A method and system for rapid detection of road path flow pollutants based on spectral analysis

Abstract

The application provides a road path flow pollutant rapid detection method based on spectrum analysis, and belongs to the field of pollutant detection.The method comprises the following steps: S1, collecting road path flow water samples, performing gradient filtration treatment and scanning, and obtaining Raman spectrum data and infrared spectrum data; S2, performing integral calculation on the Raman spectrum data and the infrared spectrum data in a preset wavelength interval; S3, performing standard normalization processing on Raman comprehensive response intensity and infrared comprehensive response intensity respectively, and performing pollutant type identification based on the similarity between the standard pollutant reference vector, so as to obtain the pollutant type; S4, calculating a parameter representing the state of the pollutant, so as to obtain a state representation parameter; and S5, constructing a decision feature vector according to a two-dimensional normalized response vector, the state representation parameter and an environmental influence parameter, performing a mixed decision based on the decision feature vector, obtaining a risk grade, and determining the detection method according to the risk grade.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of pollutant detection, and in particular to a highway runoff pollutant rapid detection method and system based on spectral analysis. BACKGROUND

[0002] Highway runoff is an important way for new pollutants to enter the environment. Various pollutants such as tire and road wear particles, heavy metals generated by brake system wear, and tire additive oxidation products generated by high-speed vehicle travel are dispersed into the water environment through runoff. Studies have shown that the concentration of tire and road wear particles in highway runoff can reach hundreds of thousands of particles per liter, and the specific toxic substances contained therein have extremely high toxicity to aquatic organisms. Existing laboratory analysis methods such as Fourier transform infrared spectroscopy and thermal cracking gas chromatography mass spectrometry, although accurate, take several hours, and cannot meet the needs of on-site rapid monitoring of highways.

[0003] The rapid characterization method based on Raman spectroscopy and infrared spectroscopy has the advantages of rapidity and non-destructiveness, but the existing methods have the following shortcomings: lack of quantitative characterization parameter system for highway runoff particulate matter, existing spectral methods are mostly qualitative analysis; lack of fast and accurate method for identifying the source of tire wear particles, making it difficult to distinguish tire wear from other carbonaceous particles; lack of fast means for assessing the aging degree of microplastics and rubber particles, as the aging degree directly affects the release rate of additives and ecological toxicity; lack of effective multi-spectral information fusion method, existing research mostly uses different spectral techniques independently without fully exploiting the advantages of Raman and infrared spectroscopy collaborative characterization; lack of layered detection strategy combining spectral rapid screening with laboratory accurate analysis method.

[0004] Chinese patent application CN117538517A discloses a microplastic quantity concentration detection method, which collects microplastics enriched on the surface of a steel membrane and a filtrate by filtering the sample to be tested through a one-micron steel membrane, places the microplastics enriched on the surface of the steel membrane under a high-resolution Raman confocal microscope to identify and count microplastic particles with a diameter of more than one micron, and injects the filtrate into a nanoparticle tracking analyzer to detect and count the number of particles with a particle size of less than one micron, thereby achieving hierarchical detection and quantity counting of microplastics of different particle sizes. However, it only targets a single pollutant type of microplastics and fails to solve the problem of rapid identification when multiple pollutants coexist. SUMMARY

[0005] Therefore, the present application provides a highway runoff pollutant rapid detection method and system based on spectral analysis, which solves the problems of rapid identification of tire wear particles and quantitative evaluation of the aging degree of microplastics, realizes rapid characterization and risk classification of highway runoff particulate matter, significantly improves the detection efficiency while ensuring the accuracy of the detection.

[0006] The technical scheme of the present application is implemented as follows:

[0007] In one aspect, the present application provides a rapid detection method for road path flow pollutants based on spectral analysis, comprising the following steps:

[0008] S1, collecting road path flow water samples and performing gradient filtration processing to obtain filtrate and filter membrane intercepted particulate matter, performing Raman spectrum scanning on the suspension of the filter membrane intercepted particulate matter to obtain Raman spectrum data, and performing Fourier transform infrared spectrum scanning on the filter membrane intercepted particulate matter after drying treatment to obtain infrared spectrum data;

[0009] S2, integrating and calculating the Raman spectrum data and the infrared spectrum data in a preset wave number interval, respectively, to obtain Raman comprehensive response intensity and infrared comprehensive response intensity;

[0010] S3, performing standard normalization processing on the Raman comprehensive response intensity and the infrared comprehensive response intensity, respectively, to obtain a two-dimensional normalized response vector, and performing pollutant source identification based on the similarity between the two-dimensional normalized response vector and a standard pollutant reference vector to obtain a pollutant source type;

[0011] S4, calculating a state characteristic parameter representing the state of the pollutant based on the characteristic peak area in the Raman spectrum data and the infrared spectrum data to obtain the state characteristic parameter;

[0012] S5, constructing a decision feature vector according to the two-dimensional normalized response vector, the state characteristic parameter, and an environmental impact parameter, performing mixed decision based on the decision feature vector to obtain a risk level, and determining the detection method according to the risk level.

[0013] On the basis of the above technical scheme, preferably, the calculation formula of the Raman spectrum comprehensive response intensity and the infrared spectrum comprehensive response intensity is:

[0014] ;

[0015] ;

[0016] wherein, Raman spectrum comprehensive response intensity, lower limit of Raman characteristic wave number interval, upper limit of Raman characteristic wave number interval, Raman spectrum intensity after background correction, Raman shift wave number, baseline intensity, infrared spectrum comprehensive response intensity, wavelength, upper limit of infrared characteristic wave number interval, lower limit of infrared characteristic wave number interval, represents the absorbance of the infrared spectrum at the wave number .

