A polyester fiber analysis method and system based on infrared spectrum

Abstract

The application discloses a polyester fiber analysis method and system based on infrared spectrum, and relates to the technical field, which comprises the following steps: acquiring a PET sample; performing sample analysis on the PET sample, collecting infrared spectrum data; extracting peak intensity or area characteristic data of a plurality of characteristic peaks in the corresponding infrared spectrum data; and constructing a PET recycling analysis model, inputting the characteristic data into the PET recycling analysis model to obtain qualitative and quantitative analysis results. Four key characteristic peaks highly related to the recycling process in the PET spectrum are selected as a minimum characteristic set, and partial least squares discriminant analysis and partial least squares regression are combined to realize rapid identification and content quantification of the recycled PET. Without an expensive chemical analysis platform and complex sample pretreatment, the detection speed is fast, the sample form is wide, the detection cost is significantly reduced, the detection flux is improved, and the demand for recycled PET proportion verification, tracing and quality control in the textile, packaging and supervision scenes is met.

Description

Technical Field

[0001] This invention relates to the field of recycled material identification technology, and more specifically to a method and system for analyzing polyester fibers based on infrared spectroscopy. Background Technology

[0002] Polyester fiber (polyethylene terephthalate), abbreviated as PET, is one of the world's largest-produced and most widely used thermoplastic polyesters. It possesses excellent mechanical strength, chemical resistance, and processability, and is widely used in beverage bottle packaging, textiles, and engineering plastics. Currently, to reduce carbon emissions and promote resource recycling, the demand for PET recycling and reuse continues to grow. Recycled PET (r-PET) is mainly obtained through two process routes: one is mechanical recycling, which includes washing, crushing, melting and regranulation, and remolding. This process is simple and easy to scale up, but repeated thermal histories can cause molecular chain breakage, changes in end-group concentration, and alterations in crystallinity. The other is chemical recycling, such as glycololysis, methanololysis, and hydrolysis processes, which can depolymerize PET into monomers such as BHET, TPA, and EG, and then repolymerize them. The product performance is close to that of virgin resin, but the process cost, reaction control, and purification procedures are more complex.

[0003] Due to significant differences between virgin PET and recycled PET in terms of production costs, quality stability, traceability, market prices, and regulatory policies in the food contact materials sector, some unscrupulous operators may profit through adulteration or mislabeling, leading to market chaos. Particularly in the food and beverage packaging and disposable consumer goods industries, strict requirements have been imposed on the permissible proportions of recycled PET used in food contact materials, as well as the purity and quality safety indicators of recycled materials. Companies are encouraged to disclose the proportions of recycled components in accordance with regulations. Therefore, simply qualitatively distinguishing between virgin and recycled PET is no longer sufficient to meet the practical needs of regulation, corporate quality control, and supply chain transparency.

[0004] Existing techniques for the identification and quantitative analysis of virgin and recycled PET largely rely on destructive chemical analysis or complex spectroscopic methods. These include determining isomer ratios, end-group content, or impurity characteristics using chromatography, mass spectrometry, nuclear magnetic resonance, and infrared / near-infrared spectroscopy, and using models such as IPA / TPA molar ratio and HPLC-PCA / PLS to infer the source and content of recycled PET. While these methods are feasible under laboratory conditions, they generally suffer from high testing costs, cumbersome sample pretreatment, strong instrument dependence, high operational barriers, and low throughput, making them unsuitable for the practical needs of on-site sampling and large-scale quality supervision in the textile and plastic packaging industries. Furthermore, full-spectrum-dependent spectroscopic discrimination methods are sensitive to instrument differences and sampling conditions, lacking stable, minimally variable characteristics suitable for engineering implementation. Therefore, there is an urgent need in this field for a rapid, non-destructive or minimally invasive, low-cost, easy-to-operate, and standardizable qualitative and quantitative analysis technique for recycled PET.

