Water environment pollution tracing method, system, medium and equipment based on UV-Vis

Abstract

The invention discloses a water environment pollution tracing method and system based on UV-Vis, relates to the technical field of environmental protection, and aims to perform field sampling in a pollution source range, jointly identify main pollution sources influencing a drainage basin water environment by using UV-Vis and MATLAB, and provide corresponding control countermeasures in a targeted manner. In addition, a pure pollutant'fingerprint database 'is designed according to typical sudden pollution events, the influence of different training group numbers on the performance is explored, and an optimal model for predicting the pollutant concentration is established. The method can improve pollution source identification efficiency, reduce artificial identification links and cost, and provide reference for water environment management of the Bunba river basin and expansion of'benchmark fingerprints' of different water pollutants in the future.

Description

Technical Field

[0001] The present invention relates to the field of environmental protection technology, and in particular to a method, system, medium and equipment for tracing the source of water pollution based on UV-Vis. Background Art

[0002] In recent years, the water environment pressure in the basin has increased significantly, and the two national monitoring sections in the basin are both located in the downstream of the basin, and there is a lack of monitoring sections in the upstream. Therefore, if the water quality in the control section exceeds the standard, the efficiency of pollution tracing is low. At present, many scholars' research on the Bahe River Basin is mainly to establish a hydrodynamic water quality model, and by analyzing the pollutant concentration at the assessment section predicted by the water quality model under different scenarios and different schemes, compare and select the best solution, and provide solutions for the urban water quality supervision department to carry out relevant treatment measures. However, when a water pollution incident occurs suddenly, it is easy to lose timeliness when spending a lot of time comparing and selecting the best solution, which is not feasible in practice. The primary task after the pollution occurs is to determine the scope of the pollution source as soon as possible, trace the source, determine the type of pollution incident and the content of pollutants, respond quickly to emergencies, and reduce the damage caused by pollution to the environment.

[0003] Traditional hydrological monitoring technology is relatively backward, with unstable sampling capacity. It mainly relies on manual information collection and processing. For most small and medium-sized basins that may have water pollution problems, there are often few or only a few control sections, and water quality targets are only determined for the control sections, while other places in the basin are not managed in a refined manner. There are many types of pollution sources in the Bahe River Basin and they are widely distributed. In recent years, the water environment pressure in the basin has increased significantly, and the two national monitoring sections in the basin are both located downstream of the basin, and there is a lack of monitoring sections upstream. Therefore, if the water quality in the control section exceeds the standard, the pollution source tracing efficiency is low. And when a water pollution incident occurs suddenly, it takes a lot of time to compare and select the best solution, which is easy to lose timeliness, and is not feasible in practice. The existing tracing methods are restricted by various conditions, the emergency response capacity is not high, the on-site monitoring management is weak, and the tracing efficiency is low, the cost is high, and it is easy to lose timeliness. Summary of the invention

[0004] In order to overcome the shortcomings of the above-mentioned prior art, such as being restricted by various conditions, low emergency response capabilities, weak on-site monitoring and management, low traceability efficiency, high cost, and easy loss of timeliness, the main purpose of the present invention is to provide a water environment pollution traceability method, system, medium and equipment based on UV-Vis.

[0005] To achieve the above object, the present invention adopts the following technical scheme, a method for tracing the source of water pollution based on UV-Vis, comprising:

[0006] Collect a number of water environment samples to be tested, and perform index determination to obtain spectral data of water quality indicators;

[0007] The principal component analysis is used to extract the features of the spectral data and obtain the spectral features;

[0008] The spectral features are classified by a K nearest neighbor algorithm to obtain a classification result, and compared and matched with K most similar samples in samples of known categories to confirm the category data with the highest comparison overlap rate;

[0009] Based on the category data with the highest overlap rate, the category of the sample to be tested is confirmed and the traceability result is obtained.