[0017] On the basis of the above technical solutions, preferably, the step S3 specifically comprises:

[0018] the Raman spectrum comprehensive response intensity and the infrared spectrum comprehensive response intensity are respectively normalized to obtain the normalized Raman response intensity and the normalized infrared response intensity , and constitute a two-dimensional normalized response vector ; wherein , represents transposition;

[0019] the similarity between the two-dimensional normalized response vector and each standard pollutant reference vector is calculated .

[0020] the standard pollutant type with the maximum similarity is selected as the pollutant source type of the sample.

[0021] On the basis of the above technical solutions, preferably, when the similarity is less than the similarity threshold value, then multiple pollutants coexist in the path flow water sample, and a non-negative least square method is used to fit the two-dimensional normalized response vector of the sample as a linear combination of multiple standard pollutant reference vectors, to solve the optimization problem , with the constraint condition being ≥0 and , and the linear combination coefficient obtained by solving is the relative contribution rate of the jth pollutant.

[0022] On the basis of the above technical solutions, preferably, the step S4 specifically comprises:

[0023] rubber characteristic peaks located in a preset wave number interval are identified from the Raman spectrum data, and rubber characteristic peak areas are extracted , D peaks and G peaks of carbon black are identified and peak intensities are extracted, a carbon quality characteristic index is calculated based on the rubber characteristic peak areas , the carbon quality characteristic index characterizes the abundance of carbon filler in the particulate matter relative to the organic polymer matrix:

[0024] ;

[0025] ;

[0026] wherein ​and the peak area of the D peak and the peak area of the G peak, is a regularization term, represents the area ratio of the D peak to the G peak, represents a structure correction coefficient;

[0027] The peak area of the D peak and the peak area of the G peak are extracted from the Raman spectrum data, and the graphitization degree index The graphitization degree index is used to assist in judging the graphitization degree of carbonaceous materials;

[0028] ;

[0029] The comprehensive aging index The comprehensive aging index characterizes the degree of oxidative aging of the polymer material:

[0030] ;

[0031] ;

[0032] wherein, is the integral area of the carbonyl characteristic peak, is the reference peak area, represents the saturated carbonyl index, represents the integral area of the hydroxyl characteristic peak, represents the saturated hydroxyl index

[0033] The comprehensive aging index is calculated by weighted combination of the carbonyl oxidation index and the double bond consumption index:

[0034] ;

[0035] ;

[0036] ;

[0037] ;

[0038] wherein, and are material adaptive coefficients, represents a material type parameter, which is identified by an infrared spectrum characteristic peak, =0 represents a high unsaturation rubber, =1 represents a medium unsaturation rubber, =2 represents a low unsaturation rubber; for saturated polymers without carbon-carbon double bonds, .

[0039] On the basis of the above technical solutions, preferably, step S5 specifically comprises:

[0040] determining the source determination feature according to the recognition confidence of the pollutant source type , the source determination feature is equal to the maximum similarity ;

[0041] determining the aging degree feature according to the comprehensive aging index in the state characterization parameter , the aging degree feature is equal to the normalized value of the comprehensive aging index ;

[0042] determining the environmental risk feature according to the environmental influence parameters such as traffic flow, rainfall intensity, and distance from the road of the sampling site , the environmental risk feature is calculated by weighted combination of multiple environmental parameters;

[0043] combining the source determination feature , the aging degree feature and the environmental risk feature into a decision feature vector ;

[0044] obtaining the risk level based on the decision feature vector through mixed decision, and the mixed decision method combines a mandatory decision rule and a risk assessment decision tree model, wherein the mandatory decision rule is that when the carbon content feature index exceeds a preset high carbon content threshold and the comprehensive aging index exceeds a preset high aging threshold, it is forcibly determined as a high-risk sample, and when the road path runoff sample does not satisfy the mandatory decision rule, the decision feature vector is input into the risk assessment decision tree model for classification to obtain the risk level.

[0045] On the basis of the above technical solutions, preferably, the construction of the risk assessment decision tree specifically comprises:

[0046] defining the risk level as a target variable , calculating the initial entropy of the training set :

[0047] ;

[0048] wherein, is the entropy of the risk level, is the sample proportion of the risk level being , is the risk level category, indicates low risk, ​​indicates a medium risk, indicates a high risk;

[0049] calculating information gain for each decision feature in the decision feature vector

[0050]

[0051] wherein, is the decision feature number, is the information gain of the th decision feature, is the conditional entropy under the given decision feature condition.

[0052] The decision feature with the maximum information gain is selected as the split feature of the current node, the sample subsets are divided based on the value range of the decision feature, and the sub-tree is recursively constructed until the stopping condition is met to form the leaf node and output the corresponding risk level.

[0053] On the basis of the above technical scheme, preferably, when the risk level is low risk, a portable Raman spectrometer is used to perform on-site rapid scanning on the path flow water sample to obtain the main pollutant type and concentration range, and a low-risk sample database is established.

[0054] When the risk level is medium risk, the path flow water sample is sent to the laboratory for qualitative and quantitative analysis of organic pollutants, heavy metal element analysis, and microplastic type identification to obtain the accurate concentration and chemical composition of the pollutants.

[0055] When the risk level is high risk, the path flow water sample is sent to a third-party detection institution with qualification, a high-resolution mass spectrometer is used to perform unknown pollutant screening and targeted quantitative analysis to obtain the trace concentration, chemical structure, morphology characteristics, and source analysis of the pollutants, and an emergency response mechanism is started.

