote memory storage devices. | BASF SE

| 220246

Those skilled in the art will also appreciate that at least parts of the invention may be practiced in a cloud computing environment. Cloud computing environments may be distributed, although this is not required. When distributed, cloud computing environments may be distributed internationally within an organization and/or have components possessed across multiple organizations. In this description and the following claims, “cloud computing” is defined as a model for enabling on-demand network access to a shared pool of configurable computing resources, e.g., networks, servers, storage, applications, and services. The definition of “cloud computing” is not limited to any of the other numerous advantages that can be obtained from such a model when deployed. The computing systems of the figures include various components orfunctional blocks that may implement the various embodiments disclosed herein as explained. The various components or functional blocks may be implemented on a local computing system or may be implemented on a distributed computing system that includes elements resident in the cloud or that implement aspects of cloud computing. The various components or functional blocks may be implemented as software, hardware, or a combination of software and hardware. The computing systems shown in the figures may include more or less than the components illustrated in the figures and some of the components may be combined as circumstances warrant.

Any reference signs in the claims should not be construed as limiting the scope.

Claims

BASF SE 220246 CLAIMS

1. A method for classifying one or more polyurethane, hereinafter called PU, objects from a waste stream comprising several PU objects according to at least one predefined property, wherein the one or more PU objects may comprise at least one impurity, the method comprising the steps of:

- providing the waste stream comprising the several PU objects, irradiating the one or more PU objects of the several PU objects with nearinfrared radiation, hereinafter called NIR, and/or middle-infrared radiation, hereinafter called MIR, recording a NIR and/or MIR spectrum of the one or more irradiated PU objects by detecting NIR and/or MIR reflected from the one or more irradiated PU objects, determining based on the recorded a NIR and/or MIR spectrum whether the one or more irradiated PU objects exhibit the at least one predefined property while verifying a possible presence of the at least one impurity in the one or more irradiated PU objects, and if it is determined that the one or more irradiated PU objects exhibit the at least one predefined property, classifying the one or more irradiated PU objects into a first group, and else, if it is determined that the one or more irradiated PU objects do not exhibit the at least one predefined property, classifying the one or more irradiated PU objects into a second group, wherein determining whether the one or more irradiated PU objects exhibit the at least one predefined property comprises comparing the recorded NIR and/or MIR spectrum to one or more reference NIR and/or MIR spectra of one or more reference PU objects exhibiting the predefined property and comprising a known amount of the at least one impurity.

2. The method according to claim 1 , wherein a first data driven model is used for determining based on the recorded a NIR and/or MIR spectrum BASF SE

220246

whether the one or more irradiated PU objects exhibit the at least one predefined property, and/or an amount of the at least one impurity within the one or more irradiated PU objects.

3. The method according to claim 1 or 2, wherein a statistical method is used for determining based on the recorded a NIR and/or MIR spectrum whether the one or more irradiated PU objects exhibit the at least one predefined property, and/or an amount of the at least one impurity within the one or more irradiated PU objects.

4. The method according to at least one of the preceding claims, comprising determining based on the recorded a NIR and/or MIR spectrum an amount of the at least one impurity within the one or more irradiated PU objects and if it is determined that the one or more irradiated PU objects exhibit the at least one predefined property, classifying the one or more irradiated PU objects into the first group, if the determined amount of the eat least one impurity is equal to or below a predefined impurity threshold, and/or the second group, if the determined amount of the at least one impurity is above the predefined impurity threshold.

5. The method according to at least one of the preceding claims, comprising capturing an image showing the one or more irradiated PU objects, using a second data driven model for determining based on the captured image whether the one or more irradiated PU objects exhibit the at least one predefined property, and for classifying the one or more irradiated PU objects into the first group or the second group, taking into account an output result of the second data driven BASF SE

220246

model indicative of whether the one or more irradiated PU objects exhibit the at least one predefined property.

6. The method according to at least one of the preceding claims, wherein the at least one impurity is one of water, blood, urine, dust, or styrene-acrylonitrile copolymers.