[0010] Compared with the prior art, the beneficial effects of the present invention are as follows: through feature extraction and classification, the accuracy of identifying the source of water pollution is improved, the accuracy of tracing is improved, the data dimension is reduced through the application of PCA, the subsequent data processing process is simplified, and the complexity of data processing is reduced. By performing fixed processing and spectral measurement at the data collection site, a rapid response is achieved, the detection efficiency is improved, and it is more real-time and efficient. The SG smoothing filter is used to reduce the noise in the spectral data, improve the signal-to-noise ratio of the data, and thus improve the reliability of the tracing results. According to different water quality indicators and pollution conditions, it has good flexibility and broad application prospects, and is more widely applicable. In short, through the comprehensive use of spectral analysis, data preprocessing, feature extraction, pattern recognition and other technical means, the rapid and accurate tracing of the source of water pollution is effectively achieved, providing strong technical support for water environment management. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The drawings described herein are used to provide further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute improper limitations on the present application.

[0012] Figure 1 It is a schematic diagram of the structure of the present invention;

[0013] Figure 2 is a water system diagram of the Bahe River Basin in an embodiment of the present invention;

[0014] Figure 3 Schematic diagram of the original UV-visible absorption spectrum of a water sample in an embodiment of the present invention;

[0015] Figure 4 is a schematic diagram of the original spectrum noise reduction processing in an embodiment of the present invention;

[0016] Figure 5 is a UV-visible absorption spectrum loading diagram of a water sample in an embodiment of the present invention;

[0017] Figure 6 It is a scree plot of the UV-visible absorption spectrum of a water sample according to an embodiment of the present invention;

[0018] Figure 7 (a) is a schematic diagram of model training and testing results in an embodiment of the present invention;

[0019] Figure 7 (b) is a schematic diagram of the model performance index results in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The present invention will be further described below in conjunction with the accompanying drawings and implementation modes.

[0021] Field sampling was conducted within the pollution source range, and UV-Vis and MATLAB were used to jointly identify the main pollution sources that affect the water environment of the basin, and corresponding control countermeasures were proposed. In addition, a pure pollutant "fingerprint database" was designed based on typical sudden pollution events to explore the impact of different numbers of training groups on its performance, and an optimal model for predicting pollutant concentrations was established. This application can improve the efficiency of pollution source identification, reduce manual identification links and reduce costs, and provide a reference for the water environment management of the Bahe River Basin and the future expansion of the "benchmark fingerprints" of different water pollutants. The specific steps include the following:

[0022] Collect a number of water environment samples to be tested, and perform index determination to obtain spectral data of water quality indexes; wherein the index determination includes the following steps:

[0023] Select the water quality indicators to be tested according to the needs;

[0024] At the collection site, sulfuric acid or nitric acid is added to the water environment sample to be tested according to the physical and chemical properties of each indicator to fix it, so as to obtain the water environment sample to be tested after fixation treatment;

[0025] Based on UV-Vis, the fixed treated water environment sample to be tested is measured to obtain the spectral data of the water quality index of the water quality to be tested.

[0026] After the spectral data of the water quality index is preprocessed, the spectral data of the water quality index is preprocessed including using SG smoothing filter to perform noise reduction on the spectral data, and the noise reduction process includes:

[0027] Select the window size and confirm the data points used in the smoothing process;

[0028] Select the polynomial order;

[0029] According to the window size and the polynomial order, the filter coefficients are obtained through matrix operations;

[0030] The filter coefficients are applied to the spectral data, and for each data point, a weighted average is calculated using the points within the window and the filter coefficients to obtain the smoothed data point, i.e., the preprocessed data.

[0031] The spectral data of the pre-processed water quality index is subjected to feature extraction using principal component analysis to obtain spectral features; wherein the feature extraction of the spectral data using principal component analysis comprises the following steps:

[0032] Standardize the spectral data of water quality indicators:

[0033]

[0034] Among them, E(X i ) is the mean matrix of the original data matrix; Var(X i 0 is the variance matrix of the original data matrix;

[0035] Get the covariance matrix S of the data matrix X:

[0036]

[0037] Get the eigenvalues ​​λ of the covariance matrix i :

[0038] S=VΛV T

[0039] In the formula: Λ is a diagonal matrix; V is an orthogonal matrix, that is, VVT=1.

[0040] The spectral features are classified by a K nearest neighbor algorithm to obtain a classification result, and compared and matched with K most similar samples in samples of known categories to confirm the category data with the highest comparison overlap rate;

[0041] Based on the category data with the highest overlap rate, the category of the sample to be tested is confirmed and the traceability result is obtained.