[0056] Further preferably, step S1 specifically comprises:

[0057] After collecting the path flow water sample, a first filter membrane with a pore size of 10 microns is used for primary filtration to intercept large particle suspensions to obtain a first filtrate and first filter membrane intercepted particulate matter;

[0058] A second filter membrane with a pore size of 0.45 microns is used for secondary filtration on the first filtrate to intercept small particulate matter to obtain a second filtrate and second filter membrane intercepted particulate matter;

[0059] The first filter membrane intercepted particulate matter and the second filter membrane intercepted particulate matter are mixed and added with ultrapure water to prepare a suspension, and Raman spectrum scanning is performed on the suspension to obtain Raman spectrum data;

[0060] ​​The second filter membrane retained particles are dried, and the dried filter membrane is subjected to Fourier transform infrared spectrum scanning to obtain infrared spectrum data.

[0061] In another aspect, the present application provides a rapid detection system for path flow pollutants based on spectral analysis, which adopts the rapid detection method for path flow pollutants based on spectral analysis as described above, and comprises:

[0062] A sample collection module is configured to collect path flow water samples and perform gradient filtration to obtain filtrate and filter membrane retained particles, perform Raman spectrum scanning on the suspension of the filter membrane retained particles to obtain Raman spectrum data, and perform Fourier transform infrared spectrum scanning on the dried filter membrane retained particles to obtain infrared spectrum data.

[0063] A spectral analysis module is configured to perform integral calculation on the Raman spectrum data and the infrared spectrum data in a preset wave number range, respectively, to obtain Raman comprehensive response intensity and infrared comprehensive response intensity.

[0064] A pollutant identification module is configured to perform standard normalization on the Raman comprehensive response intensity and the infrared comprehensive response intensity, respectively, to obtain a two-dimensional normalized response vector, and perform pollutant source identification based on the similarity between the two-dimensional normalized response vector and a standard pollutant reference vector to obtain a pollutant source type.

[0065] A parameter calculation module is configured to calculate a parameter representing the state of the pollutant based on the characteristic peak area in the Raman spectrum data and the infrared spectrum data to obtain a state representation parameter.

[0066] A grade evaluation module is configured to construct a decision feature vector based on the two-dimensional normalized response vector, the state representation parameter and an environmental impact parameter, perform mixed decision based on the decision feature vector to obtain a risk grade, and determine the detection method according to the risk grade.

[0067] The rapid detection method and system for path flow pollutants based on spectral analysis of the present application have the following beneficial effects compared with the prior art:

[0068] (1) The comprehensive response intensity is obtained by integral calculation of the Raman spectrum data and the infrared spectrum data to form a two-dimensional normalized response vector, the pollutant type is identified by vector similarity, and mixed decision is performed by constructing a decision feature vector and combining mandatory rules and a decision tree model, thereby realizing intelligent determination of the risk grade, significantly improving the detection efficiency and reducing the overall detection cost.

[0069] (2) The pollutant state representation method based on the characteristic peak intensity ratio is adopted, which improves the accuracy and stability of the state parameter and makes up for the deficiency of the traditional detection method that only provides qualitative information.

[0070] (3) The decision tree model based on spectral feature information gain is used to determine the risk level, the optimal split feature is selected by calculating the information gain of each decision feature to recursively construct the decision tree, the multi-dimensional feature information is fully utilized for comprehensive judgment, and the accuracy of the determination result is improved;

[0071] (4) The detection method is determined according to the risk level, the optimal configuration of the detection resource is realized, the limited precise instrument analysis resource is mainly used for the medium and high risk samples which need accurate quantification, the low risk samples do not need further detection after rapid screening, the overall detection efficiency is significantly improved while the detection accuracy is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0072] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0073] Figure 1 A flow chart of a road path flow pollutant rapid detection method based on spectral analysis of the present application;

[0074] Figure 2 Three spectral data schematic diagrams of a road path flow pollutant rapid detection method based on spectral analysis of the present application;

[0075] Figure 3 A two-dimensional normalized response vector schematic diagram of a road path flow pollutant rapid detection method based on spectral analysis of the present application;

[0076] Figure 4 A state characterization parameter calculation flow chart of a road path flow pollutant rapid detection method based on spectral analysis of the present application. DETAILED DESCRIPTION

[0077] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0078] As shown in Figure 1 and Figure 2 , the present application provides a road path flow pollutant rapid detection method based on spectral analysis, comprising the following steps:

[0079] S1. Collect water samples from the channel path and perform gradient filtration to obtain filtrate and particulate matter retained by the filter membrane. Perform Raman spectroscopy on the suspension of particulate matter retained by the filter membrane to obtain Raman spectral data. After drying the particulate matter retained by the filter membrane, perform Fourier transform infrared spectroscopy to obtain infrared spectral data.

[0080] Specifically, step S1 includes:

[0081] After collecting water samples from the path, primary filtration was performed using a first filter membrane with a pore size of 10 micrometers to remove large suspended particles, resulting in the first filtrate and the particulate matter removed by the first filter membrane.

[0082] The first filtrate is subjected to secondary filtration using a second filter membrane with a pore size of 0.45 micrometers to trap fine particulate matter, resulting in the second filtrate and the particulate matter trapped by the second filter membrane.

[0083] The particles retained by the first filter membrane and the particles retained by the second filter membrane were mixed and then added to ultrapure water to prepare a suspension. The suspension was subjected to Raman spectroscopy scanning with a wavenumber range of 400 to 3200 cm⁻¹, an integration time of 10 to 30 seconds, and an accumulation number of 2 to 5 times to obtain Raman spectral data.