[0005] Against this backdrop, establishing a rapid, non-destructive, and quantifiable method for accurately assessing the proportion of recycled PET blended in samples is of great significance for ensuring food packaging safety, regulating market order, implementing access requirements for recycled materials, and supporting enterprises in establishing traceability systems. Especially with the promotion of the "bottle-to-bottle" system for beverage bottles and the gradual increase in the use of food contact grade r-PET, precise quantitative analysis of recycled PET content has become a key technical requirement for industry quality control and supervision.

[0006] Therefore, how to propose a polyester fiber analysis method and system based on infrared spectroscopy to overcome the shortcomings of existing technologies is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0007] In view of this, the present invention provides a method and system for analyzing polyester fibers based on infrared spectroscopy. Using infrared spectroscopy as the testing method, and employing partial least squares discriminant analysis (PLS-DA) and partial least squares regression (PLS) for modeling, it achieves rapid, accurate, and non-destructive qualitative identification and quantitative analysis of recycled polyester fibers. To achieve the above objectives, the present invention adopts the following technical solution: An infrared spectroscopy-based method for analyzing polyester fibers, comprising: Obtain PET samples; The PET sample was analyzed, and infrared spectral data were collected. Extract peak intensity or area characteristic data of multiple characteristic peaks from the corresponding infrared spectral data; A PET recycling analysis model was constructed, and the characteristic data were input into the PET recycling analysis model to obtain qualitative and quantitative analysis results.

[0008] Optionally, the PET sample may comprise virgin PET and recycled PET.

[0009] Optionally, the extraction of peak intensity or area feature data of multiple characteristic peaks in the corresponding infrared spectral data includes: Extract the peak intensity or area characteristics of the four specified characteristic peaks in the corresponding infrared spectrum.

[0010] Optionally, the construction of the PET regeneration analysis model includes: coupling partial least squares discriminant analysis and partial least squares regression.

[0011] Optionally, the step of analyzing the PET sample and acquiring infrared spectral data includes: The test wavenumber range is 4000-600cm. - ¹, resolution is 2-4cm - ¹, the number of scans is n, and the characteristic spectral bands are the average of the n scans.

[0012] Optionally, it also includes background correction, baseline correction and absorption spectrum normalization processing of the acquired infrared spectral data.

[0013] Optionally, the plurality of characteristic peaks include: Bending vibrations of the CH bond in the benzene ring, bending vibrations of the CH bond in the benzene ring, symmetric stretching vibration bands of COC, and asymmetric stretching vibration bands of COC.

[0014] Optionally, the feature data includes: peak height, peak area, first derivative, and relative intensity ratio feature parameters, which constitute the input vector.

[0015] Optionally, the coupled partial least squares discriminant analysis and partial least squares regression include: Using the characteristic peak input vectors of known virgin PET and recycled PET samples, a partial least squares discriminant analysis model is constructed to output category labels; For samples with different blending ratios, a partial least squares regression model was established, with the characteristic peak input vector as input and the actual proportion of recycled PET as output. After collecting the infrared spectrum of the sample to be tested, the characteristic peak input vector is extracted and input into the partial least squares discriminant analysis model to make a qualitative determination of primary / regenerated PET. If it is determined to be regenerated or mixed, it is input into the partial least squares regression model to obtain the quantitative mixing ratio of regenerated PET.

[0016] Optionally, a polyester fiber analysis system based on infrared spectroscopy includes: Acquisition module: used to acquire PET samples; Preprocessing module: used to perform sample analysis on the PET sample and collect infrared spectral data; Feature extraction module: used to extract peak intensity or area feature data of multiple characteristic peaks in the corresponding infrared spectral data; Analysis module: Used to construct a PET recycling analysis model, and input the feature data into the PET recycling analysis model to obtain qualitative and quantitative analysis results.