[0042] Example

[0043] The specific application steps in the example are as follows:

[0044] 1. Sample collection

[0045] According to the layout principle, sampling was set up at the river outlet, river section and main tributaries. Samples were collected in accordance with the sampling specifications of the national environmental protection standard "Technical Regulations for the Design of Water Quality Sampling Programs (HJ495-2009)". Sampling was carried out continuously for one year, twice a month, and two bottles were collected at each sampling point. The relevant data of water quality indicators were obtained in the laboratory after field sampling.

[0046] 2. Index determination

[0047] The river water samples were tested for DO, COD, NH3-N and TP, among which DO was tested at the sampling point using a portable detector. COD, NH3-N and TP were fixed by adding sulfuric acid or nitric acid on site according to the different physical and chemical properties of the indicators, and then the collected water samples were taken to the laboratory for testing. The measurement methods of each water quality indicator are shown in Table 1.

[0048] Table 1 Water quality index determination method

[0049]

[0050] 3. Spectral data preprocessing

[0051] (1) Spectral noise processing

[0052] The accuracy of SG smoothing filter is mainly affected by the window size and the number of smoothing operations. It can meet the needs of many different occasions, and the computer only needs to run a relatively small program, which has low requirements for computer memory and data processing capabilities. Therefore, this method is relatively simpler, faster and more efficient. Moreover, compared with other similar averaging methods, this method can better retain the distribution characteristics such as relative maximum value, minimum value and width. Therefore, this application selects SG smoothing filter for noise reduction.

[0053] The principle of SG smoothing filter noise reduction is to assume that the spectral data is x, the selected window size is m=7 (m must be an odd number), the polynomial degree is n=3, the current smoothing point is x0, the first three points are: x1, x2, x3, and the last three points are: X1, X2, X3. The least squares fitting is performed using the center point and the first three points and the last three points. Each point can be expressed as a different polynomial result, so that the 7 points can be expressed as a system of equations containing (n+1=4) unknowns, m=7:

[0054]

[0055] To solve the above equation, we use the least squares method and the matrix knowledge in linear algebra. The linear equation can be expressed as:

[0056]

[0057] Right now:

[0058] A×b=x (3)

[0059] Therefore, the least squares method is used to obtain an analytical solution for b:

[0060] b * =(A t ×A) -1 ×A t ×x (4)

[0061] So the solution to this system of equations is:

[0062]

[0063] Substitute the obtained value into the equation (substitute into equation 2) to obtain the smoothed data points.

[0064] (2) Spectral feature extraction

[0065] At present, relevant research has proved that more redundant information has a great impact on the stability and prediction performance of the model. Therefore, if only noise reduction processing is used directly for UV-Vis modeling and classification, the classification accuracy of the model will be affected. Therefore, it is necessary to eliminate irrelevant variables and extract features for UV-Vis, eliminate interference information in spectral data, amplify the differences between characteristic pollutant spectral data, and reduce the dimension of UV-Vis data, so as to achieve the purpose of improving the computational efficiency and stability of the classification model. Common spectral feature extraction methods can be divided into two categories: principal component analysis (PCA) and characteristic wavelength selection. Among them, the PCA method has the characteristics of perfect theory and simple method. Therefore, this patent uses the PCA method. PCA can convert high-dimensional data into linearly unrelated variables, which is conducive to data compression and visualization, as well as the mining of compact and effective information. The reorganized new variables are called principal components, which are converted from the original variables. According to the decreasing order of the variance of the acquired data, they are the first principal component, the second principal component, and so on. For the data matrix X, there are m*n data, including m groups of observation data, each group of observation data corresponds to n variables, and the principal component analysis is performed according to the following steps:

[0066] Normalize the original data matrix:

[0067]

[0068] Among them,: E(X i ) is the mean matrix of the original data matrix; is the variance matrix of the original data matrix; obtain the covariance matrix S of the data matrix X:

[0069]

[0070] Find the eigenvalues ​​λi of the covariance matrix:

[0071] S=VΛV T (8)

[0072] Where: Λ is a diagonal matrix; V is an orthogonal matrix (ie, VVT = 1).

[0073] Find the eigenvector P i. After dimensionality reduction of the original data through principal component analysis, the m-dimensional data is compressed into k-dimensional data, which is expressed as:

[0074] X′ = t1p2 T + p2 T + … t i p i T + … t k p k T + E (9)

[0075] Among them, t i is the main component of the system; k is the number of main components, and k < n; E is the residual vector.

