Methods for assessing the overall toxicity of multiple pollutants in wastewater.

Abstract

A method for assessing the overall toxicity of multiple pollutants in wastewater is disclosed. The bending angle of each part of the Cryptococcus hypocastanum nematode is regarded as the posture of the nematode, and the posture of the nematode is identified using a posture feature extraction model, and the characteristics of the water sample to be measured are determined. The signature variables are obtained and the overall toxicity of the composite pollutants in the wastewater is further evaluated based on the characterization variables of the water sample to be measured.

Description

[Technical Field]

[0001] The present invention relates to the technical field of comprehensive toxicity assessment of pollutants, in particular to the total toxicity of multiple pollutants in wastewater. It relates to a method for assessing compound toxicity. [Background technology]

[0002] Many known and unknown harmful pollutants remain in wastewater and are discharged together with the wastewater. In this case, it poses a potential risk to the aquatic ecosystem and human health. To assess the overall toxic effect of substances, all substances contained in wastewater are analyzed by biotoxicity analysis. The actual toxicity of all contaminants can be comprehensively tested. The Cas hypocastanum nematode has a short life cycle and a rapid reproductive cycle, making it highly toxic to humans. Analytical methods are generally used to detect physiological indicators of Cryptococcus hypocastanum nematodes in wastewater. (changes in physiological indicators such as survival, growth and development, and proliferation), behavioral detection, and molecular biology detection ( The overall toxicity of multiple pollutants in wastewater can be assessed by measuring the expression of related genes or proteins in the body. Analyze. However, the above methods usually rely on manual visual counting to obtain the evaluation results. However, there are problems of low efficiency and large error, and therefore the interaction between multiple pollutants in wastewater is difficult. Cryptococcus hypocastanum nematodes if there is a synergistic, antagonistic or cumulative effect Existing methods for assessing the overall toxic effects of multiple pollutants in wastewater using There is a problem with the evaluation results not being comprehensive enough. Summary of the Invention [Problem to be solved by the invention]

[0003] The present invention provides a method for evaluating the overall toxicity of multiple pollutants in wastewater, comprising the following steps: The bending angle of each part of the Cryptococcus hypocastanum nematode is used as the posture of the nematode. The posture feature extraction model is used to identify the posture of the nematode and characterize the water sample to be measured. The overall toxicity of multiple pollutants in wastewater based on the characterization variables of the water samples to be obtained and measured. The above process is more efficient and has less error, and the interaction between multiple pollutants in wastewater is further evaluated. The effect of the compound on the behavior of Cryptococcus hypocastanum nematodes was avoided, and the evaluation results were It can be more comprehensive and more accurate. [Means for solving the problem]

