Method and system for separating waste polyurethane foams

1. A method for separating waste polyurethane foams, the method comprising recording at least one respective spectrum for each polyurethane sample of a supply stream comprising polyurethane samples from waste, wherein the at least one respective spectrum is recorded by near-infrared spectroscopy, wherein each polyurethane sample of the supply stream is classified by a classification algorithm, wherein the classification algorithm is based on machine learning, based on the respective at least one spectrum into a respective class of at least two classes, wherein the supply stream comprising polyurethane samples is separated into at least two streams according to the classification into the respective class, wherein at least one of the at least two streams is chemically recycled, and wherein each class corresponds to a type of polyurethane.

2. The method according to claim 1 , wherein at least one of the at least one respective spectrum is recorded in reflection mode.

3. The method according to claim 1 , wherein the at least one respective spectrum is recorded in a spectral range between 750 nm and 3500 nm.

4. The method according to claim 1 , wherein a respective average diameter of the polyurethane samples of the supply stream in at least two orthogonal dimensions is between 0.1 cm and 50 cm.

5. The method according to claim 1 , wherein the at least two classes comprise at least one MDI class and at least one TDI class, wherein each polyurethane sample of the supply stream is classified into the at least one MDI class when the classification algorithm identifies that polyurethane sample as comprising chemical products of monomeric and/or polymeric methylene diphenyl diisocyanates and that each polyurethane sample of the supply stream is classified into the at least one TDI class when the classification algorithm identifies that polyurethane sample as comprising chemical products of toluene diisocyanates.

6. The method according to claim 5 , wherein the at least one TDI class comprises at least two TDI classes.

7. The method according to claim 6 , wherein each polyurethane sample of the supply stream is classified into one of the at least two TDI classes based on a type of polyol mixture identified by the classification algorithm used for the production of that respective polyurethane sample.

8. The method according to claim 6 , wherein the at least two TDI classes comprise a high-resilience TDI class and/or a viscoelastic TDI class.

9. The method according to claim 6 , wherein the at least two TDI classes comprise a conventional TDI class and/or an other TDI class.

10. The method according to claim 1 , wherein the type of classification algorithm is a partial least squares regression, a logistic regression, a k-nearest neighbor algorithm, an adaptive boosting algorithm, a gradient boosting algorithm, a random forest algorithm or an artificial neural network algorithm.

11. The method according to claim 1 , wherein the classification algorithm employs a distance metric, an independent component analysis, a principal component analysis, or any combination thereof.

12. The method according to claim 1 , wherein the classification algorithm is trained with a set of reference spectra of respective reference polyurethane samples.

13. The method according to claim 12 , wherein each reference spectrum of the set of reference spectra is recorded by spectroscopy from a respective reference polyurethane sample.

14. A system for separating waste polyurethane foams, the system comprising a spectroscopy apparatus configured to record for each polyurethane sample of a supply stream comprising polyurethane samples from waste at least one respective spectrum, wherein the system further comprises a computing arrangement on which a classification algorithm based on machine learning and configured to classify each polyurethane sample of the supply stream into a respective class of at least two classes is executed, which classification is based on the respective at least one spectrum, the system further comprising a separating apparatus for separating the supply stream comprising polyurethane samples into at least two streams according to the classification into the respective class, wherein each class corresponds to a type of polyurethane, wherein the system further comprises a recycling module for chemically recycling at least one of the at least two streams.

15. The method according to claim 2 , wherein at least one of the at least one respective spectrum is recorded with a spatial resolution of less than 5 cm.

16. The method according to claim 15 , wherein at least one of the at least one respective spectrum is recorded with a spatial resolution of less than 1.5 cm.

17. The method according to claim 3 , wherein the at least one respective spectrum is recorded in a spectral range between 900 nm and 2500 nm.

18. The method according to claim 3 , wherein the at least one respective spectrum is recorded at a recording rate greater than 10 Hz.

19. The method according to claim 18 , wherein the at least one respective spectrum is recorded at a recording rate in the range of 300 to 600 Hz.

20. The method according to claim 4 , wherein a respective average diameter of the polyurethane samples of the supply stream in at least two orthogonal dimensions is between 3 cm and 15 cm.