[0076] In practical applications, to ensure the integrity of important data information, when selecting the main components, the cumulative contribution rate should be greater than 80%. At this time, the loss of information is small, and the characteristic information is retained.

[0077] 4. Construct a pollution source identification model

[0078] K-Nearest Neighbor (KNN) is a classification algorithm with a mature theory. Assume that the training set D now has m samples, each sample corresponds to a category y, and each sample has n feature values. Then the training set can be expressed as:

[0079] D = {(x1, y1), (x2, y2), …… (x m , y m )} (10)

[0080] x i = (x i (1) , x i (2) , x i (3) …… x i (n) ) (11)

[0081] Among them, x i is a vector, and ym is the category. KNN is a simple algorithm. The idea of this classification method is that it stores all available cases and classifies new cases according to a similarity metric (for example, a distance function). To classify an unknown document, the KNN algorithm identifies the K nearest neighbors in the given document space, searches for the K most similar samples of a sample among the samples with known categories, and if the number of these K samples belonging to a certain category is the largest, then this sample also belongs to that category.

[0082] 5. Pearson correlation traceability analysis

[0083] The Pearson correlation coefficient is used to determine the linear correlation between other sampling points and the three pollution sources UV-Vis, which can be defined here as the correlation coefficient of two continuously changing water sample UV-Vis pairs (X, Y). E(X), E(Y), and E(XY) are the expected ρ of X, Y, and XY, respectively. X,Y It is equal to the covariance Cov(X,Y) of each water sample UV-Vis divided by the product of the standard deviation of each parameter (σ X ,σ Y ), the specific calculation formula of Pearson correlation is:

[0084]

[0085] 6. Model evaluation criteria

[0086] Truepositive (TP) means that the true category is a positive example, and falsepositive (FP) means that the true category is a negative example; True negative (TN) means that the true category is a negative example, and False negative (FN) means that the true category is a positive example. The calculation formula is as follows:

[0087] Accuracy: Pr = TP / (TP+FP)

[0088] Specificity: Sp = TN / (TP+FN)

[0089] Sensitivity: Sn = TP / (TP+FN)

[0090] Accuracy: Acc = (TP + TN) / (TP + TN + FT + FN).

[0091] Example:

[0092] According to the layout principle, 11 sampling points were set up at the outlet of Bahe River, river sections and the main tributary Chanhe River. Figure 2As shown (Note: It can be seen from the "Xi'an Chanba Ecological Zone Environmental Management Commissioned Pollution Source Supervisory Monitoring Service Project" that the water quality in Lantian County ~ Bahekou was good from 2019 to 2020, so fewer sampling points were set up in this section to save manpower and material resources). Samples were collected in accordance with the sampling specifications of the national environmental protection standard "Technical Regulations for the Design of Water Quality Sampling Schemes (HJ495-2009)". The sampling time is from June 2020 to May 2021, with sampling twice a month, and two bottles are collected at each sampling point. The relevant data of water quality indicators are obtained by the laboratory after field sampling. It mainly includes the drainage outlet and upstream and downstream (Y1, Y2, Y4) of the third sewage treatment plant (Y3), the drainage outlet and upstream and downstream (Y6, Y8) of the fifth sewage treatment plant (Y7), and the Xi'an World Expo Park (Y5), a total of 8 types of water environment samples, and 4 bottles of water samples are collected at each sampling point (according to the HJ493-2009 standard).

[0093] Xi'an is a typical "resource-based water-scarce" city. As the water quality of the Bahe River changes significantly during the dry and wet seasons, the sampling data are classified according to the wet season (July-October), dry season (November-February), and normal water season (March-June), and the average value of the sampling data is taken. The specific data are shown in Tables 2, 3, and 4.

[0094] For specific data, see Table 2, Table 3 and Table 4.

