A micro-plastic multi-modal fusion traceability analysis method

Abstract

The application discloses a micro-plastic multi-modal fusion traceability analysis method. The method comprises the following steps: for micro-plastics from electronic waste: taking a part of micro-plastic particles for thermogravimetric analysis to obtain the peak temperature of the thermogravimetric first derivative curve, and measuring the additive release temperature interval and the plastic matrix release temperature interval in the micro-plastic particles; taking another part of micro-plastic particles for thermogravimetric analysis, switching the pyrolysis products to a gas chromatograph-mass spectrometer, and detecting the types and contents of additives according to the pyrolysis products generated in the additive release temperature interval, and detecting the plastic matrix material according to the pyrolysis products generated in the plastic matrix release temperature interval; combining the peak temperature, the additive release temperature interval, the plastic matrix release temperature interval, the types and contents of additives, the plastic matrix material, and the material type and shape characteristic parameters of the micro-plastic particles to determine the source and use of the micro-plastic particles. The application improves the accuracy of micro-plastic detection.

Description

Technical Field

[0001] This invention belongs to the field of environmental monitoring and materials analysis technology, and more specifically, relates to a method for multimodal fusion traceability analysis of microplastics. Background Technology

[0002] Microplastics have become a global environmental problem due to their environmental persistence, long-distance migration capabilities, and potential biotoxicity. The core prerequisite for source control and precision governance of microplastics lies in achieving accurate traceability of their origin. Microplastics from different sources exhibit significant differences in material composition, physical morphology, and additive characteristics. Currently, microplastic detection methods mainly rely on micro Raman spectroscopy, thermogravimetric-mass spectrometry (TGA), thermogravimetric-infrared spectroscopy (TGA), and pyrolysis-gas chromatography-mass spectrometry (PCMS). Spectroscopy can only obtain material information, not the morphological characteristics of microplastics, and cannot establish a correlation with their source. While mass spectrometry can detect the plastic matrix and its additives, the sample is consumed after a single test, making repeated verification impossible. Furthermore, the signals from the plastic matrix and additives easily overlap, making accurate separation and identification difficult.

[0003] Current research suffers from the following shortcomings: it cannot achieve one-to-one acquisition of multi-dimensional data such as images, spectra, and chemical composition of the same microplastic sample, making it difficult to establish effective correlations between various characteristics; it lacks a systematic understanding of the differences in shape characteristics of microplastics from different sources, making it impossible to use shape features to assist in source identification; the detection process is fragmented, making it difficult to integrate data from high-throughput screening, image acquisition, spectral detection, and component analysis to form a multi-dimensional fingerprint spectrum; and source identification is limited to a single dimension, lacking a multi-modal traceability model that integrates "shape + material + additives + pyrolysis behavior." These shortcomings mean that the accuracy of current microplastic detection still has significant room for improvement. Summary of the Invention

[0004] The main objective of this invention is to provide a multimodal fusion traceability analysis method for microplastics to overcome the shortcomings of existing technologies.

[0005] To achieve the above-mentioned objectives, the present invention adopts the following technical solution:

[0006] This invention provides a multimodal fusion-based source tracing analysis method for microplastics. The method includes: detecting and acquiring the material type and shape characteristic parameters of microplastic particles; determining whether the microplastic particles are microplastics from electronic waste based on the shape characteristic parameters; for microplastics from electronic waste: performing thermogravimetric analysis on a portion of the microplastic particles to obtain the peak temperature of the first derivative curve corresponding to the thermogravimetric curve; introducing pyrolysis gas into an infrared spectrometer and a mass spectrometer to measure the release temperature range of additives and the release temperature range of the plastic matrix in the microplastic particles, respectively; performing thermogravimetric analysis on another portion of the microplastic particles, switching the pyrolysis products to a gas chromatography-mass spectrometry (GC-MS) instrument, and detecting the type and content of additives based on the pyrolysis products generated under the additive release temperature range, and detecting the material of the plastic matrix based on the pyrolysis products generated under the plastic matrix release temperature range; and combining the material type, the shape characteristic parameters, the peak temperature, the additive release temperature range, the plastic matrix release temperature range, the type and content of additives, and the plastic matrix material to determine the source and use of the microplastic particles.

[0007] Preferably, the detection and acquisition of the material type and shape characteristic parameters of microplastic particles specifically includes: attaching filter paper covered with microplastic particles to a positioning grid below, and recording the grid coordinates of each microplastic particle; taking microscopic images of each microplastic particle, and extracting the shape characteristic parameters of the microplastic particles from the microscopic images; acquiring the spectrum of each microplastic particle using a Raman spectrometer to determine the material type of the microplastic particles; and establishing a correspondence between microplastic particles, shape characteristic parameters, and material type based on the grid coordinates.

[0008] Preferably, the shape feature parameters include at least roundness, rectangularity, aspect ratio, convexity, and edge smoothness; determining whether the microplastic particles are electronic waste-source microplastics based on the shape feature parameters specifically includes: inputting the roundness, rectangularity, aspect ratio, convexity, and edge smoothness of the microplastic particles into a preset screening model, and outputting a category label for the microplastic particles, wherein the category label is electronic waste-source microplastics or household-source microplastics.

[0009] Preferably, the kernel function of the screening model is:

[0010] ; in, For kernel function, To train the shape-material fusion feature vector of the i-th microplastic particle in the database, This is the shape-material fusion feature vector of the microplastic particles to be tested. , , , , , They are respectively The fingerprint parameters include roundness, rectangularity, aspect ratio, convexity, edge smoothness, and material origin. , , , , , They are respectively The fingerprint parameters include roundness, rectangularity, aspect ratio, convexity, edge smoothness, and material origin. The kernel width parameter. This is the roundness weighting coefficient. This is the rectangularity weighting coefficient. The aspect ratio weighting coefficient. For convexity weighting coefficients, This is the edge smoothness weighting coefficient. The fingerprint parameter weighting coefficient is used to determine the material source.