[0017] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a method and system for analyzing polyester fibers based on infrared spectroscopy, which has the following beneficial effects: This invention proposes a polyester fiber analysis method based on infrared spectroscopy, comprising: acquiring a PET sample; performing sample analysis on the PET sample and collecting infrared spectral data; extracting peak intensity or area characteristic data of multiple characteristic peaks in the corresponding infrared spectral data; constructing a PET recycling analysis model and inputting the characteristic data into the PET recycling analysis model to obtain qualitative and quantitative analysis results.

[0018] This invention, by constructing a simplified spectral feature set based on four key infrared characteristic peaks and combining partial least squares discriminant analysis and regression models, achieves rapid and reliable qualitative identification and quantitative analysis of blending ratios between virgin and recycled PET, resulting in the following significant effects in practical applications: First, this invention significantly reduces detection costs and lowers technical barriers. The proposed method can complete the analysis using only conventional infrared spectroscopy equipment, eliminating the need for expensive chromatography, mass spectrometry, or nuclear magnetic resonance platforms, as well as complex chemical pretreatments such as dissolution, extraction, and purification, thus greatly reducing detection costs and shortening the analysis cycle.

[0019] Secondly, this invention significantly improves detection speed and throughput. Because it requires only the collection of a small number of characteristic peak variables, the model has low computational load and fast inference speed, enabling rapid discrimination at the second or minute level. It is suitable for batch quality sampling inspection of samples such as textiles, bottle flakes, and films, solving the high-frequency, high-throughput regulatory needs that traditional chemical analysis cannot meet.

[0020] Third, this invention has higher model stability and cross-scenario transferability. By defining four stable characteristic peaks that are highly correlated with regeneration and are less affected by instrument differences, the model constructed by this invention can maintain consistent discrimination performance under different instruments, different batches of samples, and different regeneration processes (mechanical regeneration, chemical regeneration).

[0021] Fourth, this invention enables precise quantification of the blending ratio of recycled PET, providing reliable quantitative data for food contact materials, bottle-to-bottle recycling systems, and fiber-grade green certification. This is of great significance for implementing regulatory requirements regarding the proportion of recycled PET used, food packaging safety, and the construction of traceability systems.

[0022] Fifth, this invention possesses excellent engineering feasibility. The measurement method is simple, the model structure is lightweight, and the variable set can be standardized, facilitating rapid deployment by upstream and downstream enterprises in the industry chain in quality inspection, production processes, supply chain audits, and law enforcement spot checks. This can significantly improve the transparency and quality consistency of recycled materials in the industry. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of a polyester fiber analysis method based on infrared spectroscopy provided by the present invention.

[0025] Figure 2 The PET infrared spectrum provided for this invention.

[0026] Figure 3 The identification diagram of virgin and mechanically recycled PET provided by this invention.

[0027] Figure 4 The identification diagram of virgin and chemically recycled PET provided by this invention.

[0028] Figure 5 A quantitative regression diagram of PET mixtures provided by the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] This invention discloses a method for analyzing polyester fibers based on infrared spectroscopy, such as... Figure 1 As shown, it includes: Obtain PET samples; The PET sample was analyzed, and infrared spectral data were collected. Extract peak intensity or area characteristic data of multiple characteristic peaks from the corresponding infrared spectral data; A PET recycling analysis model was constructed, and the characteristic data were input into the PET recycling analysis model to obtain qualitative and quantitative analysis results.

[0031] Furthermore, the PET sample comprises virgin PET and recycled PET.

[0032] Furthermore, the extraction of peak intensity or area feature data of multiple characteristic peaks in the corresponding infrared spectral data includes: Extract the peak intensity or area characteristics of the four specified characteristic peaks in the corresponding infrared spectrum.

[0033] Furthermore, the construction of the PET regeneration analysis model includes: coupled partial least squares discriminant analysis and partial least squares regression.

[0034] Furthermore, the sample analysis of the PET sample, including the acquisition of infrared spectral data, includes: The test wavenumber range is 4000-600cm. - ¹, resolution is 2-4cm - ¹, the number of scans is n, and the characteristic spectral bands are the average of the n scans.