[0095] Table 2 Water quality indicators at monitoring points in the Bahe River Basin during the flood season (mg / L)

[0096] Flood season DO COD <![CDATA[NH3-N]]> TP S1 8.136 15.625 0.225 0.106 S2 8.675 15.125 0.187 0.078 S3 9.639 20.250 0.254 0.145 S4 8.209 18.125 0.306 0.137 S5 8.304 19.250 0.254 0.143 S6 9.843 14.875 0.195 0.119 S7 9.339 14.875 0.194 0.156 S8 8.961 13.475 0.298 0.167 S9 8.389 18.625 0.448 0.213 S10 9.064 18.750 0.785 0.217 S11 11.615 24.750 0.440 0.136

[0097] Table 3 Water quality indicators of monitoring points in the Bahe River Basin during the normal water period from June 2020 to May 2021 (mg / L)

[0098]

[0099]

[0100] Table 4 Water quality indicators at monitoring points in the Bahe River Basin during the dry season from June 2020 to May 2021 (mg / L)

[0101] Dry season DO COD <![CDATA[NH3-N]]> TP S1 10.096 13.750 1.809 0.163 S2 11.089 11.375 0.740 0.156 S3 10.846 15.500 0.749 0.230 S4 11.406 15.375 0.693 0.201 S5 10.330 14.500 0.728 0.212 S6 9.798 11.875 0.913 0.291 S7 11.967 10.875 0.655 0.212 S8 10.981 11.125 0.825 0.156 S9 10.001 18.125 0.897 0.336 S10 10.119 20.500 0.995 1.264 S11 9.405 20.500 0.380 0.104

[0102] The Sanlang Village section where the Bahe River flows into the Weihe River is not only a national key assessment section, but also a key assessment section in the "Shaanxi Province Water Function Zoning" and the "Xi'an City Water Environment Function Zoning". In this embodiment, the water quality of the sampling points is evaluated according to the following standards.

[0103] The L6S water quality detection system was used to measure each bottle of water 40 times, and a total of 1280 groups of samples were measured. The UV-Vis average value of each sampling point was calculated and plotted. Considering that there are a lot of original spectral data, only the average value is shown in this embodiment. The original spectrum is shown in Figure 3 shown.

[0104] Depend on Figure 4 It can be seen that the water sample has absorption in the ultraviolet band and has an obvious absorption peak at 210nm, indicating that the water sample components contain conjugated double bonds or monocyclic aromatic substances. The UV-Vis maximum absorption wavelengths of sampling points Y1, Y2, Y3, Y4, Y5, Y6, Y7, and Y8 are very close. The maximum absorption wavelengths of Y3 and Y4 are between 204 and 206nm, the UV-Vis maximum absorption wavelengths of Y7 and Y8 are between 202 and 204nm, the UV-Vis maximum absorption wavelength of Y5 is between 204 and 206nm, and the UV-Vis maximum absorption wavelength of Y6 is between 200 and 201nm. Studies have shown that the absorbance value at the maximum ultraviolet absorption wavelength can reflect the DOM content to a certain extent. The DOM content of sampling points in the Bahe River Basin is ranked as follows: Y7>Y6>Y8>Y5>Y4>Y3>Y2>Y1; that is, through the analysis of the UV-Vis absorbance value, it can be seen that the concentration of organic matter in the section from the intersection of Chanba River to Sanlang Village is relatively high, and the risk of exceeding the standard in this area is more likely.

[0105] Spectral noise processing, using the SG method for smoothing filtering and noise reduction, the noise reduction processing results are as follows Figure 4 shown.

[0106] Spectral feature extraction, eliminating irrelevant variables and extracting features from the denoised spectra, and enhancing the differences between pollutant spectral data. The results are as follows Figure 5 As shown in the UV-Vis absorption spectrum load diagram of water samples. From the analysis of the UV-Vis absorption spectrum load diagram of water samples, it can be seen that the contribution rate of the first two principal components can reach 94.3%. Generally speaking, when the cumulative contribution rate of the components reaches more than 80%, it indicates that the corresponding number of components can represent all the characteristic information of the original data. Figure 6 The scree plot analysis of the UV-visible absorption spectrum of water samples shows that the best number of principal components is 4.

[0107] The pollution source identification model and model evaluation were constructed. The unprocessed full spectrum and the feature-extracted spectrum were used to classify and identify UV-Vis in combination with different algorithms. 89 groups of original data from sampling points Y3, Y5, and Y7 were randomly selected with equal probability as training sets, and trained with 7:3 cross-validation. 31 groups of data were used as test sets. The specific results of the 8 classification models are shown in Table 5. The training and test results are shown in Table 5. Figure 7 As shown in a, the performance of eight classification models is evaluated by calculation. The results are as follows Figure 7 As shown in b.

