Water pollution detection method based on three-dimensional fluorescence spectrum curved surface characteristics

Abstract

The invention relates to a water pollution detection method based on three-dimensional fluorescence spectrum curved surface characteristics, which comprises the following steps: constructing a spectrum curved surface taking excitation wavelength, emission wavelength and fluorescence intensity as dimensions by collecting three-dimensional fluorescence spectrum data of a water body sample; the total curvature is calculated through the curved surface integral of the Gaussian curvature, and whether new pollutants with different fluorescence wavelengths appear or not is judged through the positive value and the negative value of the total curvature; for pollutants with fluorescence peaks close to original components, identifying tiny inclination of the convex curved surface by analyzing the direction change of the normal vector sum of each point of the curved surface; fitting a spectral curved surface by adopting a singular value decomposition method, and capturing potential morphological anomaly caused by tiny change of a substance component proportion through a fitting function; and integrating difference descriptors such as the total curvature, the normal vector sum and the fitting function, and establishing a multi-dimensional feature analysis model to identify water pollution. According to the method, difference descriptors such as the total curvature, the normal vector sum and the fitting function are integrated, a multi-dimensional feature analysis model is established, accurate recognition of water pollution is achieved, and the water pollution detection level is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of water pollution detection, and in particular to a water pollution detection method based on three-dimensional fluorescence spectrum surface morphology analysis. Background Art

[0002] In recent years, with the acceleration of urbanization and the increasing complexity of industrial activities, water pollution has shown new characteristics such as diverse components, low concentrations, and hidden sources. The limitations of traditional water pollution detection technologies in early warning and accurate identification have become increasingly prominent. While existing detection methods such as chemical analysis and biosensors can achieve quantitative detection of specific pollutants, they generally suffer from long detection cycles, insufficient multi-dimensional feature analysis capabilities, and delayed response to weak anomalies. These issues make it difficult to meet the demand for early detection and early response to pollution incidents in complex water environments.

[0003] As an emerging means of water pollution detection, three-dimensional fluorescence spectroscopy can simultaneously characterize the fluorescence characteristics of multiple components in water bodies, providing rich multi-dimensional information for pollution detection. However, the current analysis of spectral data mostly stays at the two-dimensional statistical level, and the changes in the morphological characteristics of free surfaces are more sensitive than the changes in two-dimensional matrices. When pollutants with fluorescence wavelengths close to the background are introduced into the water body or the proportion of trace components changes, the spectral surface only shows local tilt or non-structural fluctuations, and traditional methods are difficult to effectively identify such pollution. In addition, the existing technology lacks differentiated detection logic for different types of pollution, resulting in insufficient pertinence and timeliness of the warning results.

[0004] To address these challenges, this technology proposes a multidimensional feature extraction method based on three-dimensional fluorescence spectral surface morphology analysis. This method constructs a monitoring system through the collaborative analysis of parameters such as the total Gaussian curvature, normal vector and direction, and spectral surface fitting error. This method overcomes the bottleneck of traditional technologies' inadequate utilization of surface morphology, achieving full-chain optimization from pollution signal capture and type identification to trend prediction. This significantly improves the detection sensitivity and anti-interference ability for subtle changes in water quality, providing a scientifically feasible technical path for the precise prevention and control of trace pollutants in water bodies, and has important practical significance for ensuring the safety of the water environment. Summary of the Invention