[0084] After the particulate matter trapped in the second filter membrane is dried, the dried filter membrane is subjected to Fourier transform infrared spectroscopy scanning in the wavenumber range of 400 to 4000 cm⁻¹. -1 The resolution is 4 cm. -1 The infrared spectral data were obtained by accumulating 16 to 64 scans.

[0085] Understandably, automatic sampling devices are installed at highway stormwater collection wells or drainage pipe outlets. The sampling process automatically starts when rainfall reaches a preset threshold and runoff forms. The collected runoff samples are transported to a pretreatment module via a peristaltic pump for gradient filtration. Primary filtration removes large particles such as leaves and stones, while secondary filtration effectively traps tire and road wear particles, as well as colloidal heavy metals. Raman spectroscopy identifies the main characteristic peaks of tire wear particles, including butadiene and styrene peaks, as well as the D and G peaks of carbon black; infrared spectroscopy reflects the characteristic absorption of aromatic organic matter and humic substances.

[0086] S2. Integrate the Raman and infrared spectral data within the preset wavenumber intervals to obtain the Raman and infrared combined response intensities. The calculation formula is as follows:

[0087] ;

[0088] ;

[0089] in, Indicates the overall response intensity of the Raman spectrum. This represents the lower limit of the Raman characteristic wavenumber range. This represents the upper limit of the Raman characteristic wavenumber range. This represents the Raman spectral intensity after background correction. Represents the Raman shift wavenumber. Indicates baseline strength. Indicates the overall intensity of the infrared spectrum response. Indicates wavelength. This indicates the upper limit of the infrared characteristic wavenumber range. This represents the lower limit of the infrared characteristic wavenumber range. Indicates the infrared spectrum at wavenumber The absorbance at that location.

[0090] Understandably, the overall Raman spectral response intensity is the total Raman scattering intensity beyond the baseline within the entire characteristic range, and it is proportional to the total concentration of pollutants and the Raman scattering cross section. The overall infrared spectral response intensity is the total absorption capacity within the infrared characteristic range, and it is related to the concentration of aromatic pollutants.

[0091] like Figure 3 As shown in Figure S3, the Raman integrated response intensity and the infrared integrated response intensity are respectively normalized to obtain a two-dimensional normalized response vector, and the pollutant source is identified based on the similarity with the standard pollutant reference vector to obtain the pollutant source type.

[0092] Specifically, step S3 includes:

[0093] Comprehensive response intensity of Raman spectra Combined response intensity with infrared spectrum Normalization was performed separately to obtain the normalized Raman response intensity. and normalized infrared response intensity And constitute a two-dimensional normalized response vector. ;in , Indicates transpose;

[0094] ;

[0095] ;

[0096] in, Indicates the overall response intensity of the Raman spectrum. This represents the minimum intensity of the overall Raman spectrum response. This represents the maximum value of the overall Raman spectrum response intensity. Indicates the overall response intensity of the infrared spectrum. This represents the minimum value of the overall infrared spectral response intensity. This represents the maximum value of the overall infrared spectral response intensity;

[0097] Calculate the two-dimensional normalized response vector Similarity with each standard pollutant reference vector ;

[0098] ;

[0099] ;

[0100] in, This indicates the similarity between the water sample from the channel path and the j-th reference pollutant. This represents the two-dimensional normalized reference vector for the j-th standard pollutant. Represents the diagonal weight matrix. The weighting coefficients representing the overall intensity of the Raman spectrum response. Weighting coefficients representing the overall intensity of the infrared spectral response;

[0101] Select similarity The largest standard pollutant type is used as the pollutant source type for the sample.

[0102] Understandable. and This can be adjusted according to the actual usage scenario. For example, when the main focus is on carbonaceous pollutants, the Raman weight can be increased to... When the primary focus is on the aging of organic polymers, the infrared weighting can be increased to... .

[0103] In one embodiment of the present invention, when the similarity When the similarity is less than the threshold, multiple pollutants coexist in the water sample along the channel path. The two-dimensional normalized response vector of the sample is fitted using the non-negative least squares method, which is a linear combination of multiple standard pollutant reference vectors. The optimization problem is then solved. The constraints are ≥0 and The linear combination coefficients obtained by solving denoted as the relative contribution rate of pollutant j.

[0104] Understandable. This represents the contribution rate of the j-th pollutant, if the similarity... A very high value indicates that the road runoff sample mainly consists of the j-th type of pollutant. The solution process is actually searching for an optimal combination of coefficients in M-dimensional space that minimizes the Euclidean distance between the "composite vector" constructed from these coefficients and the actual measured road runoff sample vector, while satisfying the constraint that all coefficients are non-negative and their sum is 1. If the residual obtained from the non-negative least squares fitting is still large (poor fitting quality), it may mean that there are novel pollutants in the sample that are not included in the database, requiring further in-depth analysis. If the fitting quality is very good, a quantitative report on the mixture composition can be given, such as "the sample mainly consists of 60% tire wear particles, 25% polyethylene microplastics, and 15% asphalt particles".

[0105] like Figure 4 As shown in Figure S4, parameters characterizing the state of pollutants are calculated based on the characteristic peak areas in Raman and infrared spectral data, thus obtaining the state characterization parameters.