[0108] Table 5 Classification results of three pollution sources under different methods

[0109]

[0110] From Table 5 and Figure 7 (a) Analysis shows that the eight classification models have relatively high recognition accuracy for the training set data, all above 95%, but the recognition accuracy for the test set data is relatively low, between 80% and 90%, indicating that the model's prediction ability is relatively poor, but the overall effect is still good. It was also found that the KMN classification model established with spectral data extracted by PCA features improved the accuracy of the training set from 95.5% to 97.8%, and the running time was reduced from 2.4692 seconds to 0.8880 seconds, more than three times faster; the SVM classification model based on PCA feature extraction of spectral data has no change in accuracy, but the running speed is more than twice as fast. Figure 7 b Analysis shows that except for model 1 (full spectrum + KNN) with a sensitivity lower than 95%, the accuracy, specificity, sensitivity and accuracy of the other models are relatively high, all above 95%, showing good generalization and robustness. The sensitivity and accuracy of model 5 are lower than those of other models (2, 3, 4, 6, 7, 8), ranging from 97% to 98%, but not much different from other models, and from Figure 7 b. It can be clearly seen that model 5 has the fastest running time and the highest running efficiency.

[0111] By integrating the classification effect of the model and comparing the classification accuracy and running speed of KNN and SVM models for different spectral data, it can be concluded that the classification model based on "PCA+KNN" is the best.

[0112] Pearson correlation tracing analysis: IBM SPSS was used to conduct Pearson correlation analysis on the UV-Vis average values ​​of sampling points Y1 to Y8 in the Bahe River Basin. The specific numerical results of similarity matching are shown in Table 6.

[0113] Table 6 Similarity matching results of UV-visible absorption spectra in the Bahe River Basin

[0114]

[0115]

[0116] From the analysis of Table 6, we can see that the UV-Vis correlation of sampling points Y1 to Y8 is highly significant. Figure 4The results are consistent. The UV-Vis of Y1~Y8 have some problems such as partial overlap of absorption wavelengths and serious collinearity between the data of long points. The UV-Vis data of similar characteristic pollutants are less different and have strong correlation. The similarity between UV-Vis at points Y1 and Y2 and the pollution source Y5 is 99.8%; the similarity between Y4 and Y3 is 99.8%, the similarity between Y4 and Y5 is 99.1%, and the similarity between Y7 is 99.2%; the similarity between Y6 and Y3 is 99.0%, the similarity between Y5 and Y7 is 99.8%, and the similarity between Y6 and Y3 is 99.0%, the similarity between Y6 and Y3 is 99.8%, and the similarity between Y6 and Y5 is 99.0%; the similarity between Y8 and Y3 is 99.1%, the similarity between Y8 and Y5 is 99.6%, and the similarity between Y7 is 99.8%. Among the sampling points Y1~Y8, the UV-Vis downstream of the drainage outlet is very similar to the drainage outlet, and the UV-Vis of the water sample at the Y8 sampling point is more similar to Y5 and Y7 than to Y3. The results of the correlation analysis show that the main source of pollution downstream of the Chan-Ba intersection (Y6) is more likely to be the Xi'an World Expo Park (Y5), and the main source of pollution in the Sanlang Village section (Y8) is more likely to be the Xi'an World Expo Park (Y5) and the Fifth Wastewater Treatment Plant (Y7). This is consistent with the results of the pollution source tracing model analysis, that is, the UV-Vis method of establishing a source tracing model is reliable.

[0117] Propose water pollution control measures

[0118] 1. Results of tracing the source of the Bahe River

[0119] UV-Vis numerical analysis shows that the amount of urban sewage discharged from the tributaries of the Chan River and along the section from the confluence of the Chan-Ba River to Sanlang Village is large and the water quality is unstable. The sections that meet the standards are still at risk of pollution. Through tracing the source of typical pollution discharge outlets, it is known that the Fifth Sewage Treatment Plant and the Xi'an World Expo Park are more likely to be the source of pollution in the Sanlang Village section.

[0120] 2. Water pollution control measures

[0121] (1) Carry out ecological restoration of the wetlands along the S107 Huanshan Road in Tangyu, Kuyu, and Daiyu of the Chan River to improve the water quality and ecological landscape of the Chan River; above the mouth of the Chan River, focus on water conservation and ecological protection, and below the mouth of the Chan River, focus on embankment management and ecological restoration. Implement ecological wetlands in the Taohuatan section of the Chan River to improve water purification capacity and reduce the amount of sewage discharged into the Ba River. It is recommended to give priority to the combination of wetlands in the river and wetlands outside the river according to local conditions.