[0035] Furthermore, it also includes background correction, baseline correction, and absorption spectrum standardization processing of the collected infrared spectral data.

[0036] Furthermore, the plurality of characteristic peaks include: Bending vibrations of the CH bond in the benzene ring, bending vibrations of the CH bond in the benzene ring, symmetric stretching vibration bands of COC, and asymmetric stretching vibration bands of COC.

[0037] Furthermore, the feature data includes: peak height, peak area, first derivative, and relative intensity ratio feature parameters, which constitute the input vector.

[0038] Furthermore, the coupled partial least squares discriminant analysis and partial least squares regression include: Using the characteristic peak input vectors of known virgin PET and recycled PET samples, a partial least squares discriminant analysis model is constructed to output category labels; For samples with different blending ratios, a partial least squares regression model was established, with the characteristic peak input vector as input and the actual proportion of recycled PET as output. After collecting the infrared spectrum of the sample to be tested, the characteristic peak input vector is extracted and input into the partial least squares discriminant analysis model to make a qualitative determination of primary / regenerated PET. If it is determined to be regenerated or mixed, it is input into the partial least squares regression model to obtain the quantitative mixing ratio of regenerated PET.

[0039] In a specific embodiment, a polyester fiber analysis system based on infrared spectroscopy includes: Acquisition module: used to acquire PET samples; Preprocessing module: used to perform sample analysis on the PET sample and collect infrared spectral data; Feature extraction module: used to extract peak intensity or area feature data of multiple characteristic peaks in the corresponding infrared spectral data; Analysis module: Used to construct a PET recycling analysis model, and input the feature data into the PET recycling analysis model to obtain qualitative and quantitative analysis results.

[0040] In a specific embodiment, this invention provides a polyester fiber analysis method based on infrared spectroscopy. Based on infrared spectroscopy and chemometric modeling, it achieves rapid qualitative identification and quantitative analysis of blending ratios between virgin and recycled PET. This method selects four characteristic peaks highly correlated with the recycling process from the mid-infrared absorption spectrum of PET as a minimum feature set. Combining partial least squares discriminant analysis (PLS-DA) and partial least squares regression (PLS), it achieves accurate prediction of material type identification and recycled component content. The specific technical implementation process is as follows: I. Sample Collection and Spectroscopic Testing The spectral data of the PET samples under test were acquired using an infrared spectrometer. The wavenumber range was 4000-600 cm⁻¹. - ¹, The preferred resolution is 2-4cm - ¹, The number of scans was 32, and the characteristic spectral bands were the average of the 32 scans. Background correction, baseline correction and absorption spectrum standardization were performed to ensure the comparability of samples from different batches and in different forms.

[0041] II. Selection of Key Characteristic Peaks and Feature Construction

[0042] Based on the spectral statistical patterns of a large number of virgin and recycled PET samples, four characteristic peaks with high discriminative power and high regeneration sensitivity were selected as core variables, including but not limited to: 865-885cm - ¹: Bending vibration of the CH bond in the benzene ring; 1010-1030cm - ¹: Bending vibration of the CH bond in the benzene ring; 1080-1160cm - ¹: The COC symmetric stretching vibration band is closely related to the changes in chain segment configuration during the regeneration process; 1220-1260cm - ¹: The asymmetric stretching vibration band of COC is closely related to the changes in chain segment configuration during the regeneration process; For the four peaks mentioned above, feature parameters such as peak height, peak area, first derivative, and relative intensity ratio can be extracted to form an input vector.

[0043] Specifically, a four-peak combination is used as the minimum feature set to ensure the model's lightweight nature, stability, and cross-instrument reproducibility.

[0044] III. Construction of the PLS-DA Qualitative Discriminant Model

[0045] A partial least squares discriminant analysis (PLS-DA) model is constructed using the characteristic peak input vectors of known virgin and recycled PET samples. During model training, cross-validation is performed on the data to optimize the number of latent variables, ensuring the model can stably distinguish between virgin and recycled samples. The PLS-DA outputs class labels to achieve qualitative identification of unknown samples.