[0106] Specifically, step S4 includes:

[0107] Identify rubber characteristic peaks located within a preset wavenumber range from Raman spectroscopy data and extract the area of ​​these peaks. The D and G peaks of carbon black were identified and their intensities were extracted, based on the characteristic peak area of ​​rubber. Calculation of carbonaceous characteristic index Carbonaceous characteristic index Characterizing the abundance of carbonaceous fillers relative to the organic polymer matrix in particulate matter:

[0108] ;

[0109] ;

[0110] in and These are the peak areas of peak D and peak G, respectively. For regularization terms, This represents the area ratio of peak D to peak G. Indicates the structural correction factor;

[0111] The peak areas of the D and G peaks of carbon black were extracted from Raman spectroscopy data, and the graphitization index was calculated. Graphitization index Used to assist in determining the degree of graphitization of carbonaceous materials;

[0112] ;

[0113] Calculate the comprehensive aging index Comprehensive aging index Characterizing the degree of oxidative aging of polymer materials:

[0114] ;

[0115] ;

[0116] in, The integral area of ​​the carbonyl characteristic peak. For reference peak area, Indicates the saturated carbonyl index. This represents the integral area of ​​the hydroxyl characteristic peak. Indicator of saturated hydroxyl index

[0117] Comprehensive Aging Index Calculated by weighted combination of carbonyl oxidation index and double bond consumption index:

[0118] ;

[0119] ;

[0120] ;

[0121] ;

[0122] in, and For material adaptive coefficients, Material type parameters are indicated by characteristic peaks in the infrared spectrum. =0 indicates highly unsaturated rubber. =1 indicates medium-unsaturated rubber. =2 indicates low unsaturation rubber; for saturated polymers that do not contain carbon-carbon double bonds, .

[0123] Understandably, the rubber characteristic peaks located within a preset wavenumber range are identified from Raman spectroscopy data, and the intensity of the rubber characteristic peaks is extracted. The type of rubber is determined by peak position matching. If a strong peak is detected near a specific wavenumber, it is determined to be styrene-butadiene rubber. If a strong peak is detected at other specific wavenumbers, it is determined that it may contain natural rubber or nitrile rubber components.

[0124] The D and G peaks of carbon black are identified and their intensities are extracted. The characteristic peak intensity of carbon black is calculated as the sum of the peak intensities of the D and G peaks. The carbonaceous characteristic index is calculated based on the characteristic peak intensity of rubber. The regularization term is used to avoid unstable ratios due to excessively small denominators. This is achieved through the numerator term... Quantitative characterization of the total amount of carbonaceous components in a sample, denominator term The ratio of organic polymer matrix content to carbonaceous filler content reflects the relative abundance of the carbonaceous filler. An introduced correction factor is also included. Based on Raman spectroscopy for carbon material characterization, the structural characteristics (degree of graphitization, defect density) reflected by the D / G peak area ratio are used to help distinguish carbonaceous particles from different sources. (Tire carbon black is an example.) Typical values ​​are 0.8-1.2, while diesel soot and asphalt carbonaceous particles... Typically, within varying ranges, this correction can improve the accuracy of identifying tire wear particles. When A value >0.6 indicates a high carbon content, possibly indicating tire wear particles; when 0.3 < ≤0.6 indicates a medium carbon content, which needs to be judged in conjunction with other parameters; when If the concentration is ≤0.3, it indicates a low carbon content, with organic polymers being the main component.

[0125] According to the classical theory of Raman spectroscopy for characterizing carbon materials (Ferrari-Robertson model), the intensity of the D peak reflects the degree of defects and disorder in the carbon lattice, while the intensity of the G peak reflects the graphitized ordered structure. A larger value indicates more defects and a lower degree of graphitization; conversely, a smaller value indicates a more ordered structure and a higher degree of graphitization. This can be achieved through a conversion formula. The graphitization index is positively correlated with the degree of graphitization.

[0126] By setting a saturation threshold and By mapping the aging index of different materials to a standardized range of 0-1, the aging degree of different types of pollutants can be compared. The design is based on polymer aging chemistry theory: the oxidation and consumption of unsaturated bonds, along with the formation of carbonyl and hydroxyl groups, are synergistic oxidation reactions. The relative abundance of carbonyl and hydroxyl groups can approximately reflect the degree of double bond consumption, thus avoiding the difficulty of finding a "fresh reference" from samples of unknown origin. The material adaptive weighting coefficients dynamically adjust the contributions of COI and VCI according to the polymer's unsaturation. For highly unsaturated materials (such as natural rubber), double bond consumption is more significant, and the weights are relatively balanced; for low-unsaturated materials, carbonyl formation is the main aging marker, increasing the COI weight. ≥0.7 indicates high aging (severe photo-oxidative degradation), while 0.4≤ <0.7 indicates moderate aging (significant signs of degradation); when A value of <0.4 indicates mild aging or fresh particulate matter.

[0127] This invention employs a pollutant state characterization method based on characteristic peak intensity ratio. By calculating the intensity ratio of carbon black characteristic peaks to rubber characteristic peaks, it determines whether the pollutant originates from tire wear particles. By comprehensively considering the generation of carbonyl oxidation products and the retention of carbon-carbon double bonds, it assesses the aging degree of microplastics. This peak intensity ratio-based method improves the accuracy and stability of state parameters and compensates for the shortcomings of traditional detection methods that only provide qualitative information.

[0128] In one embodiment of the present invention, the structure correction factor is 0.3, which is used to distinguish the structural differences of carbonaceous particles from different sources.

[0129] In one embodiment of the present invention, tire carbon black is typically a moderately graphitized furnace black. Typical values ​​are 0.45-0.55; smoke and dust produced by high-temperature combustion The value is relatively high (0.6-0.8); the carbon produced by biomass combustion The values ​​are relatively low (0.3-0.45). Combining the CCI index and GDI can improve the accuracy of pollutant source tracing.