[0005] To overcome the insensitivity of existing water pollution detection methods to subtle changes and their inability to accurately distinguish pollution types, this paper provides a water pollution detection method based on three-dimensional fluorescence spectral surface features. This method significantly improves the sensitivity and accuracy of detecting subtle changes in water quality, enabling the identification of different types of pollutants in complex water environments, providing a more efficient and comprehensive technical solution for water pollution monitoring.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] A water pollution detection method based on the surface characteristics of three-dimensional fluorescence spectra. By collecting the three-dimensional fluorescence spectrum data of water samples, a spectral surface is constructed with the excitation wavelength, emission wavelength, and fluorescence intensity as dimensions; the total curvature is calculated by the surface integral of the Gaussian curvature, and whether new pollutants with different fluorescence wavelengths appear is judged by the positive and negative values of the total curvature; for pollutants with fluorescence peaks close to the original components, by analyzing the direction change of the sum of the normal vectors of each point on the surface, the slight inclination of the convex surface is identified; the singular value decomposition method is used to fit the spectral surface, and the potential morphological anomalies caused by the slight change of the substance component ratio are captured by the fitting function; a multi-dimensional feature analysis model is established by integrating difference descriptors such as total curvature, normal vector sum, and fitting function to identify water pollution.

[0008] Further, the method includes the following steps:

[0009] Step 1: Construct a spectral surface

[0010] Construct a three-dimensional fluorescence spectral surface through the excitation wavelength, emission wavelength, and fluorescence intensity to visually display the spatial morphology of the spectral data;

[0011] Step 2: Spectral surface morphology analysis and characteristic parameter calculation

[0012] The change of the free surface morphology characteristics is more sensitive than the change of the two-dimensional matrix and is more conducive to reflecting the slight changes of the three-dimensional fluorescence spectra. Therefore, a series of difference descriptors of the three-dimensional fluorescence spectral surface morphology characteristics can be extracted for water pollution detection. The process is as follows:

[0013] 2.1 Calculation of curvature characteristic parameters

[0014] The spectral surface of normal water quality is mainly convex. When pollutants with a large difference in fluorescence peak position from the water body itself are added to the water body, a saddle surface structure will appear on the spectral surface, resulting in changes in the original spectral surface morphology characteristics, such as Figure 2 (A). Therefore, the three-dimensional fluorescence spectrum is divided into several local regions according to the wavelength, and the morphological characteristics of each sub-surface region are calculated respectively;

[0015] The Gaussian curvature K is the product of the two principal curvatures K1 and K2 at a certain point on the surface and is an intrinsic property of the surface, as shown in formula (1):

[0016] (A). Therefore, the three-dimensional fluorescence spectrum is divided into several local regions according to the wavelength, and the morphological characteristics of each sub-surface region are calculated respectively;

[0015] The Gaussian curvature K is the product of the two principal curvatures K1 and K2 at a certain point on the surface and is an intrinsic property of the surface, as shown in formula (1):

[0016] K = K1 · K2 (1)

[0017] Among them, K represents the Gaussian curvature at a certain point, K1 represents the curvature in the direction with the largest bending degree at a certain point, K2 represents the curvature in the direction with the smallest bending degree at a certain point. Under the saddle surface structure, K1 and K2 are in opposite directions, K is negative, and the total curvature Φ of the region is also negative, as shown in formula (2):

[0018]

[0019] Among them, Φ represents the total curvature, K represents the Gaussian curvature, T represents the subsurface area, and dA represents the area element. Therefore, by monitoring whether the total curvature of each local area has a new negative value, it can be determined whether pollutants have been introduced into the water body;

[0020] 2.2 Calculation of normal vector and characteristic parameters

[0021] The presence of pollutants in water with fluorescence peaks close to those of the original sample will cause changes in the direction of the surface convexity, such as Figure 2 (B) As shown; for the three-dimensional fluorescence spectrum of normal water quality with convex surface as the main feature, the direction of the normal vector of each point is approximately pointing to the direction of increasing fluorescence intensity. Therefore, the normal vectors of each point on the surface are superimposed, and the normal vectors and It can intuitively reflect the overall convex trend of the surface, such as formula (3):

[0022]

[0023] in, Denotes the normal vector sum, represents the normal vector of the i-th point on the surface, and n represents the total number of points. Therefore, when a contaminant with a fluorescence peak close to that of the original sample is added, the normal vector sum can capture the change in the convex direction of the spectral surface;