[0011] Preferably, the measurement process for the additive release temperature range and the plastic matrix release temperature range specifically includes: pyrolysis gas entering an infrared spectrometer to monitor the change of characteristic absorption peaks with temperature in real time, in order to determine the additive release temperature range; and pyrolysis gas entering a mass spectrometer to monitor the change of characteristic ion intensity with temperature in real time, in order to determine the plastic matrix release temperature range.

[0012] Preferably, another portion of the microplastic particles is subjected to thermogravimetric analysis, specifically including: setting a corresponding segmented heating program according to the release temperature range of the additive and the release temperature range of the plastic matrix; and under the heating control of the segmented heating program, taking another portion of the microplastic particles for thermogravimetric analysis.

[0013] Preferably, the upper limit of the additive release temperature range is lower than the lower limit of the plastic matrix release temperature range. The segmented heating procedure includes: heating the thermogravimetric analyzer to a first temperature at a first heating rate, the first temperature being lower than the lower limit of the additive release temperature range and differing from a first set temperature; heating the thermogravimetric analyzer from the first temperature to a second temperature at a second heating rate, the second temperature being higher than the upper limit of the additive release temperature range and differing from a second set temperature, the second heating rate being lower than the first heating rate; heating the thermogravimetric analyzer from the second temperature to a third temperature at a third heating rate, the third temperature being lower than the lower limit of the plastic matrix release temperature range and differing from a third set temperature, the third heating rate being higher than the first heating rate; and heating the thermogravimetric analyzer from the third temperature to a fourth temperature at a fourth heating rate, the fourth temperature being higher than the upper limit of the plastic matrix release temperature range and differing from a fourth set temperature, the fourth heating rate being lower than the third heating rate.

[0014] Preferably, the fusion determination of the source and use of the microplastic particles specifically includes: inputting the material type, the shape characteristic parameters, the peak temperature, the additive release temperature range, the plastic matrix release temperature range, the type and content of additives, and the plastic matrix material into a preset fusion discrimination model. The fusion discrimination model adopts a weighted voting mechanism to output the source and use of the microplastic particles.

[0015] Preferably, the fusion discrimination model is:

[0016] ; in, The result output by the fusion discriminant model. The weights of the d-th dimension information are... For the d-th dimension standard template information from the j-th source, For similarity functions; The information is d-th dimension. The first dimension includes the shape feature parameters, the second dimension includes the material type, the third dimension includes the peak temperature, the additive release temperature range and the plastic matrix release temperature range, and the fourth dimension includes the type and content of additives and the plastic matrix material.

[0017] Preferred, for:

[0018] ;

[0019] in, This represents the availability coefficient of the d-th dimension of information. This represents the source specificity coefficient of the d-th dimension of information. This represents the detection reliability coefficient of the d-th dimension information. The data quality coefficient represents the information in the d-th dimension.

[0020] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention provides a multimodal fusion traceability analysis method for microplastics. First, it distinguishes whether microplastic particles are from electronic waste sources based on shape feature parameters, enabling preliminary determination of the source of large batches of samples within minutes, effectively improving detection efficiency. In in-depth component analysis, a staged detection strategy is adopted. After determining the release temperature range through online detection, segmented offline collection is then used for precise analysis, realizing the separation and detection of plastic matrix and additives, effectively solving the problems of signal overlap and difficulty in component identification in existing technologies. Furthermore, it integrates four-dimensional information such as shape features, material type, pyrolysis behavior, and additive fingerprints to establish a hierarchical traceability method from rapid screening to precise confirmation, significantly improving the accuracy of microplastic source identification. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application 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 some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 The flowchart illustrates the microplastic multimodal fusion traceability analysis method provided in this embodiment of the invention.

[0023] Figure 2 This is a schematic diagram of the positioning grid provided in an embodiment of the present invention.

[0024] Figure 3 This is a schematic diagram of the synchronous positioning and detection of graphics and materials provided in an embodiment of the present invention.

[0025] Figure 4 This is a schematic diagram illustrating the extraction of shape feature parameters provided in an embodiment of the present invention.

[0026] Figure 5 This is a process diagram illustrating the staged detection of microplastics from electronic waste sources, provided in an embodiment of the present invention. Detailed Implementation

[0027] In view of the shortcomings of the prior art, the inventors of this invention, through long-term research and extensive practice, have proposed the technical solution of this invention. The following will further explain and illustrate this technical solution, its implementation process, and its principles.

[0028] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0029] Furthermore, in the description of this invention, it should be understood that the terms "upper," "lower," "inner," "outer," "horizontal," "vertical," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0030] In the description of this specification, the references to terms such as "an embodiment," "a particular embodiment," or "the embodiment" indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0031] Figure 1 This is a flowchart illustrating the microplastic multimodal fusion traceability analysis method provided in an embodiment of the present invention. (See also...) Figure 1 , combined Figures 2-5 The present invention provides a detailed description of the microplastic multimodal fusion traceability analysis method provided in the embodiments of the present invention, the method including operation S100-operation S500.

[0032] Operate S100 to detect and acquire the material type and shape characteristic parameters of the microplastic particles.

[0033] Operate S200 to determine whether microplastic particles are microplastics from electronic waste sources based on shape characteristic parameters.

[0034] Operating S300, for microplastics from electronic waste sources: take a portion of the microplastic particles for thermogravimetric analysis to obtain the peak temperature of the first derivative curve corresponding to the thermogravimetric curve; pyrolysis gas enters the infrared spectrometer and mass spectrometer to measure the release temperature range of additives and plastic matrix in the microplastic particles, respectively.

[0035] Operate the S400, take another portion of microplastic particles for thermogravimetric analysis, switch the pyrolysis products to gas chromatography-mass spectrometry, and detect the type and content of additives based on the pyrolysis products generated under the additive release temperature range, and detect the material of the plastic matrix based on the pyrolysis products generated under the plastic matrix release temperature range.