[0122] (2) Restore the self-purification function of the water body by dredging the bottom mud in the rubber dam at the Sanlang Village section of the Bahe River or adopting measures such as aeration and oxygenation, constructing wetlands, and adding ecological floating islands.

[0123] 3. Accurately control the dam collapse flow generated by the operation of the A# rubber dam on the Chan River, increase the flow velocity of the lower reaches of the Chan River, make the water flow smooth at the intersection of the Chan and Ba Rivers, accelerate the diffusion and dilution of pollutants in the Xi'an World Expo Park and the third sewage treatment plant, and reduce the negative impact of the backflow at the intersection of the Chan and Ba Rivers caused by the erection of the A# rubber dam on the Ba River, the slowdown of the water flow velocity, and the weakening of the diffusion effect.

[0124] It should be noted that, in the present invention, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0125] The above embodiments are merely examples of the present invention and do not limit the protection scope of the present invention. All designs that are the same or similar to the present invention fall within the protection scope of the present invention.

Claims

1. A method for tracing the source of water pollution based on UV-Vis, characterized in that: include: Collect several water environment samples to be tested, and perform index determination to obtain UV-Vis spectrum data of water quality indexes; After preprocessing the UV-Vis spectral data of the water quality index, principal component analysis is used to perform feature extraction to obtain spectral features; The spectral features are classified by a K nearest neighbor algorithm to obtain a classification result, and compared and matched with K most similar samples in samples of known categories to confirm the category data with the highest comparison overlap rate; Based on the category data with the highest overlap rate, the category of the sample to be tested is confirmed and the traceability result is obtained.

2. The method for tracing the source of water pollution based on UV-Vis according to claim 1, characterized in that: The method of extracting features from spectral data using principal component analysis comprises the following steps: Standardize the spectral data of water quality indicators: Among them, E(X i ) is the mean matrix of the original data matrix; Var(X i ) is the variance matrix of the original data matrix; Get the covariance matrix S of the data matrix X: Get the eigenvalues ​​λ of the covariance matrix i : S=VΛV T In the formula: Λ is a diagonal matrix; V is an orthogonal matrix, that is, VVT=1.

3. The method for tracing the source of water pollution based on UV-Vis according to claim 1, characterized in that: The preprocessing of the spectral data of the water quality index includes performing noise reduction processing on the spectral data using SG smoothing filtering.

4. The method for tracing the source of water pollution based on UV-Vis according to claim 1, characterized in that: The noise reduction process comprises: Select the window size and confirm the data points used in the smoothing process; Select the polynomial order; According to the window size and the polynomial order, the filter coefficients are obtained through matrix operations; The filter coefficients are applied to the spectral data, and for each data point, a weighted average is calculated using the points within the window and the filter coefficients to obtain the smoothed data point, i.e., the preprocessed data.

5. The method for tracing the source of water pollution based on UV-Vis according to claim 1, characterized in that: The index determination comprises the following steps: Select the water quality indicators to be tested according to the needs; At the collection site, sulfuric acid or nitric acid is added to the water environment sample to be tested according to the physical and chemical properties of each indicator to fix it, so as to obtain the water environment sample to be tested after fixation treatment; Based on UV-Vis, the fixed treated water environment sample to be tested is measured to obtain the spectral data of the water quality index of the water quality to be tested.

6. A water environment pollution tracing system based on UV-Vis, characterized in that: include: The data acquisition module is used to obtain a number of water environment samples to be tested, and to perform index determination to obtain UV-Vis spectrum data of water quality indicators, and to perform feature extraction on the UV-Vis spectrum data using principal component analysis to obtain spectrum features; A data analysis module is used to classify the spectral features by using a K nearest neighbor algorithm to obtain a classification result, and compare and match the spectral features with the K most similar samples in the known categories to confirm the category data with the highest comparison overlap rate; The traceability result processing module is used to confirm the category of the sample to be tested based on the category data with the highest overlap rate and obtain the traceability result.

7. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

8. A computer device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method described in any one of claims 1 to 5 is implemented.

Images