[0046] IV. Establishment of PLS ​​Quantitative Model

[0047] Partial least squares regression models were established for samples with different blending ratios. The characteristic peak variable set was used as input, and the actual proportion of recycled PET was used as output to obtain a quantitative prediction model. The model used an external validation set to evaluate the prediction error, ensuring high quantitative accuracy under different sources and recycling process conditions. The different recycling processes included mechanical recycling and chemical recycling.

[0048] V. Detection Procedure for Unknown Samples

[0049] After collecting the infrared spectrum of the sample to be tested, the intensity parameters of the four characteristic peaks mentioned above are extracted and input into the PLS-DA model to achieve qualitative determination of virgin / regenerated; if it is determined to be regenerated or there is mixing, it is further input into the PLS regression model to obtain the quantitative mixing ratio of regenerated PET.

[0050] This embodiment of the method requires no dissolution, extraction, separation, or chemical derivatization, and the entire detection process can be completed within minutes. This embodiment, for the first time, screens and limits four characteristic peaks from the infrared spectrum of PET that are highly correlated with the regeneration process and have stable discriminative capabilities as a minimum feature set. Input variables are constructed using the peak height characteristics of these characteristic peaks. Combined with partial least squares discriminant analysis and partial least squares regression, rapid qualitative identification of virgin PET and recycled PET, as well as precise quantification of their blending ratio, are achieved. Compared with existing destructive chemical analysis methods relying on chromatography, mass spectrometry, and nuclear magnetic resonance, this invention significantly reduces detection costs and eliminates complex sample pretreatment. Compared with full-spectrum modeling methods, this invention uses fewer variable characteristic peaks for modeling, resulting in stronger model stability, transferability, and engineering feasibility. It maintains consistent analytical performance under different instruments, different batches of samples, and different regeneration process conditions. This invention offers advantages such as speed, non-destructive / minimally destructive operation, ease of operation, and on-site deployment, making it particularly suitable for the verification needs of recycled PET proportions in textile sampling, circulation supervision, and food contact materials.

[0051] In a specific implementation, an infrared spectroscopy-based method for analyzing polyester fibers is experimentally verified as follows: Example 1 1. Materials and Methods (1) Material selection Two types of 100% single-component PET were selected: virgin PET (n-PET) and mechanically recycled PET (m-rPET) with a purity ≥99%.

[0052] (2) Spectral acquisition

[0053] like Figure 2 As shown, ATR-FTIR was used, and the test conditions were as follows: wavenumber range: 4000-600cm. - ¹, Resolution: 4cm - ¹, Number of scans: 32, automatically finding peaks and extracting peak parameters within a fixed window. PLS-DA qualitative model, focusing on 865-885 cm⁻¹. - ¹、1010-1030cm - ¹、1080-1160cm - ¹、1220-1260cm - ¹Four feature windows, response variables: n-PET=1, m-rPET=0, latent variable LV=12.

[0054] 2. Results and Analysis

[0055] Qualitative classification performance: The PLS-DA model in the validation set showed an average classification accuracy of 100% with 5-fold cross-validation; indicating that the four-peak feature set in this embodiment can stably distinguish between purely mechanically recycled PET and virgin PET.

[0056] 3. Conclusion

[0057] like Figure 3 As shown, the PLS-DA model established based on four key infrared characteristic peaks in this embodiment can accurately distinguish between pure n-PET and pure m-rPET. The model is stable and has a clear response, and can be used for quality inspection and rapid supervision of recycled materials.

[0058] Example 2

[0059] 1. Materials and Methods

[0060] (1) Material selection

[0061] Two types of 100% single-component PET were selected: virgin PET (n-PET) and chemically recycled PET (c-rPET) with a purity ≥99%.