[0036] By operating the S500, the source and purpose of microplastic particles are determined by combining material type, shape characteristic parameters, peak temperature, additive release temperature range, plastic matrix release temperature range, additive type and content, and plastic matrix material.

[0037] In a preferred embodiment, operation S100 specifically includes the following sub-operations S110-S130.

[0038] In sub-operation S110, filter paper coated with microplastic particles is attached and placed below the positioning grid, and the grid coordinates of each microplastic particle are recorded.

[0039] In sub-operation S120, microscopic images of each microplastic particle are captured, and shape feature parameters of the microplastic particles are extracted from the microscopic images; the spectra of each microplastic particle are acquired using a Raman spectrometer to determine the material type of the microplastic particles.

[0040] In sub-operation S130, the correspondence between microplastic particles, shape feature parameters, and material type is established based on the grid coordinates.

[0041] First, samples were collected from various sources (e.g., electronic waste dismantling plants, wastewater treatment plants). After pretreatment including digestion, density separation, and filtration, microplastics were enriched on glass fiber filter paper. Subsequently, methods such as... Figure 2 The positioning grid shown divides the filter paper into sections for positioning. Through the coordinated work of the micro Raman spectrometer and the microscope, a one-to-one correspondence between the Raman spectroscopy detection and high-definition image acquisition of the same microplastic particle is achieved.

[0042] Specifically, filter paper coated with microplastic particles is placed under a positioning grid, ensuring close contact between the filter paper and the grid. The grid coordinates of each microplastic particle are recorded. Under a microscope, the microplastic particles are located based on the grid coordinates, high-resolution microscopic images are captured, and shape feature parameters are extracted. Keeping the positions of the filter paper and the positioning grid unchanged, the spectra of the same microplastic particle are acquired using a Raman spectrometer based on the same grid coordinates. The material type is determined by comparing it with a standard spectral library. The grid coordinates, high-resolution microscopic images, and material types of each microplastic particle are associated and stored to construct a graph-material association database. The overall process is as follows: Figure 3 As shown. Based on the aforementioned graphic-material association database, shape feature parameters are extracted.

[0043] In a preferred embodiment, the shape characteristic parameters include at least roundness, rectangularity, aspect ratio, convexity, and edge smoothness. Operation S200 specifically includes: inputting the roundness, rectangularity, aspect ratio, convexity, and edge smoothness of the microplastic particles into a preset screening model, and outputting a category label for the microplastic particles, wherein the category label is electronic waste source microplastics or household source microplastics.

[0044] See Figure 4 Circularity C indicates the degree to which a shape approximates a circle. A represents the area, P represents the perimeter; rectangularity R indicates how closely the shape approximates a rectangle. , The area of ​​the minimum bounding rectangle; aspect ratio Indicates the elongation or slenderness of the shape. , , These are the long and short sides of the smallest bounding rectangle, respectively; convexity. Indicates the degree of concavity or convexity of a shape. , This represents the area of ​​the convex hull.

[0045] Edge smoothness is used to quantify the local curvature of the edge profile of microplastic particles, reflecting the smoothness of the edge. The calculation process is as follows: The particle edge profile is parameterized, with a total of n profile points. The curvature of each point is then calculated. For discrete contour points, curvature can be approximated using the three-point concyclic method or the difference method. This embodiment of the invention employs a curvature estimation method based on adjacent point vectors:

[0046] ;

[0047] in, Let i be the change in the tangent direction angle at the contour point i. Let be the arc length between adjacent contour points i and i+1. Preferably, in this embodiment of the invention, edge smoothness is defined. for:

[0048] ;

[0049] ;

[0050] ;

[0051] in, The normalized edge smoothness The normalized relative curvature This is the perimeter. It should be noted that edge smoothness can also be calculated using other existing edge smoothness calculation methods. Edge smoothness The larger the value, the smoother and more even the edges. The smaller the value, the rougher and more curved the edge. Microplastics from electronic waste have smooth edges due to the mechanical cutting process, resulting in a higher edge smoothness value; microplastics from household sources have rough edges due to environmental weathering, microbial erosion, and other factors, resulting in a lower edge smoothness value.

[0052] For any microplastic particle to be tested, construct a shape feature-material fusion feature vector. ,in, , , , , and These represent the normalized circularity, rectangularity, aspect ratio, convexity, edge smoothness, and material source fingerprint parameters, respectively. Normalization can eliminate the influence of differences in the dimensions and numerical ranges of different features on the model's discrimination results.

[0053] Among them, the material originates from fingerprint parameters. This can be constructed based on the category of microplastic material. For example, materials highly relevant to e-waste dismantling activities, such as polycarbonate (PC), acrylonitrile-butadiene-styrene copolymer (ABS), styrene-acrylonitrile copolymer (SAN), and polystyrene (PS), can be categorized as e-waste source-related materials. Materials with high relevance to daily life sources, such as polyethylene (PE), polypropylene (PP), polyethylene terephthalate (PET), and polyvinyl chloride (PVC), are categorized as the set of materials related to daily life sources. The material source fingerprint parameters can be represented as:

[0054] ;

[0055] in, This indicates that the microplastic particles being tested have been identified as material. The indicator variable or matching confidence level. When using deterministic material identification results, if the particle material is... ,but ,otherwise When using spectral matching probability or confidence level results, This represents the matching probability or confidence level for the corresponding material. Therefore, The higher the value, the closer the fingerprint of the particle material is to the characteristics of electronic waste sources; The smaller the value, the closer the fingerprint of the particle material is to the characteristics of a real-life source.