[0130] In one embodiment of the present invention, the carbonyl characteristic peak is 1650-1850 cm⁻¹. -1 The reference peak is typically at 1450 cm⁻¹. -1 The nearby -CH2- bending vibration peak, saturated carbonyl index The value is 3, and the characteristic peak of the hydroxyl group is 3200-3600 cm⁻¹. -1 saturated hydroxyl index It is 1.5.

[0131] In one embodiment of the present invention, when =0 (such as natural rubber). =0.5, =0.5; when =1 (e.g., styrene-butadiene rubber). =0.5, =0.5; =2 (e.g., nitrile rubber). =0.5, =0.5.

[0132] S5. Construct a decision feature vector based on the two-dimensional normalized response vector, state characterization parameters, and environmental impact parameters. Perform hybrid decision-making based on the decision feature vector to obtain the risk level, and determine the detection method based on the risk level.

[0133] This invention calculates the comprehensive response intensity by integrating Raman and infrared spectral data and constructs a two-dimensional normalized response vector. It uses vector similarity to identify pollutant types, solving the problem of limited selectivity of single spectral techniques. Furthermore, by constructing decision feature vectors and combining mandatory rules with decision tree models for hybrid decision-making, it achieves intelligent risk level determination, significantly improving detection efficiency and reducing overall detection costs.

[0134] Specifically, step S5 includes:

[0135] Source identification characteristics are determined based on the confidence level of the pollutant source type. Source determination characteristics Equal to maximum similarity , ;

[0136] Based on the comprehensive aging index in the state characterization parameters Determine the characteristics of aging Characteristics of aging Equal to the comprehensive aging index The normalized value;

[0137] Environmental risk characteristics are determined based on environmental impact parameters such as traffic flow, rainfall intensity, and distance from roads at the sampling sites. Environmental risk characteristics Calculated by weighted combination of multiple environmental parameters;

[0138] Source determination characteristics Characteristics of aging and environmental risk characteristics Combined into decision feature vector ;

[0139] Risk levels are derived through a hybrid decision-making process based on decision feature vectors. This hybrid decision-making method combines mandatory judgment rules with a risk assessment decision tree model. The mandatory judgment rule is based on the carbonaceous characteristic index... Exceeding the preset high carbon content threshold and the comprehensive aging index Samples exceeding a preset high aging threshold are forcibly classified as high-risk. When a flow sample along a path does not meet the mandatory classification rule, the decision feature vector is... The risk assessment decision tree model is used to classify the risk levels.

[0140] In one embodiment of the present invention, the construction of the risk assessment decision tree specifically includes:

[0141] Define risk level as the target variable Calculate the initial entropy of the training set. :

[0142] ;

[0143] in, Entropy for risk level The risk level is The sample proportion Risk level category Indicates low risk. Indicates medium risk. Indicates high risk;

[0144] Calculate the information gain for each decision feature in the decision feature vector. :

[0145] ;

[0146] ;

[0147] in, For decision feature index, For the first Information gain of each decision feature Given decision features Conditional entropy under certain conditions Representation of features The set of possible values, Representation of features Values The sample proportion Indicates in Entropy of risk level under certain conditions;

[0148] The decision feature with the largest information gain is selected as the splitting feature of the current node. Based on the value range of the decision feature, the sample subset is divided and the subtree is recursively constructed until the stopping condition is met to form a leaf node and output the corresponding risk level.

[0149] Understandably, the training set data sources include historical channel flow sample spectral data and their corresponding precise laboratory analysis results and risk ratings, water pollution monitoring data from relevant fields, and simulation training samples based on expert knowledge. A higher initial entropy value indicates greater uncertainty in the risk level.

[0150] Stopping conditions include all samples in the subset having the same risk level, reaching a preset maximum tree depth to avoid overfitting, the number of samples in the subset being less than a preset minimum number of split samples, or the information gain of all features being less than a preset threshold. The risk level label of a leaf node is the risk level of the majority of samples in that node.

[0151] This invention employs a decision tree model based on spectral feature information gain for risk level determination. By calculating the information gain of each decision feature, the optimal splitting feature is selected and the decision tree is recursively constructed. This fully utilizes multi-dimensional feature information for comprehensive judgment, avoiding the one-sidedness of single-indicator judgment, ensuring reliable identification of high-risk samples, and improving the accuracy of the judgment results.

[0152] Furthermore, the determination of the detection method based on the risk level specifically includes:

[0153] When the risk level is low, a portable Raman spectrometer is used to quickly scan the water samples along the channel path to obtain the main pollutant types and concentration ranges, and to establish a low-risk sample database.

[0154] When the risk level is medium risk, the water samples from the road path will be sent to the laboratory for qualitative and quantitative analysis of organic pollutants, analysis of heavy metal elements, and identification of microplastic types to obtain the precise concentration and chemical composition of pollutants.

[0155] When the risk level is high, the water samples from the route will be sent to a qualified third-party testing institution for screening and targeted quantitative analysis of unknown pollutants using a high-resolution mass spectrometer to obtain trace concentrations, chemical structures, morphological characteristics and source analysis of the pollutants, and an emergency response mechanism will be activated.

[0156] This invention determines the detection method based on the risk level. Low-risk samples are rapidly scanned on-site using a portable spectrometer, medium-risk samples are sent to a laboratory for standard instrument analysis, and high-risk samples are sent to a third-party testing institution for comprehensive characterization and activation of an emergency response mechanism. This tiered detection strategy optimizes the allocation of testing resources, focusing limited precision instrument analysis resources on medium- and high-risk samples that require accurate quantification. Low-risk samples do not require further testing after rapid screening, thus significantly improving overall detection efficiency while ensuring detection accuracy.