[0024] 2.3 Calculation of surface fitting characteristic parameters

[0025] When the proportion of material components in the water environment changes, it does not lead to the appearance of a new saddle-shaped surface on the spectral surface, nor does it cause a significant change in the convex surface direction, such as Figure 2 As shown in (C), by using the singular value decomposition method, such as formula (4), we can find the slight changes in the three-dimensional fluorescence spectrum morphology:

[0026]

[0027] Where Z is the three-dimensional fluorescence spectrum matrix, f(E x ,E m ) is the fitting function of the spectral surface, r is the rank of the spectral matrix Z, σ k is the kth singular value u k (E x )The kth left singular vector, v k (E m ) is the kth right singular vector, therefore, small changes in the fluorescence surface are monitored by surface fitting;

[0028] Step 3: Comprehensive judgment of difference descriptors

[0029] Compare the calculated total curvature, normal vector sum, and fitting function parameters with the baseline values of normal water quality:

[0030] (3.1) If the local total curvature Φ shows a new negative value, it is determined that there are pollutants with fluorescence peak positions different from the fluorescence peak position of the water sample itself, such as laboratory wastewater, construction mud, etc.

[0031] (3.2) If the normal vector and the deviation θ are greater than 10°, as shown in formula (5), it is determined that the pollutant intrusion or component content change is close to the fluorescence peak, such as cafeteria wastewater or rainfall:

[0032]

[0033] Among them, θ is the normal vector and deviation, N is the normal vector and reference direction, The normal vector and the actual direction are used to determine whether there is contamination by judging the normal vector and the deviation.

[0034] (3.3) If the parameters of the fitting function of the sample to be tested are abnormal, it indicates that the spectral surface has undergone a potential subtle change. Specifically, if the rate of change of the total singular value |ΔS| is greater than 10%, as shown in formula (6) or the angle deviation of the singular vector θ u and θ v If it is greater than 10°, as shown in formulas (7) and (8), it means that there are trace pollutants or changes in environmental factors such as water temperature and pH, which cause slight anomalies in the spectral surface;

[0035]

[0036] Among them, δ k 、u k 、v k are the singular values, left singular vectors, and right singular vectors of the reference three-dimensional fluorescent surface, δ k 、u k 、v k are the singular values, left singular vectors, and right singular vectors of the three-dimensional fluorescent surface to be measured. The presence of contamination is determined by judging the error of the fitting function parameters.

[0037] Preferably, in 2.1, the three-dimensional fluorescence spectrum is evenly divided into 5×5 sub-regions according to wavelength.

[0038] Through the above process, this technical solution realizes the full-chain automated analysis from spectral data collection to feature extraction, anomaly identification and pollution judgment, significantly improving the detection sensitivity of subtle changes in water quality.

[0039] The technical concept of the present invention is as follows: when there are significant differences between the positions of the pollutant fluorescence peaks and the fluorescence peaks of the water sample itself, extract the curvature information of the spectral surface, obtain the total curvature by calculating the surface integral of the Gaussian curvature of each local area, and determine whether pollutants appear based on whether new negative values appear in the total curvature; when the pollutant fluorescence peak is close to the original sample and the total curvature does not change significantly, use the change in the sum of the normal vectors of each point on the surface to determine the pollution change trend; when the proportion of substance components in the water environment changes slightly, no new saddle-shaped surface will appear, and the convex surface direction will not change significantly either. Use the singular value decomposition method to fit the spectral surface, and monitor the slight changes in the proportion of substance components in the water environment through the surface fitting function, so as to improve the sensitivity of water pollution detection, with reference to Figure 1 。

[0040] The present invention focuses on the morphological analysis of three-dimensional fluorescence spectral surfaces, aiming to achieve highly sensitive detection of water pollution. By collecting three-dimensional fluorescence spectral data of water samples, a spectral surface is constructed with the excitation wavelength, emission wavelength, and fluorescence intensity as dimensions. Calculate the total curvature using the surface integral of the Gaussian curvature, and judge whether new pollutants with different fluorescence wavelengths appear based on the positive and negative values of the total curvature; for pollutants with fluorescence peaks close to the original components, identify the slight inclination of the convex surface by analyzing the change in the direction of the sum of the normal vectors of each point on the surface; use the singular value decomposition method to fit the spectral surface, and capture potential morphological anomalies caused by slight changes in the proportion of substance components through the fitting function. Integrate difference descriptors such as total curvature, normal vector sum, and fitting function to establish a multi-dimensional feature analysis model, achieve accurate identification of water pollution, and improve the automation level of water pollution detection.