[0056] Based on this, the following screening model is established. :

[0057] ;

[0058] in, To train the shape-material fusion feature vector of the i-th microplastic particle in the database, The shape and material fusion feature vector of the microplastic particles to be tested; For source labeling, when the particles belong to electronic waste microplastics. When the particles belong to household microplastics ; These are the support vector coefficients obtained during model training. For bias terms; This represents the number of support vectors. This is the kernel function. The acquisition method is as follows: Microplastic particles from known source categories in the training database are used as training samples, and their... as well as As input, the constrained optimization problem is solved through the support vector machine training process to obtain the corresponding values ​​for each training sample. The constrained optimization problem to be solved is: [The problem is incomplete and requires further context to be fully understood.] and Under the condition of , the classification margin is maximized while the classification error is minimized; where C is a penalty parameter used to adjust the model's tolerance for misclassified samples. When When the value is greater than 0, the corresponding training sample becomes a support vector and participates in the determination of the source of the microplastic particles to be tested.

[0059] when At that time, the microplastic particles to be tested were initially identified as microplastics from electronic waste; when At that time, the microplastic particles to be tested were initially identified as microplastics from household sources. Therefore, this model not only uses single material information for judgment, but also comprehensively integrates the particle's roundness, rectangularity, aspect ratio, convexity, edge smoothness / roughness, and material origin fingerprint to achieve rapid preliminary screening of microplastics from electronic waste and household sources.

[0060] In a preferred embodiment, to further reflect the contribution of different shape parameters and material fingerprint parameters in source screening, the kernel function is:

[0061] ;

[0062] in, , , , , , They are respectively The fingerprint parameters include roundness, rectangularity, aspect ratio, convexity, edge smoothness, and material origin. , , , , , They are respectively The fingerprint parameters include roundness, rectangularity, aspect ratio, convexity, edge smoothness, and material origin. The kernel width parameter. This is the roundness weighting coefficient. This is the rectangularity weighting coefficient. The aspect ratio weighting coefficient. For convexity weighting coefficients, This is the edge smoothness weighting coefficient. The weighting coefficients for fingerprint parameters representing the material source. The weighting coefficients can be obtained by training with microplastic samples from known sources in the training database.

[0063] The acquisition method is as follows: Microplastic particles from known source categories in the training database are used as training samples. Feature parameters such as roundness, rectangularity, aspect ratio, convexity, edge smoothness, and material type are extracted. The model is then trained based on electronic waste and household waste source labels to obtain the optimal model. .

[0064] , , , , and To characterize the relative contribution of different feature parameters in source screening, the aforementioned weight coefficients can be determined based on the ability of each feature to distinguish between microplastics from electronic waste sources and microplastics from household sources. Specifically, the significance of differences between samples from different sources, the degree of dispersion between classes, the degree of dispersion within classes, the importance of features, or the contribution to model validation are calculated, and the weight coefficients are determined accordingly.

[0065] In a preferred embodiment, the measurement process of the additive release temperature range and the plastic matrix release temperature range in operation S300 specifically includes: pyrolysis gas entering an infrared spectrometer to monitor the change of characteristic absorption peaks with temperature in real time, so as to determine the additive release temperature range; and pyrolysis gas entering a mass spectrometer to monitor the change of characteristic ion intensity with temperature in real time, so as to determine the plastic matrix release temperature range.

[0066] In a preferred embodiment, in operation S400, another portion of the microplastic particles is taken for thermogravimetric analysis, specifically including: setting a corresponding segmented heating program according to the additive release temperature range and the plastic matrix release temperature range; and taking another portion of the microplastic particles for thermogravimetric analysis under the heating control of the segmented heating program.

[0067] Further preferably, the upper limit of the additive release temperature range is lower than the lower limit of the plastic matrix release temperature range, and the segmented heating process includes the following four stages.

[0068] First stage: The thermogravimetric analyzer is heated to a first temperature at a first heating rate. The first temperature is lower than the lower limit of the additive release temperature range and differs from a first set temperature.

[0069] Second stage: The thermogravimetric analyzer is heated from the first temperature to the second temperature at a second heating rate. The second temperature is higher than the upper limit of the additive release temperature range and differs from the second set temperature. The second heating rate is lower than the first heating rate.

[0070] The third stage: The thermogravimetric analyzer is heated from the second temperature to the third temperature at the third heating rate. The third temperature is lower than the lower limit of the release temperature range of the plastic matrix and differs from the third set temperature. The third heating rate is higher than the first heating rate.

[0071] Fourth stage: The thermogravimetric analyzer is heated from the third temperature to the fourth temperature at the fourth heating rate. The fourth temperature is higher than the upper limit of the release temperature range of the plastic matrix and differs from the fourth set temperature. The fourth heating rate is lower than the third heating rate.

[0072] For microplastic particles initially identified as originating from electronic waste in operation S200, further in-depth component analysis is performed using thermogravimetric-infrared-gas chromatography-mass spectrometry (TG-IR-GCMS) to obtain the additive composition and pyrolysis behavior fingerprint, which is achieved through operations S300-S400.

[0073] See Figure 5 This analysis employs a phased detection method. The first phase involves online continuous detection, where microplastic particles are placed in a thermogravimetric analyzer and heated according to a set temperature program. The thermogravimetric curve (TG) and its first derivative curve (DTG) are recorded in real time. Pyrolysis gases are detected in real time using an infrared spectrometer and a mass spectrometer. By monitoring the changes in characteristic absorption peaks and characteristic ion intensities with temperature, the additive release temperature range is determined. Release temperature range of plastic matrix The second stage involves segmented offline detection. Based on the temperature range determined in the first stage, a segmented heating program is set up. The pyrolysis products in the additive release temperature range are switched to gas chromatography-mass spectrometry (GC-MS) to detect the type and content of additives. The pyrolysis products in the plastic matrix release temperature range are switched to GC-MS to detect the characteristics of plastic pyrolysis products to verify the material type.

[0074] In a preferred embodiment, the fusion determination of the source and use of microplastic particles in operation S500 specifically includes: inputting the material type, shape characteristic parameters, peak temperature, additive release temperature range, plastic matrix release temperature range, additive type and content, and plastic matrix material into a preset fusion discrimination model. The fusion discrimination model adopts a weighted voting mechanism to output the source and use of microplastic particles.