[0004] To achieve the above objectives, the present invention adopts the following technical solutions: The method for assessing the overall toxicity of multiple pollutants in wastewater comprises the following steps: Cryptococcus hypocastana in water samples to be measured using a posture feature extraction model Identifying multiple motility videos of the nematode and determining the characterization variables of the water sample to be measured; Here, the posture feature extraction model is a graph neural network model, and the posture feature to be measured is The water sample was wastewater containing multiple pollutants, and the multiple pollutants were perfluorooctane sulfonyl ether (PFA) and methyl methyl ether (PMMA). Contains carboxylic acid, perfluorooctanoic acid, and nonylphenol, and each The movement videos were all video images of the posture of Cryptococcus hypocastanum nematodes. The posture of the nematode includes the bending angle of each part of the Cryptococcus hypocastanum nematode. Based on the characterization variables of the water sample to be measured, cryptococcal Determine the nematode behavior score for Bacillus hypocastanum nematodes Characterization variables of the water sample to be measured and Cryptococcus in the water sample to be measured · Based on the nematode behavior score of Hippocastanum nematodes, determine the combined toxicity of the combined pollutants in the water sample to be measured of. As another aspect of the present invention, by means of the posture feature extraction model, identify multiple motion videos of Cryptococcus in the water sample to be measured The step of determining the characterization variables of the water sample to be measured includes the following: For each motion video in the multiple motion videos of Cryptococcus hippcastanum nematodes in the water sample to be measured perform image analysis for each frame of each motion video, obtain an angle matrix including the temporal and spatial change characteristics of the nematode posture and identify the angle matrix of each motion video in the multiple motion videos of Cryptococcus hippcastanum nematodes in the water sample to be measured by means of the posture feature extraction model, and obtain the characterization variables of the water sample to be measured As another aspect of the present invention, for each motion video in the multiple motion videos of Cryptococcus hippcastanum nematodes in the water sample to be measured, performing image analysis for each frame of each motion video and obtaining an angle matrix including the temporal and spatial change characteristics of the nematode posture includes the following : Identify Cryptococcus hippcastanum nematodes as the identification target, perform identification for each frame of each motion video, obtain the identification target in the video screen of each frame of each motion video, and evenly divide the identification target into k ( 10 < k ≤ 21, k ∈ Z) segments, and use them as k target variables Measure the curvature angle θ (0 ≤ θ ≤ π, r ad) of the k target variables at consecutive t (t ≥ 7200) time points, arrange the curvature angles θ of the k target variables in each motion video in a time series, and form an angle row ​​​​​​where the angle matrix is ​​a t×k matrix, the rows of the angle matrix represent the target variables, and the columns represent the time The elements in the angle matrix represent the curvature angles of k target variables at t consecutive time points. The angle is θ degrees. In another aspect of the present invention, the cryptograms in the water sample to be measured by the posture feature extraction model are Identifying the angle matrix of each locomotion video in multiple locomotion videos of the nematode Coccus hypocastanum , Obtaining characterization variables of the water sample to be measured includes: Multiple movements of Cryptococcus hypocastanum nematodes in a water sample to be measured in a video For each exercise video: Assemble the angle matrix of the motion video in chronological order into a two-dimensional array dataset with k columns and m rows. Here, the rows of the two-dimensional array dataset represent time points, the columns represent the target variable, and The elements in the sequence dataset are determined by the bending angles of k target variables at m consecutive time points. θ, The posture feature extraction model is based on k eyes corresponding to the first time point to the m-1th time point. The bending angles of the k target variables corresponding to the mth time point are predicted from the bending angles of the target variables, and Calculate the characterization variables of the exercise videos, and each characterization variable of the exercise video satisfies the following formula: JPEG0007774204000002.jpg13101where, JPEG0007774204000003.jpg915 represents the characterization variables of each motion video, JPEG0007774204000004.jpg915 is a 1×(m-1) matrix, JPEG0007774204000005.jpg69 represents k target variables, JPEG0007774204000006.jpg730 indicates that x and y are two target variables that interact with each other among k target variables, and 1 ≦x≦k, 1≦y≦k, x,y∈Z, JPEG0007774204000007.jpg1321 is the first m - 1 time points JPEG0007774204000008.jpg55 represents the effect of the bending angle on the bending angle at the first m - 1 time points y, JPEG0007774204000009.jpg1321 is a (m - 1)×1 matrix, JPEG0007774204000010.jpg1012 is the predicted bending angle of the i - th target variable at the m - th time point, 1 ≤ i ≤ k, i ∈ Z, f (·) is a function of the posture feature extraction model, Among multiple motion videos of Cryptococcus hippostanamus nematodes in the water sample to be measured collect the characterization variables of each motion video, and obtain the characterization variables { JPEG0007774204000011.jpg915, 1 ≤ x ≤ k, 1 ≤ y ≤ k, x, y ∈ Z} of the water sample to be measured. As another aspect of the present invention, based on the characterization variables of the water sample to be measured, determine the nematode behavior score of Cryptococcus hippostanamus nematodes in the water sample to be measured includes the following: For the characterization variables { JPEG0007774204000012.jpg915, 1 ≤ x ≤ k, 1 ≤ y ≤ k, x, y ∈ Z} of the water sample to be measured, treat the variables with x > y in the characterization variables of the water sample to be measured as forward characterization variables, and treat the variables with x < y in the characterization variables of the water sample to be measured as backward characterization variables, and treat the variables with x = y in the characterization variables of the water sample to be measured as self - characterization variables, calculate the nematode behavior score of Cryptococcus hippostanamus nematodes in the water sample to be measured, and the nematode behavior score of Cryptococcus hippostanamus nematodes in the water sample to be measured satisfies (|F(Ψ)| = |0.26a + 0.26b + 0.48c|), where and calculate, and the nematode behavior score of Cryptococcus hippostanamus nematodes in the water sample to be measured (|F(Ψ)| = |0.26a + 0.26b + 0.48c|), where |F(Ψ)| is the number of Cryptococcus hypocastanum nematodes in the water sample to be measured represents the behavior score, a represents the mean of the forward characterization variables, and b represents the mean of the backward characterization variables represents the value, and c represents the mean value of the self-characterization variable. In another aspect of the present invention, the characterization variables of the water sample to be measured and the water sample to be measured are Based on the nematode behavior score of Cryptococcus hypocastanum nematodes in the pool, Determining the overall toxicity of multiple contaminants in a water sample includes: Nematode behavior score |F for Cryptococcus hypocastanum nematodes in the water sample to be measured Determine the magnitude of (Ψ)|: When |F(Ψ)|≦1.1, the composite contaminants in the water sample to be measured are Cryptococcus -Does not affect the behavior of the nematodes of the nematodes of Hippocastanum nematode and does not affect the behavior of the complex contaminants in the water sample to be measured. Demonstrate that the dyes pose no environmental risks, When |F(Ψ)|>1.1, the composite contaminants in the water sample to be measured are Cryptococcus · Multiple contaminants in water samples that affect the nematode behavior of the nematode Hippocastanum nematode and should be measured The toxicity of the water sample is shown to be present and the characterization variables to be measured are integrated into the comprehensive toxicity model. Input and integrated toxicity model to measure Cryptococcus hypocastanum in water samples The acute toxicity index of the nematode is output. num nematode and 10% increase in Cryptococcus hypocastanum nematode Including growth inhibitory concentrations, Next, the 10% body length inhibitory concentration of Cryptococcus hypocastanum nematodes and Cryptococcus The mean 10% growth inhibitory concentration of the nematode, Cas hypocastanum, in the water sample to be measured This is the evaluation result of the overall toxicity of complex pollutants. In another aspect of the present invention, the comprehensive toxicity model is a support vector machine model, a gradient booster model, or a Sting tree model, Bayesian network model, k-nearest neighbor model, random forest The models are: st model, decision tree model, Lasso model, or XGBoost model. [Effects of the Invention]