[0041] The beneficial technical effects of the present invention are mainly manifested in: 1. Based on the morphological analysis of three-dimensional fluorescence spectral surfaces, using multi-dimensional characteristic parameters such as Gaussian curvature, normal vector sum, and surface fitting, different types of pollution signals can be accurately captured: the total curvature is sensitive to new pollutants with large differences in fluorescence wavelengths, the normal vector sum can identify the convex surface inclination of near-wavelength pollution, and the fitting error can capture the subtle changes in trace components, realizing full-chain coverage from strong-feature pollution to potential risks. 2. By setting different thresholds, a rough judgment of the pollution source can be achieved, significantly improving the timeliness and pertinence of detection, and avoiding missed detections and misjudgments of traditional methods. Description of the Drawings

[0042] Figure 1 It is a system flow chart.

[0043] Figure 2 It is a schematic diagram of difference descriptors of spectral surface characteristics under three water pollution conditions. Detailed Embodiments

[0044] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer and more understandable, the following further describes the present invention in detail with specific embodiments and accompanying drawings. Herein, the illustrative embodiments of the present invention and their descriptions are used to explain the operation process of the present invention, but not as the only operation of the present invention.

[0045] Referring to Figure 1 and Figure 2 , a water pollution detection method based on the surface characteristics of three-dimensional fluorescence spectra measures the total curvature, normal vector sum, and surface fitting value of a water sample to obtain different differential descriptors, and compares the deviation between the water sample to be measured and the reference water sample to achieve the judgment of the pollution source.

[0046] Furthermore, the method includes the following steps:

[0047] Step 1: Construct a spectral surface

[0048] Construct a three-dimensional fluorescence spectral surface through the excitation wavelength, emission wavelength, and fluorescence intensity to visually display the spatial form of spectral data;

[0049] Step 2: Analysis of the spectral surface morphology and calculation of characteristic parameters

[0050] The change in the morphological characteristics of a free-form surface is more sensitive than that of a two-dimensional matrix and is more conducive to reflecting the subtle changes in three-dimensional fluorescence spectra. Therefore, a series of differential descriptors of the morphological characteristics of the three-dimensional fluorescence spectral surface can be extracted for water pollution detection. The process is as follows:

[0051] 2.1 Calculation of curvature characteristic parameters

[0052] The spectral surface of normal water quality is mainly a convex surface. When pollutants with a large difference in fluorescence peak position from the water body itself are added to the water body, a saddle surface structure will appear on the spectral surface, resulting in changes in the original spectral surface morphological characteristics, such as Figure 2 (A). Therefore, the three-dimensional fluorescence spectrum is divided into several local regions according to the wavelength, and the morphological characteristics of each sub-surface region are calculated respectively;

[0053] The Gaussian curvature K is the product of the two principal curvatures K1 and K2 at a certain point on the surface and is an intrinsic property of the surface, as shown in formula (1):

[0054] K = K1 · K2 (1)

[0055] Among them, K represents the Gaussian curvature at a certain point, K1 represents the curvature in the direction with the largest bending degree at a certain point, K2 represents the curvature in the direction with the smallest bending degree at a certain point. Under the saddle surface structure, K1 and K2 are in opposite directions, K is negative, and the total curvature Φ of the region is also negative, as shown in formula (2):

[0056]

[0057] Among them, Φ represents the total curvature, K represents the Gaussian curvature, T represents the sub-surface area, and dA represents the area element. Therefore, by monitoring whether new negative values appear in the total curvature of each local area, it can be judged whether pollutants are introduced into the water body;