[0075] In a preferred embodiment, the fusion discriminant model is:

[0076] ;

[0077] in, To integrate the results output by the discriminant model, The weights of the d-th dimension information are... For the d-th dimension standard template information from the j-th source, For similarity functions; The d-th dimension represents information. The first dimension includes shape characteristic parameters, the second dimension includes material type, the third dimension includes peak temperature, additive release temperature range, and plastic matrix release temperature range, and the fourth dimension includes additive type and content, as well as plastic matrix material. j represents the candidate source category, including electronic waste sources and residential sources.

[0078] To adapt the weight settings to the microplastic source identification scenario, this invention determines the weights based on the source specificity, detection reliability, and data integrity of different information dimensions. The design principles for the weights of each information dimension are as follows.

[0079] (1) Shape features (first dimension information) can reflect the particle formation process and environmental weathering process. Microplastics from electronic waste sources may have regular shapes, high rectangularity, or high edge smoothness formed by mechanical cutting, crushing, or dismantling processes; microplastics from household sources may exhibit stronger irregularity and edge roughness due to environmental weathering, wear, or bio-erosion. However, since shape features are easily affected by particle size, aging degree, shooting angle, and image segmentation accuracy, the shape feature dimension can be assigned a relatively low to medium weight, mainly used for rapid screening and auxiliary discrimination.

[0080] (2) Material type (second dimension information) is an important basic information for distinguishing the source. Engineering plastics or styrene plastics such as PC, ABS, SAN, and PS are highly correlated with electronic waste dismantling activities, while materials such as PE, PP, PET, and PVC are more common in microplastics from household sources. Therefore, the material type dimension can be given a high weight, but its weight is lower than or close to that of the additive fingerprint dimension, so as to avoid making absolute source judgments based solely on the material.

[0081] (3) Pyrolysis behavior (third dimension information) is used to assist in verifying the material type and supplement the additive release characteristics. The DTG peak temperature, additive release temperature range, and plastic matrix release temperature range can reflect the thermal stability of microplastics, additive release behavior, and matrix pyrolysis characteristics, and can serve as supplementary evidence for material identification and additive fingerprinting. Therefore, the pyrolysis behavior dimension can be given a moderate weight.

[0082] (4) Additive fingerprints (the fourth dimension of information) have high source specificity. Microplastics from electronic waste sources may contain flame retardants, plasticizers, or other characteristic additives related to electronic and electrical products, flame-retardant plastics, or engineering plastics. Therefore, the types and contents of additives have a strong indicative effect on microplastics from electronic waste sources. Thus, the additive fingerprint dimension can be assigned a high weight.

[0083] For example, the weights of the four information dimensions can be set sequentially as follows: , , , The weights are summed to 1. For example, the four weights are 0.15, 0.30, 0.20, and 0.35 respectively. The weights can be adjusted according to the actual distribution and detection results of microplastics from different sources in the training database.

[0084] In a preferred embodiment, for:

[0085] ;

[0086] in, This represents the availability coefficient of the d-th dimension of information. This represents the source specificity coefficient of the d-th dimension of information. This represents the detection reliability coefficient of the d-th dimension information. This represents the data quality coefficient for the d-th dimension of information. When the d-th dimension of information is available... When the data for this dimension is missing or unavailable . , and It can be obtained based on the training database, the results of the detection method verification, and the current sample detection results.

[0087] This can be calculated from a database of microplastic samples from known sources. Specifically, it compares the differences between microplastics from electronic waste and microplastics from household sources in the d-th dimension. The more significant the difference, the stronger the indicative role of that dimension in source identification. The higher the value, the better. For example, if the additive fingerprint or material type differs significantly between electronic waste and residential waste samples, then the corresponding dimension... A higher value can be taken.

[0088] This can be obtained through detection method validation or quality control experiments. Specifically, it can be determined based on standard sample test results, repeatability test results, spiked recovery results, or instrument matching accuracy. The more stable, repeatable, and accurate the test results, the higher the accuracy of this dimension. The higher the value, the better. For example, when the accuracy of Raman spectroscopy material identification is high, the value of the material type dimension is also high. Higher values ​​can be taken; when the recovery and repeatability of additive detection by gas chromatography-mass spectrometry (GC-MS) are good, the additive fingerprint dimension... A higher value can be taken.

[0089] This can be obtained based on the actual detection data of the microplastic particles being tested. Specifically, it can be determined based on image clarity, particle boundary integrity, Raman spectral signal-to-noise ratio, spectral matching confidence level, DTG peak clarity, GC-MS characteristic peak signal-to-noise ratio, and peak matching. The clearer, more complete, and reliable the data for this dimension of the current sample, the better. The higher the value, the better. If there are issues such as blurry images, weak spectral signals, overlapping peaks, or low matching confidence, then... The value decreases accordingly.

[0090] In practical applications, Typically, it is pre-calculated from a known source sample database. This information is typically obtained through experiments validating the detection method. The actual detection quality of each sample is determined in real time. This allows information dimensions with strong source specificity, high detection reliability, and good current data quality to receive higher weight in the fusion and discrimination process.

[0091] When data for a certain dimension is missing or the detection quality is below a set threshold, the data availability coefficient can be used. or data quality coefficient The weights are dynamically adjusted. For example, when additive fingerprint data is missing, let... The weights of the remaining dimensions are renormalized according to the above weight calculation formula; when the signal-to-noise ratio or matching confidence of a certain dimension is low, its weight is reduced. This reduces the impact of this dimension on the final discrimination result. Therefore, the fusion discrimination method of this invention maintains good adaptability under different detection conditions and different data completeness levels.

[0092] The similarity functions for each dimension can be set according to the data type. For continuous data such as shape features and pyrolysis behavior, a distance-decay similarity function can be used.

[0093] ;

[0094] in, For the first The scaling parameter for the d-th dimension can be determined based on the numerical range or standard deviation of that dimension in the training database, or it can be obtained through optimization using the training and validation sets. Specifically, when the numerical fluctuation range of the d-th dimension is large, a smaller scaling parameter can be set. To reduce the excessive influence of this dimension on the similarity results; when the numerical fluctuation range of the d-th dimension is small, a larger value can be set. This enhances the model's ability to identify differences in information in this dimension.