[0062] (2) Spectral acquisition

[0063] ATR-FTIR was used, and the test conditions were as follows: wavenumber range: 4000-600cm. - ¹, Resolution: 4cm -¹, Number of scans: 32, automatically finding peaks and extracting peak parameters within a fixed window. PLS-DA qualitative model, focusing on 865-885 cm⁻¹. - ¹、1010-1030cm - ¹、1080-1160cm - ¹、1220-1260cm - ¹Four feature windows, response variables: n-PET=1, c-rPET=0, latent variable LV=12.

[0064] 2. Results and Analysis

[0065] In the qualitative classification performance validation set, the PLS-DA model showed an average classification accuracy of 100% with 5-fold cross-validation. This demonstrates that the four-peak feature set in this embodiment can stably distinguish between pure chemically recycled PET and virgin PET.

[0066] 3. Conclusion

[0067] like Figure 4 As shown, the PLS-DA model, based on four key infrared characteristic peaks, can accurately distinguish between pure n-PET and pure c-rPET. The model is stable and has a clear response, and can be used for quality inspection and rapid supervision of recycled materials.

[0068] Example 3

[0069] 1. Materials and Methods

[0070] (1) Sample preparation

[0071] Two types of 100% single-component PET were selected: virgin PET (n-PET) and chemically recycled PET (c-rPET) with a purity ≥99%.

[0072] (2) Spectral acquisition

[0073] ATR-FTIR was used, and the test conditions were as follows: wavenumber range: 4000-600cm. - ¹, Resolution: 4cm - ¹, Number of scans: 32, automatically find peaks and extract peak parameters within a fixed window.

[0074] (3) Construct a simulated mixed spectrum. Randomly select one original spectrum and one regenerated spectrum from the pure original and regenerated spectra mentioned above, and linearly superimpose them at a random ratio (step size 0.01) within the range of 0-1 to generate a set of 1000 spectra covering the entire mixing ratio range.

[0075] (4) Construct a PLS regression model, focusing on the 865-885cm range. - ¹、1010-1030cm - ¹、1080-1160cm- ¹、1220-1260cm - ¹With four feature windows and a latent variable LV=30, this model is used to directly and quantitatively characterize the recycled content in PET mixtures.

[0076] 2. Results and Analysis

[0077] Regression model performance: The validation set results are as follows (typical values): R²=0.955, mean absolute error (MAE)=0.048, the model shows good linearity and no significant systematic bias.

[0078] 3. Conclusion

[0079] like Figure 5 As shown, the four characteristic peaks + PLS regression method proposed in this embodiment can directly and rapidly predict the recycled content of PET samples. The method in this embodiment has a simpler testing process, stronger engineering feasibility, and is suitable for actual industrial production and recycling process monitoring.

[0080] This invention proposes a method for identifying and quantifying the blending ratio of virgin PET and recycled PET based on infrared spectroscopy. This method involves screening four key characteristic peaks (865-885 cm⁻¹) in the PET spectrum that are highly correlated with the recycling process. - ¹、1010-1030cm - ¹、1080-1160cm - ¹、1220-1260cm - ¹) This method uses a minimal feature set, combined with partial least squares discriminant analysis and partial least squares regression, to achieve rapid identification and quantification of recycled PET content. It eliminates the need for expensive chemical analysis platforms and complex sample pretreatment, offering fast detection speeds and applicability to a wide range of sample types. This significantly reduces detection costs, increases throughput, and meets the practical needs of textile, packaging, and regulatory scenarios for verifying the proportion, traceability, and quality control of recycled PET, effectively overcoming the shortcomings of existing technologies.