[0058] 2.2 Calculation of normal vector and characteristic parameters

[0059] The appearance of pollutants with fluorescence peaks close to the original sample in the water body will cause a change in the convex direction of the surface, as shown in Figure 2 (B); for the three-dimensional fluorescence spectrum of normal water quality mainly composed of convex surfaces, the directions of the normal vectors at each point generally approximately point to the direction of increasing fluorescence intensity. Therefore, by superimposing the normal vectors of each point on the surface, the sum of the normal vectors can intuitively reflect the overall convex trend of the surface, as shown in formula (3):

[0060]

[0061] Among them, represents the sum of the normal vectors, represents the normal vector of the i-th point on the surface, n represents the total number of points. Therefore, when pollutants with fluorescence peaks close to the original sample are added, the sum of the normal vectors can capture the change in the convex direction of the spectral surface;

[0062] 2.3 Calculation of surface fitting characteristic parameters

[0063] When the change in the proportion of substance components in the water environment does not cause a new saddle-shaped surface to appear on the spectral surface, nor cause an obvious change in the convex direction, as shown in Figure 2 (C), through the singular value decomposition method, as shown in formula (4), the subtle changes in the three-dimensional fluorescence spectrum morphology can be found:

[0064]

[0065] Among them, Z is the three-dimensional fluorescence spectrum matrix, f(E x ,E m ) is the fitting function of the spectral surface, r is the rank of the spectral matrix Z, σ k is the k-th singular value, u k (E x ) is the k-th left singular vector, v k (E m ) is the k-th right singular vector. Therefore, the subtle changes in the fluorescence surface are monitored through surface fitting;

[0066] Step 3: Comprehensive judgment of difference descriptors

[0067] Compare the calculated total curvature, sum of normal vectors, and fitting function parameters with the benchmark values of normal water quality:

[0068] (3.1) If a new negative value appears in the local total curvature Φ, it is determined that there are pollutants with fluorescence peak positions different from those of the water sample itself, such as laboratory wastewater, construction mud, etc.;

[0069] (3.2) If the deviation angle θ of the normal vector sum is greater than 10°, as shown in formula (5), it is determined that the pollutants approaching the fluorescence peak have invaded or the component content has changed, such as canteen wastewater or rainfall:

[0070]

[0071] Among them, θ is the deviation angle of the normal vector sum, is the reference direction of the normal vector sum, is the actual direction of the normal vector sum. Whether there are pollutants is determined by judging the deviation angle of the normal vector sum;

[0072] (3.3) If the fitting function parameters of the sample to be measured are abnormal, it indicates that there are potential subtle changes in the spectral surface. Specifically, if the change rate of the sum of singular values |ΔS| is greater than 10%, as shown in formula (6) or the angle deviation θ of the singular vectors u and θ v is greater than 10°, as shown in formulas (7) and (8), it indicates that there are trace pollutants or subtle abnormalities in the spectral surface caused by changes in environmental factors such as water temperature and pH;

[0073]

[0074] Among them, δ k 、u k 、v k are the singular values, left singular vectors, and right singular vectors of the reference three-dimensional fluorescence surface respectively, and δ k 、u k 、v k are the singular values, left singular vectors, and right singular vectors of the three-dimensional fluorescence surface to be measured respectively. Whether there are pollutants is judged by judging the error of the fitting function parameters.

[0075] Through the above process, this technical solution realizes the full-chain automatic analysis from spectral data acquisition to feature extraction, anomaly recognition, and pollution source recognition, significantly improving the sensitivity of water pollution detection.