[0095] For categorical data such as material type, a category matching similarity function can be used. When the material of the particle to be tested belongs to the high-frequency material in the source template, the similarity value is higher, and when it does not belong to the high-frequency material in the template, the similarity value is lower.

[0096] For multi-component data such as additive fingerprints, the cosine similarity function can be used:

[0097] ;

[0098] Finally, the similarity of the microplastic particles under test across four dimensions—shape features, material type, pyrolysis behavior, and additive fingerprint—was weighted and summed to obtain their comprehensive matching score against templates from different sources. :

[0099] ;

[0100] When the comprehensive matching score corresponding to the electronic waste source template is greater than that corresponding to the residential source template, the electronic waste source identification result is output; when the comprehensive matching score corresponding to the residential source template is higher, the residential source identification result is output. This multi-dimensional information fusion identification method avoids misjudgments caused by relying solely on shape, material, or additive information, thus improving the accuracy and robustness of microplastic source identification.

[0101] The microplastic multimodal fusion traceability analysis method provided in this invention has the following advantages compared with the prior art.

[0102] (1) By using a positioning grid, the synchronous positioning and detection of graphics and materials are realized. High-definition images and Raman spectra of the same microplastic particles can be collected one by one, and a morphology-material association database is constructed, thereby solving the problem of image and spectral data being separated and difficult to correlate and analyze in the existing technology.

[0103] (2) Based on the constructed morphology-material association database, the system analyzed the differences in shape characteristics between microplastics from electronic waste and microplastics from daily life. It was clarified that microplastics from electronic waste have regular polygonal shapes and smooth edges, while microplastics from daily life have irregular shapes and curved edges. This provides a data basis and discrimination criteria for rapid screening based on shape characteristics.

[0104] (3) Based on this, a screening model based on shape features and material type was established, which can complete the preliminary judgment of the source of a large number of samples within minutes, effectively improving the detection efficiency.

[0105] (4) Further, the four-dimensional information of shape features, material type, pyrolysis behavior and additive fingerprints were integrated to establish a hierarchical traceability method from rapid screening to accurate confirmation, which significantly improved the accuracy of microplastic source identification.

[0106] (5) In the in-depth component analysis, a staged detection strategy is adopted. After the release temperature range is determined by online detection, the data is collected offline in segments for accurate analysis. This realizes the separation and detection of plastic matrix and additives, and effectively solves the problems of signal overlap and component identification in the existing technology.

[0107] The technical solution of the present invention will be further described in detail below with reference to several preferred embodiments and 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. Test methods in the following embodiments that do not specify specific conditions are generally performed under conventional conditions.

[0108] Wastewater samples from an electronic waste dismantling plant and sewage samples from a sewage treatment plant were used as test objects to verify the effectiveness of the microplastic multimodal fusion traceability analysis method provided by this invention. The specific verification process is as follows.

[0109] Samples were collected using in-situ separation and filtration sampling equipment; 2m³ samples were taken from wastewater from the electronic waste dismantling plant. 3 Wastewater samples from the wastewater treatment plant were taken at a depth of 3m. 3 After on-site filtration, microplastics were enriched on glass fiber filter paper with a diameter of 47 mm. At the same time, plastic products (such as mobile phone casings and wire sheaths) and household plastic products (such as food packaging bags and beverage bottles) generated during the dismantling of electronic waste were collected as control samples.

[0110] The filter paper enriched with microplastics was dried in a 60℃ oven for 24 hours, and then digested with 30% hydrogen peroxide solution in a 60℃ water bath for 24 hours to remove organic impurities. A saturated sodium chloride or zinc chloride solution (density 1.2 g / cm³) was then added. 3 The microplastics were allowed to float to the surface after standing for 24 hours. The supernatant was then filtered through glass fiber filter paper with a pore size of 0.45 μm, where the microplastics accumulated on the surface of the filter paper. The solution was then dried at 40°C for later use. The positioning grid was 47 mm × 47 mm in size, consisting of 1 mm × 1 mm square grids made of crisscrossing metal wires. Each grid had a unique coordinate code. The grid was attached tightly to the surface of the filter paper and used as a common positioning reference for both the microscope and the Raman spectrometer.

[0111] Place a gridded filter paper on the microscope stage and scan the screen grid by grid. When a microplastic particle is detected, record its grid coordinates and take a high-resolution microscopic image at the same location using the microscope's built-in camera. Extract shape feature parameters for each microplastic particle.

[0112] Keeping the filter paper and grid in place, transfer the sample to the Raman spectrometer stage. Using the recorded grid coordinates, precisely locate the same microplastic particle using the spectrometer software. Set the Raman spectrometer parameters: laser wavelength 785 nm, power 50 mW, scanning range 505–1620 cm⁻¹. -1 In static scanning mode, with an integration time of 10 seconds and a cumulative total of 3 scans, the Raman spectrum of the particle was acquired. This spectrum was processed using Raman spectroscopy software and compared with a built-in polymer spectral library (containing standard spectra for ABS, PC, PS, PE, PP, PET, PVC, etc.). Particles with a matching degree higher than 80% were labeled as microplastics, and their material type was recorded. The grid coordinates, high-resolution microscopic image, shape characteristic parameters, Raman spectrum, and material type of each microplastic particle were associated and stored to form a graphic-material association database.

[0113] Based on the established graph-material association database, statistical analysis was conducted on the shape characteristics of microplastics from electronic waste sources (n=352) and household sources (n=425). The results showed that microplastics from electronic waste sources exhibited high roundness (0.72±0.12), high rectangularity (0.85±0.10), an aspect ratio close to 1 (1.18±0.25), high convexity (0.93±0.06), and low edge smoothness (0.09±0.04), presenting a regular geometric shape. Microplastics from household sources, on the other hand, exhibited low roundness (0.45±0.18), low rectangularity (0.52±0.22), a large aspect ratio (2.35±1.20), low convexity (0.78±0.15), and high edge smoothness (0.28±0.12), presenting an irregular and complex shape. The differences between the two groups were statistically significant (p<0.01) according to the t-test.