[0081] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0082] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for analyzing polyester fibers based on infrared spectroscopy, characterized in that, include: Obtain PET samples; The PET sample includes virgin PET and recycled PET, and the recycled PET includes mechanically recycled PET and chemically recycled PET. The PET sample was analyzed, and infrared spectral data were collected. Extract peak intensity or area characteristic data of multiple characteristic peaks from the corresponding infrared spectral data; Extract the peak intensity or area characteristics of the four specified characteristic peaks in the corresponding infrared spectrum; The plurality of characteristic peaks include: 865-885cm-¹: Bending vibration of the CH bond in the benzene ring; 1010-1030cm-¹: Bending vibration of the CH bond in the benzene ring; 1080-1160cm-¹: COC symmetrical stretching vibration band; 1220-1260cm-¹: COC asymmetric stretching vibration belt; A PET recycling analysis model is constructed, and the characteristic data is input into the PET recycling analysis model to obtain qualitative and quantitative analysis results; The construction of the PET regeneration analysis model includes: coupling partial least squares discriminant analysis and partial least squares regression; The coupled partial least squares discriminant analysis and partial least squares regression include: Using the characteristic peak input vectors of known virgin PET and recycled PET samples, a partial least squares discriminant analysis model is constructed to output category labels; For samples with different blending ratios, a partial least squares regression model was established, with the characteristic peak input vector as input and the actual proportion of recycled PET as output. After collecting the infrared spectrum of the sample to be tested, the characteristic peak input vector is extracted and input into the partial least squares discriminant analysis model to make a qualitative judgment on whether it is virgin or regenerated. If it is judged to be regenerated or mixed, it is input into the partial least squares regression model to obtain the quantitative mixing ratio of regenerated PET.

2. The method for analyzing polyester fibers based on infrared spectroscopy according to claim 1, characterized in that, The sample analysis of the PET sample, including the acquisition of infrared spectral data, includes: The test wavenumber range is 4000-600cm. - ¹, resolution is 2-4cm - ¹, the number of scans is n, and the characteristic spectral bands are the average of the n scans.

3. The method for analyzing polyester fibers based on infrared spectroscopy according to claim 2, characterized in that, It also includes background correction, baseline correction, and absorption spectrum standardization processing of the collected infrared spectral data.

4. The method for analyzing polyester fibers based on infrared spectroscopy according to claim 1, characterized in that, The feature data includes peak height, peak area, first derivative, and relative intensity ratio feature parameters, which constitute the input vector.

5. A polyester fiber analysis system based on infrared spectroscopy, characterized in that, include: Acquisition module: used to acquire PET samples; the PET samples include virgin PET samples and recycled PET samples, the recycled PET including mechanically recycled PET samples and chemically recycled PET samples; Preprocessing module: used to perform sample analysis on the PET sample and collect infrared spectral data; Feature extraction module: used to extract peak intensity or area feature data of multiple characteristic peaks in the corresponding infrared spectral data; Extract the peak intensity or area characteristics of the four specified characteristic peaks in the corresponding infrared spectrum; The plurality of characteristic peaks include: 865-885cm-¹: Bending vibration of the CH bond in the benzene ring; 1010-1030cm-¹: Bending vibration of the CH bond in the benzene ring; 1080-1160cm-¹: COC symmetrical stretching vibration band; 1220-1260cm-¹: COC asymmetric stretching vibration belt; Analysis module: Used to construct a PET recycling analysis model, and input the feature data into the PET recycling analysis model to obtain qualitative and quantitative analysis results; The construction of the PET regeneration analysis model includes: coupling partial least squares discriminant analysis and partial least squares regression; The coupled partial least squares discriminant analysis and partial least squares regression include: Using the characteristic peak input vectors of known virgin PET and recycled PET samples, a partial least squares discriminant analysis model is constructed to output category labels; For samples with different blending ratios, a partial least squares regression model was established, with the characteristic peak input vector as input and the actual proportion of recycled PET as output. After collecting the infrared spectrum of the sample to be tested, the characteristic peak input vector is extracted and input into the partial least squares discriminant analysis model to make a qualitative judgment on whether it is virgin or regenerated. If it is judged to be regenerated or mixed, it is input into the partial least squares regression model to obtain the quantitative mixing ratio of regenerated PET.

Images