7. The method according to at least one of the preceding claims, wherein the predefined property of the one or more PU objects is a PU type or composition, an added compound type present in the one or more PU objects or a ratio of ingredients of the one or more PU objects.

8. A classification system for classifying one or more PU objects from a waste stream comprising several PU objects according to at least one predefined property, wherein the one or more PU objects may comprise at least one impurity, the classification system comprising: a PU object providing unit that is configured for providing the waste stream comprising the several PU objects, a radiation source that is configured for irradiating the one or more PU objects of the several PU objects with NIR and/or middle-infrared radiation, a sensor unit that is configured for recording a NIR and/or MIR spectrum of the one or more irradiated PU objects by detecting NIR and/or MIR reflected from the one or more irradiated PU objects, an evaluation unit that is configured for determining based on the recorded a NIR and/or MIR spectrum whether the one or more irradiated PU objects exhibit the at least one predefined property while verifying a possible presence of the at least one impurity in the one or more irradiated PU objects, and a classification unit that is configured classifying the one or more irradiated PU objects into a first group, if it is determined that the one or more irradiated PU objects exhibit the at least one predefined property and else, for classifying the one or more irradiated PU objects into a second group, if it is determined that the one or more irradiated PU objects do not exhibit the at least one predefined property, wherein BASF SE

220246

the evaluation unit is further configured for determining whether the one or more irradiated PU objects exhibit the at least one predefined property comprises comparing the recorded NIR and/or MIR spectrum to one or more reference NIR and/or MIR spectra of one or more reference PU objects exhibiting the predefined property and comprising a known amount of the at least one impurity.

9. The classification system according to claim 8, wherein the evaluation unit comprises a first data driven model and/or a statistical model that is configured for determining based on the recorded a NIR and/or MIR spectrum whether the one or more irradiated PU objects exhibit the at least one predefined property, and/or

- an amount of the at least one impurity within the one or more irradiated PU objects.

10. The classification system according to claim 8 or 9, wherein the sensor unit comprises a first and a second NIR and/or MIR sensor that are configured for detecting a first and a second NIR and/or MIR spectrum, respectively, the first and a second NIR and/or MIR sensors having a different spectral absorption range.

11. The classification system according to at least one of claim 8 to 10, wherein the sensor unit further comprises a loss-on-drying sensor that is configured for determining a water content of the one or more PU objects.

12. The classification system according to at least one of claim 8 to 11 , wherein the sensor unit further comprises a VIS spectroscopy sensor that is configured for capturing an image from the one or more PU objects and wherein the evaluation unit comprises a second data driven model that is configured for determining based on the captured image whether the one or more irradiated PU objects exhibit the at least one predefined property and for providing output result indicative of whetherthe one or more irradiated PU objects exhibit the at least one predefined property.

13. The classification system according to at least one of claim 8 to 12, wherein the PU object providing unit comprises a shredding unit that is configured for shredding a PU block to obtain the several PU objects. BASF SE

220246

14. A computer program comprising instructions for classifying one or more PU objects from a waste stream comprising several PU objects according to at least one predefined property, wherein the one or more PU objects may comprise at least one impurity, the computer program being configured for: receiving a recorded NIR and/or MIR spectrum of the one or more PU objects that have been irradiated with NIR and/or MIR, determining based on the recorded a NIR and/or MIR spectrum whether the one or more irradiated PU objects exhibit the at least one predefined property while verifying a possible presence of the at least one impurity in the one or more irradiated PU objects, and if it is determined that the one or more irradiated PU objects exhibit the at least one predefined property, classifying the one or more irradiated PU objects into a first group, and else, if it is determined that the one or more irradiated PU objects do not exhibit the at least one predefined property, classifying the one or more irradiated PU objects into a second group, wherein determining whether the one or more irradiated PU objects exhibit the at least one predefined property comprises comparing the recorded NIR and/or MIR spectrum to one or more reference NIR and/or MIR spectra of one or more reference PU objects exhibiting the predefined property and comprising a known amount of the at least one impurity, and providing an output result indicative of whether one or more irradiated PU objects have been classified into the first group or the second group.

15. A non-transitory computer-readable data medium storing the computer program according to claim 14.