[0114] A support vector machine (SVM) classification model was constructed using shape features (circularity, rectangularity, aspect ratio, convexity, and edge smoothness) and material type as input features. A radial basis function (RBF) kernel was employed, and the penalty and kernel parameters were optimized through grid search. The training and test sets were split in a 7:3 ratio, and five-fold cross-validation was used to evaluate the model's performance. The model achieved a classification accuracy of 87.3%, indicating that the shape-material correlation-based screening model has high reliability and can complete the preliminary determination of the origin of a large number of samples within minutes.

[0115] Samples identified as electronic waste sources by the sieving model were subjected to in-depth TG-IR-GCMS analysis. 1–5 mg of microplastic particles were placed in the sample pan of the thermogravimetric analyzer. The temperature program was set as follows: 40 °C increased to 600 °C at a rate of 10 °C / min under a nitrogen atmosphere at a flow rate of 50 mL / min. TG and DTG curves were recorded in real time. Pyrolysis gases were transferred to the infrared spectrometer and mass spectrometer via a 250 °C transfer line. The infrared spectrometer scanned the wavenumber range from 4000 to 400 cm⁻¹. -1 4cm resolution -1 The mass spectrometer scanning range was 50–650 m / z, and the electron impact source was 70 eV. By monitoring the peak position of the DTG curve and the intensity of characteristic ions in the mass spectrometry, the release temperature range of the additive was determined to be 220–280 °C, and the release temperature range of the plastic matrix was determined to be 380–450 °C.

[0116] Based on the above temperature range, a segmented temperature ramp program was set: the temperature was ramped at 10℃ / min to 200℃ and held for 2 min; the temperature was ramped at 5℃ / min to 300℃, and the pyrolysis products were introduced into the gas chromatography-mass spectrometry (GC-MS) instrument within this range via a six-way valve; the temperature was ramped at 20℃ / min to 360℃ and held for 2 min; the temperature was ramped at 10℃ / min to 470℃, and the pyrolysis products were introduced into the GC-MS instrument again within this range. GC-MS conditions: DB-5MS capillary column (30m × 0.25mm × 0.25μm), helium carrier gas flow rate 1.0mL / min, injection port temperature 250℃, split ratio 10:1, chromatographic temperature program: 50℃ held for 2 min, ramped at 10℃ / min to 300℃ and held for 5 min, mass spectrometry was set to full scan mode (m / z 50~650). Test results: Tris(2-chloropropyl) phosphate (TCPP) was detected in the 220~280℃ range, confirming that the sample contains flame retardants; styrene, acrylonitrile and other characteristic pyrolysis products were detected in the 380~450℃ range, verifying that the material is ABS.

[0117] Shape features (circularity 0.78, rectangularity 0.91, aspect ratio 1.08, convexity 0.96, edge smoothness 0.07), material type (ABS), pyrolysis behavior (additive release temperature range 220-280℃, plastic matrix release temperature range 380-450℃), and additive fingerprint (including TCPP) are input into the fusion discrimination model. A weighted voting mechanism is used to calculate the output result as the source of electronic waste, specifically for electronic product casings, with a confidence level of 94%.

[0118] To verify the technical effectiveness of this invention, three control experiments were set up for comparison with the complete technical solution of this invention. Control group 1 used only Raman spectroscopy to identify materials, control group 2 used only thermogravimetric-gas chromatography-mass spectrometry to detect additives, control group 3 used the present invention but without shape features (i.e., only material + additives + pyrolysis behavior), and the experimental group used the complete technical solution of this invention. 100 microplastic samples from known sources (50 from electronic waste and 50 from residential sources) were tested. The results showed that the accuracy rate of control group 1 was 68% (62% for electronic waste and 74% for residential sources), control group 2 was 72% (78% for electronic waste and 66% for residential sources), control group 3 was 86% (88% for electronic waste and 84% for residential sources), and the accuracy rate of the experimental group was 93% (96% for electronic waste and 90% for residential sources). The source tracing accuracy of the experimental group was significantly better than that of single detection methods and methods without shape feature integration, verifying the effectiveness of the multimodal fusion strategy of this invention.

[0119] It should be understood that the above embodiments are merely illustrative of the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All equivalent changes or modifications made in accordance with the spirit and essence of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for multimodal fusion traceability analysis of microplastics, characterized in that, The methods include: The material type and shape characteristic parameters of microplastic particles are detected and obtained; Based on the shape characteristic parameters, determine whether the microplastic particles are microplastics from electronic waste sources; For microplastics from electronic waste sources: a portion of the microplastic particles were subjected to thermogravimetric analysis to obtain the peak temperature of the first derivative curve corresponding to the thermogravimetric curve; pyrolysis gas was introduced into an infrared spectrometer and a mass spectrometer to measure the release temperature range of additives and plastic matrix in the microplastic particles, respectively. Another portion of microplastic particles was subjected to thermogravimetric analysis. The pyrolysis products were switched to a gas chromatography-mass spectrometry instrument. The type and content of additives were detected based on the pyrolysis products generated under the additive release temperature range, and the material of the plastic matrix was detected based on the pyrolysis products generated under the plastic matrix release temperature range. By combining the material type, shape characteristic parameters, peak temperature, additive release temperature range, plastic matrix release temperature range, additive type and content, and plastic matrix material, the source and purpose of the microplastic particles are determined.