[0157] In one embodiment of the present invention, the trend of pollutant concentration change is analyzed based on historical monitoring data. When the spectral response amplitude of the current sample increases significantly compared with the historical average level (relative change rate Δ>0.5), it indicates that there may be an aggravation of pollution, and the risk level is upgraded by one level accordingly.

[0158] This invention also provides a rapid detection system for road runoff pollutants based on spectral analysis, employing the rapid detection method for road runoff pollutants based on spectral analysis as described above, including:

[0159] The sample acquisition module is used to collect water samples from the channel path and perform gradient filtration to obtain filtrate and particulate matter retained by the filter membrane. The Raman spectrum data is obtained by Raman spectroscopy scanning of the suspension of particulate matter retained by the filter membrane, and the particulate matter retained by the filter membrane is dried and then subjected to Fourier transform infrared spectroscopy scanning to obtain infrared spectral data.

[0160] The spectral analysis module is used to perform integral calculations on Raman spectral data and infrared spectral data within preset wavenumber intervals to obtain the Raman comprehensive response intensity and the infrared comprehensive response intensity.

[0161] The pollutant identification module is used to perform standard normalization processing on the Raman integrated response intensity and the infrared integrated response intensity to obtain a two-dimensional normalized response vector, and to identify the pollutant source based on the similarity between the vector and the standard pollutant reference vector to obtain the pollutant source type.

[0162] The parameter calculation module is used to calculate parameters characterizing the state of pollutants based on the characteristic peak areas in Raman and infrared spectral data, and obtain state characterization parameters.

[0163] The risk assessment module is used to construct a decision feature vector based on the two-dimensional normalized response vector, state characterization parameters, and environmental impact parameters. Based on the decision feature vector, a hybrid decision is made to obtain the risk level, and the detection method is determined according to the risk level.

[0164] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A rapid detection method for road runoff pollutants based on spectral analysis, characterized in that... Includes the following steps: S1. Collect water samples from the channel path and perform gradient filtration to obtain filtrate and particulate matter retained by the filter membrane. Perform Raman spectroscopy on the suspension of particulate matter retained by the filter membrane to obtain Raman spectral data. After drying the particulate matter retained by the filter membrane, perform Fourier transform infrared spectroscopy to obtain infrared spectral data. S2. Integrate the Raman spectral data and infrared spectral data within the preset wavenumber range to obtain the Raman comprehensive response intensity and the infrared comprehensive response intensity; S3. The Raman and infrared integrated response intensities are standardized and normalized to obtain a two-dimensional normalized response vector. Based on the similarity to the standard pollutant reference vector, the pollutant source is identified to determine the pollutant source type. Step S3 specifically includes: Comprehensive response intensity of Raman spectra Combined response intensity with infrared spectrum Normalization was performed separately to obtain the normalized Raman response intensity. and normalized infrared response intensity And form a two-dimensional normalized response vector. ;in , Indicates transpose; Calculate the two-dimensional normalized response vector Similarity with each standard pollutant reference vector ; Select similarity The largest standard pollutant type is used as the pollutant source type of the sample. S4. Calculate the parameters characterizing the pollutant state based on the characteristic peak areas in the Raman and infrared spectral data to obtain the state characterization parameters; Step S4 specifically includes: Identify rubber characteristic peaks located within a preset wavenumber range from Raman spectroscopy data and extract the area of ​​these peaks. Identify the D and G peaks of carbon black and extract their peak intensities, based on the characteristic peak area of ​​rubber. Calculation of carbonaceous characteristic index Carbonaceous characteristic index Characterizing the abundance of carbonaceous fillers relative to the organic polymer matrix in particulate matter: ; ; in and The peak areas of peak D and peak G are respectively. For regularization terms, This represents the area ratio of peak D to peak G. Indicates the structural correction factor; The peak areas of the D and G peaks of carbon black were extracted from Raman spectroscopy data, and the graphitization index was calculated. Graphitization index Used to assist in determining the degree of graphitization of carbonaceous materials; ; Calculate the comprehensive aging index Comprehensive Aging Index Characterizing the degree of oxidative aging of polymer materials: ; ; in, The integral area of ​​the carbonyl characteristic peak. For reference peak area, Indicates the saturated carbonyl index, The integral area representing the characteristic peak of hydroxyl groups. Indicates the saturated hydroxyl index; Comprehensive Aging Index Calculated by weighted combination of carbonyl oxidation index and double bond consumption index: ; ; ; ; in, and For material adaptive coefficients, Material type parameters are indicated by characteristic peaks in the infrared spectrum. =0 indicates highly unsaturated rubber. =1 indicates medium-unsaturated rubber. =2 indicates low-unsaturation rubber; for saturated polymers that do not contain carbon-carbon double bonds, ; S5. Construct a decision feature vector based on the two-dimensional normalized response vector, state characterization parameters, and environmental impact parameters. Perform hybrid decision-making based on the decision feature vector to obtain the risk level, and determine the detection method according to the risk level. Step S5 specifically includes: Source identification characteristics are determined based on the confidence level of the pollutant source type. Source determination characteristics Equal to maximum similarity ; Based on the comprehensive aging index in the state characterization parameters Determine the characteristics of aging Characteristics of aging Equal to the comprehensive aging index The normalized value; Environmental risk characteristics were determined based on environmental impact parameters such as traffic flow, rainfall intensity, and distance from roads at the sampling sites. Environmental risk characteristics Calculated by weighted combination of multiple environmental parameters; Source determination characteristics Characteristics of aging and environmental risk characteristics Combined into decision feature vector ; Risk levels are derived through a hybrid decision-making process based on decision feature vectors. This hybrid decision-making method combines mandatory judgment rules and a risk assessment decision tree model. The mandatory judgment rule is based on the carbonaceous characteristic index... Exceeding the preset high carbon content threshold and the comprehensive aging index Samples exceeding a preset high aging threshold are forcibly classified as high-risk. When a stream sample from a different path does not meet the mandatory classification rule, the decision feature vector is... The risk assessment decision tree model is used to classify the risk levels.