[0076] Taking the detection of whether there are pollutants in the rivers on the campus of Zhejiang University of Technology as an example, the specific implementation steps are as follows:

[0077] Step 1: Collect 10 groups of water samples from the upstream, middle stream, and downstream respectively, including 30 samples, measure the fluorescence intensity of each sample in the excitation wavelength range of 220 - 600 nm and the emission wavelength range of 220 - 600 nm, and draw a three-dimensional fluorescence surface;

[0078] Step 2: Spectral Surface Morphology Analysis and Characteristic Parameter Calculation

[0079] 2.1 Total Curvature Calculation

[0080] Divide the spectral surface into 5×5 local regions (for example, take Ex = 220 - 296 nm and Em = 220 - 296 nm as a sub-region), calculate the surface integral of the Gaussian curvature K for each region, and obtain the total curvature Φ;

[0081] If Φ < 0 in a certain region, mark it as a potential pollution region, indicating the existence of new pollutants with fluorescence peaks different from the original water body;

[0082] 2.2 Normal Vector and Analysis

[0083] Calculate the normal vector of each point and superimpose them to obtain the sum of the normal vectors of the region. Compare the sum of the normal vectors of each region with the reference direction of the upstream water sample control point. If the direction deviation angle θ exceeds 10°, it is determined that there are pollutants with fluorescence peaks close to the original components;

[0084] 2.3 Surface Fitting

[0085] Use the singular value decomposition method to fit the spectral matrix, obtain the fitting function Z, retain the first 3 principal components (r = 3) to extract the main features, and calculate the fitting error;

[0086] Step 3: Pollution Situation Judgment

[0087] The curvature sum shows a new negative value: the saddle surface exists, and it is determined as a definite pollution event. It is necessary to immediately check the surrounding sewage outlets, such as laboratory wastewater, construction mud, etc.;

[0088] The direction deviation angle θ of the sum of the normal vectors exceeds 10°: The fluorescence peak of the pollutant is close to the original component wavelength or the proportion of the water body components changes, resulting in local tilting of the convex surface but no saddle shape formed. It may be caused by the discharge of cafeteria wastewater or rainfall;

[0089] The change rate of the sum of singular values |ΔS| is greater than 10%, or the angle deviation θ of the singular vectors u and θ v is greater than 10°: It may be due to trace pollutants or changes in environmental factors such as water temperature and pH that cause slight abnormalities in the spectral surface;

[0090] Normal state: If all parameters meet the reference values of the control point, it is determined that there is no significant abnormality in the water quality.

[0091] The content described in the embodiments of this specification is only a list of the implementation forms of the inventive concept and is only for illustrative purposes. The protection scope of the present invention should not be regarded as limited to the specific forms stated in this embodiment. The protection scope of the present invention also extends to equivalent technical means that can be conceived by those of ordinary skill in the art based on the inventive concept of the present invention.

Claims

1. A water pollution detection method based on the curved surface characteristics of three-dimensional fluorescence spectra, characterized in that, By collecting the three-dimensional fluorescence spectral data of water samples, a spectral surface is constructed with the excitation wavelength, emission wavelength, and fluorescence intensity as dimensions; the total curvature is calculated by the surface integral of the Gaussian curvature, and whether new pollutants with different fluorescence wavelengths appear is judged by the positive and negative values of the total curvature; For pollutants with fluorescence peaks close to the original components, by analyzing the direction change of the sum of the normal vectors of each point on the surface, the slight inclination of the convex surface is identified; the spectral surface is fitted by the singular value decomposition method, and the potential morphological anomalies caused by the slight change in the proportion of substance components are captured by the fitting function; Based on the differential descriptors such as total curvature, sum of normal vectors, and fitting function, a multi-dimensional feature analysis model is established to identify water pollution.