2. The microplastic multimodal fusion traceability analysis method according to claim 1, characterized in that, The material type and shape characteristic parameters of microplastic particles are detected and obtained, specifically including: Filter paper coated with microplastic particles is attached and placed below the positioning grid, and the grid coordinates of each microplastic particle are recorded. Microscopic images of each microplastic particle are captured, and shape feature parameters of the microplastic particles are extracted from the microscopic images; the spectra of each microplastic particle are acquired using a Raman spectrometer to determine the material type of the microplastic particles. Based on the grid coordinates, establish the correspondence between microplastic particles, shape feature parameters, and material types.

3. The microplastic multimodal fusion traceability analysis method according to claim 1, characterized in that, The shape feature parameters include at least roundness, rectangularity, aspect ratio, convexity, and edge smoothness; Determining whether the microplastic particles are electronic waste-sourced microplastics based on the shape characteristic parameters specifically includes: inputting the roundness, rectangularity, aspect ratio, convexity, and edge smoothness of the microplastic particles into a preset screening model, and outputting the category label of the microplastic particles, wherein the category label is electronic waste-sourced microplastics or household-sourced microplastics.

4. The microplastic multimodal fusion traceability analysis method according to claim 3, characterized in that, The kernel function of the screening model is: ； in, For kernel function, To train the shape-material fusion feature vector of the i-th microplastic particle in the database, This is the shape-material fusion feature vector of the microplastic particles to be tested. , , , , , They are respectively The fingerprint parameters include roundness, rectangularity, aspect ratio, convexity, edge smoothness, and material origin. , , , , , They are respectively The fingerprint parameters include roundness, rectangularity, aspect ratio, convexity, edge smoothness, and material origin. The kernel width parameter. This is the roundness weighting coefficient. This is the rectangularity weighting coefficient. The aspect ratio weighting coefficient. For convexity weighting coefficients, This is the edge smoothness weighting coefficient. The fingerprint parameter weighting coefficient is used to determine the material source. Material source fingerprint parameters for: ； in, This indicates that the microplastic particles being tested have been identified as material. The indicator variable or matching confidence level, This represents a collection of materials related to electronic waste sources. This represents the set of materials related to everyday sources; when using deterministic material identification results, if the material of the microplastic particles to be tested is... , ,otherwise, When using spectral matching probability or confidence level results, This represents the matching probability or confidence level for the corresponding material.

5. The microplastic multimodal fusion traceability analysis method according to claim 1, characterized in that, The measurement process for the additive release temperature range and the plastic matrix release temperature range specifically includes: Pyrolysis gas is introduced into an infrared spectrometer, and the change of characteristic absorption peaks with temperature is monitored in real time to determine the temperature range for additive release. The pyrolysis gas enters the mass spectrometer, and the change in the intensity of characteristic ions with temperature is monitored in real time to determine the release temperature range of the plastic matrix.

6. The microplastic multimodal fusion traceability analysis method according to claim 1, characterized in that, Another portion of the microplastic particles was subjected to thermogravimetric analysis, specifically including: Based on the additive release temperature range and the plastic matrix release temperature range, a corresponding segmented heating program is set; Under the heating control of the segmented heating program, another portion of the microplastic particles was taken for thermogravimetric analysis.

7. The microplastic multimodal fusion traceability analysis method according to claim 6, characterized in that, The upper limit of the additive release temperature range is lower than the lower limit of the plastic matrix release temperature range, and the segmented heating program includes: The thermogravimetric analyzer is heated to a first temperature at a first heating rate, the first temperature being lower than the lower limit of the additive release temperature range and differing from a first set temperature. The thermogravimetric analyzer is heated from the first temperature to the second temperature at a second heating rate. The second temperature is higher than the upper limit of the additive release temperature range and differs from a second set temperature. The second heating rate is lower than the first heating rate. The thermogravimetric analyzer is heated from the second temperature to the third temperature at a third heating rate. The third temperature is lower than the lower limit of the release temperature range of the plastic matrix and differs from a third set temperature. The third heating rate is higher than the first heating rate. The thermogravimetric analyzer is heated from the third temperature to the fourth temperature at a fourth heating rate. The fourth temperature is higher than the upper limit of the release temperature range of the plastic matrix and differs from the fourth set temperature. The fourth heating rate is lower than the third heating rate.

8. The microplastic multimodal fusion traceability analysis method according to claim 1, characterized in that, The determination of the source and purpose of the microplastic particles includes: The material type, shape characteristic parameters, peak temperature, additive release temperature range, plastic matrix release temperature range, additive type and content, and plastic matrix material are input into a preset fusion discrimination model. The fusion discrimination model adopts a weighted voting mechanism to output the source and use of the microplastic particles.

9. The microplastic multimodal fusion traceability analysis method according to claim 8, characterized in that, The fusion discrimination model is as follows: ； in, The result output by the fusion discriminant model. The weights of the d-th dimension information are... For the d-th dimension standard template information from the j-th source, It is a similarity function; The information is d-th dimension. The first dimension includes the shape feature parameters, the second dimension includes the material type, the third dimension includes the peak temperature, the additive release temperature range and the plastic matrix release temperature range, and the fourth dimension includes the type and content of additives and the plastic matrix material.

10. The microplastic multimodal fusion traceability analysis method according to claim 9, characterized in that, for: ； in, This represents the availability coefficient of the d-th dimension of information. This represents the source specificity coefficient of the d-th dimension of information. This represents the detection reliability coefficient of the d-th dimension information. The data quality coefficient represents the information of the d-th dimension; when the d-th dimension information is available. When the d-th dimension information is missing or unavailable ; The degree of difference between microplastics from electronic waste and microplastics from household sources in the d-th dimension is determined; the higher the degree of difference, the better. The higher the value; Determined by the results of standard sample testing, repeated testing results, spiked recovery results, or instrument matching accuracy; The peak shape was determined by the sharpness of the microscopic image, the integrity of the particle boundary, the signal-to-noise ratio of the Raman spectrum, the confidence level of the spectral matching, the sharpness of the peak shape of the first derivative curve corresponding to the thermogravimetric curve, the signal-to-noise ratio of the characteristic peaks of the gas chromatography-mass spectrometry, and the peak matching status.

Images