2. The rapid detection method for road runoff pollutants based on spectral analysis as described in claim 1, characterized in that... The formulas for calculating the combined Raman spectral response intensity and the combined infrared spectral response intensity are as follows: ; ; in, Indicates the overall response intensity of the Raman spectrum. This represents the lower limit of the Raman characteristic wavenumber range. This represents the upper limit of the Raman characteristic wavenumber range. Indicates the Raman spectral intensity after background correction. Represents the Raman shift wavenumber. Indicates baseline intensity. Indicates the overall response intensity of the infrared spectrum. Indicates wavelength. This indicates the upper limit of the infrared characteristic wavenumber range. This represents the lower limit of the infrared characteristic wavenumber range. Indicates the infrared spectrum at wavenumber The absorbance at that location.

3. The rapid detection method for road runoff pollutants based on spectral analysis as described in claim 1, characterized in that... When similarity When the similarity is less than the threshold, multiple pollutants coexist in the water sample along the flow path. The two-dimensional normalized response vector of the sample is fitted using the non-negative least squares method, which is a linear combination of multiple standard pollutant reference vectors. The optimization problem is then solved. The constraints are: ≥0 and The linear combination coefficients obtained by solving Let be the relative contribution rate of pollutant j, where Represents a two-dimensional normalized response vector. Let represent the two-dimensional normalized reference vector of the j-th standard pollutant.

4. The rapid detection method for road runoff pollutants based on spectral analysis as described in claim 1, characterized in that... The construction of the risk assessment decision tree specifically includes: Define risk level as the target variable Calculate the initial entropy of the training set. : ; in, Entropy of risk level, The risk level is The sample proportion Risk level category, Indicates low risk. Indicates medium risk. Indicates high risk; Calculate the information gain for each decision feature in the decision feature vector. : ; in, For decision feature index, For the first Information gain of each decision feature Given decision features Conditional entropy under certain conditions; The decision feature with the largest information gain is selected as the splitting feature of the current node. Based on the value range of the decision feature, the sample subset is divided and the subtree is recursively constructed until the stopping condition is met to form a leaf node and output the corresponding risk level.

5. The rapid detection method for road runoff pollutants based on spectral analysis as described in claim 4, characterized in that... The method for determining the detection method based on the risk level specifically includes: When the risk level is low, a portable Raman spectrometer is used to quickly scan the water samples along the channel path to obtain the main pollutant types and concentration ranges, and to establish a low-risk sample database. When the risk level is medium risk, the water samples from the roadway will be sent to the laboratory for qualitative and quantitative analysis of organic pollutants, analysis of heavy metal elements, and identification of microplastic types to obtain the precise concentration and chemical composition of the pollutants. When the risk level is high, the water samples from the route will be sent to a qualified third-party testing institution. High-resolution mass spectrometry will be used to screen for unknown pollutants and perform targeted quantitative analysis to obtain trace concentrations, chemical structures, morphological characteristics and source analysis of the pollutants, and an emergency response mechanism will be activated.

6. The rapid detection method for road runoff pollutants based on spectral analysis as described in claim 1, characterized in that... Step S1 specifically includes: After collecting water samples from the path, primary filtration was performed using a first filter membrane with a pore size of 10 micrometers to remove large suspended particles, resulting in the first filtrate and the particulate matter retained by the first filter membrane. The first filtrate was subjected to secondary filtration using a second filter membrane with a pore size of 0.45 micrometers to trap fine particulate matter, resulting in the second filtrate and the particulate matter trapped by the second filter membrane. The particulate matter retained by the first filter membrane and the particulate matter retained by the second filter membrane were mixed and then added to ultrapure water to prepare a suspension. Raman spectroscopy was performed on the suspension to obtain Raman spectral data. The particulate matter trapped in the second filter membrane is dried, and the dried filter membrane is subjected to Fourier transform infrared spectroscopy to obtain infrared spectral data.

7. A rapid detection system for road runoff pollutants based on spectral analysis, characterized in that... The method for rapid detection of road runoff pollutants based on spectral analysis as described in any one of claims 1-6 includes: The sample acquisition module is used to collect water samples from the channel path and perform gradient filtration to obtain filtrate and particulate matter retained by the filter membrane. The Raman spectrum data of the suspension of particulate matter retained by the filter membrane is obtained by Raman spectroscopy, and the particulate matter retained by the filter membrane is dried and then subjected to Fourier transform infrared spectroscopy to obtain infrared spectral data. The spectral analysis module is used to perform integral calculations on Raman and infrared spectral data within preset wavenumber intervals to obtain the Raman integrated response intensity and the infrared integrated response intensity. The pollutant identification module is used to standardize and normalize the Raman and infrared integrated response intensities to obtain a two-dimensional normalized response vector, and to identify the pollutant source type based on the similarity between the vector and the standard pollutant reference vector. The parameter calculation module is used to calculate parameters characterizing the state of pollutants based on the characteristic peak areas in Raman and infrared spectral data, thus obtaining state characterization parameters. The risk assessment module is used to construct a decision feature vector based on the two-dimensional normalized response vector, state characterization parameters, and environmental impact parameters. Based on the decision feature vector, a hybrid decision is made to obtain the risk level, and the detection method is determined according to the risk level.

Images