2. The water pollution detection method based on the three-dimensional fluorescence spectrum surface characteristics as described in claim 1, characterized in that: The water pollution detection method includes the following steps: Step 1: Construct a spectral surface A three-dimensional fluorescence spectral surface is constructed through the excitation wavelength Ex, emission wavelength Em, and fluorescence intensity Z to intuitively display the spatial morphology of the spectral data; Step 2: Analysis of the spectral surface morphology and calculation of characteristic parameters, the process is as follows: 2.1 Calculation of curvature characteristic parameters The three-dimensional fluorescence spectrum is divided into several local regions according to the wavelength, and the morphological characteristics of each sub-surface region are calculated respectively; The Gaussian curvature K is the product of the two principal curvatures K1 and K2 at a certain point on the surface, which is the intrinsic property of the surface, as shown in formula (1): K = K1·K2 (1) Where K represents the Gaussian curvature at a certain point, K1 represents the curvature in the direction with the largest bending degree at a certain point, K2 represents the curvature in the direction with the smallest bending degree at a certain point. Under the saddle surface structure, the directions of K1 and K2 are opposite, K is negative, and the total curvature Φ of the region is also negative, as shown in formula (2): Among them, Φ represents the total curvature, K represents the Gaussian curvature, T represents the sub-surface region, and dA represents the area element. Therefore, by monitoring whether a new negative value appears in the total curvature of each local region, it can be judged whether pollutants are introduced into the water body; 2.2 Calculation of the sum of normal vectors characteristic parameters For the three-dimensional fluorescence spectrum of normal water quality mainly featuring convex surfaces, the directions of the normal vectors at each point approximately point towards the direction of increasing fluorescence intensity as a whole. By superimposing the normal vectors at each point on the surface, the sum of the normal vectors can intuitively reflect the overall convexity trend of the surface, as shown in Equation (3): Among them, represents the sum of normal vectors, represents the normal vector of the i-th point of the surface, and n represents the total number of points. Therefore, when a contaminant with a fluorescence peak close to the original sample is added, the sum of normal vectors can capture the change in the convex direction of the spectral surface; 2.3 Calculation of surface fitting characteristic parameters When the change in the proportion of substance components in the water environment does not cause a new saddle surface to appear on the spectral surface, nor cause an obvious change in the convex surface direction, through the singular value decomposition method, as shown in formula (4), the slight change in the three-dimensional fluorescence spectrum morphology can be found: Among them, Z is a three-dimensional fluorescence spectrum matrix, and f(E x , E m ) is the fitting function of the spectral surface, r is the rank of the spectral matrix Z, and σ k is the k-th singular value u k (E x ) is the k-th left singular vector, and v k (E m ) is the k-th right singular vector. Therefore, the small changes in the fluorescence surface are monitored by surface fitting; Step 3: Comprehensive judgment of differential descriptors, compare the calculated total curvature, sum of normal vectors, and fitting function parameters with the benchmark values of normal water quality, the process is as follows: (3.1) If a new negative value appears in the total curvature Φ of the local region, it is determined that there are pollutants with fluorescence peak positions different from those of the water sample itself; (3.2) If the deviation degree θ of the sum of normal vectors is greater than 10°, as shown in formula (5), it is determined that pollutants with close fluorescence peaks invade or the component content changes: where θ is the normal vector and the deviation degree, is the normal vector and the reference direction, is the normal vector and the actual direction, and it is determined whether there is a pollutant by judging the normal vector and the deviation degree; (3.3) If the fitting function parameters of the sample to be measured are abnormal, it indicates that there are potential subtle changes in the spectral surface. If the change rate of the sum of singular values |ΔS| is greater than 10%, as shown in formula (6), or the deviation of the singular vector angle θ u and θ v is greater than 10°, as shown in formulas (7) and (8), it indicates that there are trace pollutants or environmental factor changes that cause subtle abnormalities in the spectral surface; Among them, δ k , u k , v k are the singular value, left singular vector, and right singular vector of the reference three-dimensional fluorescence surface respectively, and δ k , u k , v k are the singular value, left singular vector, and right singular vector of the three-dimensional fluorescence surface to be measured respectively. Whether there are pollutants is judged by judging the parameter error of the fitting function.

3. The water pollution detection method based on the three-dimensional fluorescence spectrum surface features according to claim 2, wherein: In the above 2.1, the three-dimensional fluorescence spectrum is evenly divided into 5×5 sub-regions according to the wavelength.

Images