[0005] Compared with the prior art, the present invention has the following significant beneficial results: The present invention provides a method for evaluating the overall toxicity of multiple pollutants in wastewater, and Multiple Cryptococcus hypocastanum nematodes in water samples to be measured by the model Identifying the bending angle of each part of the Cryptococcus hypocastanum nematode through the motion video and characterization of the nematode posture of Cryptococcus hypocastanum After obtaining the characterization variables, the clean water content in the water sample to be measured is calculated from the characterization variables. Determine the Nematode Behavior Score of Putococcus hypocastanum Nematode and Characterize Variables and Nematodes Based on the behavior score, the comprehensive toxicity of multiple pollutants in wastewater is evaluated. Identifying nematode postures using the extracted model is less time-consuming and labor-intensive than manual counting. It is more efficient and has less error, saving power. By adjusting the bending angle of each part to the posture of the nematode, synergy and antagonism between the multiple pollutants in the wastewater can be or cumulative effect of Cryptococcus hypocastanum nematode on multiple behavioral indicators (head shaking frequency) This can avoid the possibility that factors such as the degree of bending, frequency of bending, etc. may interfere with the evaluation process, resulting in better evaluation results. It will be more accurate and comprehensive. [Brief explanation of the drawings]

[0006] [Figure 1]1 is a schematic diagram of the method for evaluating the overall toxicity of multiple pollutants in wastewater provided by the examples of the present application. [Figure 2] FIG. 2 is a second schematic diagram of the method for evaluating the overall toxicity of multiple pollutants in wastewater provided by the examples of the present application. DETAILED DESCRIPTION OF THE INVENTION

[0007] The methods and apparatus provided by the examples of this application are useful for determining the overall toxicity of multiple pollutants in wastewater. Regarding the evaluation, the nematode posture of Cryptococcus hypocastanum in the wastewater was It is possible to evaluate the overall toxicity of multiple pollutants. Cryptococcus hypocastanum nematode (Caenorhabditis elegans) It should be understood that nematodes are non-toxic and harmless nematodes that can survive on their own. Dyes are complex contaminant mixtures with multiple contaminants present simultaneously in wastewater. The nematode S. hypocastanum has a short life cycle and a fast reproduction cycle, making it suitable for livestock. In the method of chemical toxicity analysis, the physiological status of Cryptococcus hypocastanum nematodes in wastewater is usually Analyzing the integrated toxicity of multiple pollutants in wastewater using indicator, behavioral, and molecular biological detection In the examples of this application, we use behavioral detection of Cryptococcus hypocastanum nematodes. Realize comprehensive toxicity analysis of complex pollutants in wastewater. Background of the study: Detection of multiple pollutants in wastewater using Cryptococcus hypocastanum nematodes Existing methods for assessing overall toxic effects have high uncertainty and the assessment results are not sufficiently comprehensive. To solve this problem, we investigated the curvature of each part of the Cryptococcus hypocastanum nematode. The angle is the posture of the nematode, and the posture of the nematode is identified using a posture feature extraction model. The characterization variables of the water sample to be measured are obtained, and the characteristics of the wastewater are calculated based on the characterization variables of the water sample to be measured. The overall toxicity of multiple contaminants is further evaluated. Compared with manual counting, the above process: It saves time and effort, is more efficient and has less error, and then By determining the bending angle of each part of the nematode as the posture of the nematode, the correlation between multiple pollutants in wastewater can be determined. Interactions between Cryptococcus hypocastanum nematodes and multiple behavioral indicators (head shaking frequency, body flexion) This avoids the possibility that factors such as song frequency may interfere with the evaluation process, making the evaluation results more accurate. It will be more inclusive. For example, as shown in FIG. 1, the method for determining the concentration of multiple pollutants in wastewater provided by the embodiment of the present application is as follows: Methods for assessing overall toxicity include S1 to S3. S1: Using the posture feature extraction model, Cryptococcus hypocaus in the water sample to be measured Identifying multiple motility videos of Stannum nematodes and determining the characterization variables of water samples to be measured do. Here, the posture feature extraction model is a graph neural network model. The characterization variables of the water sample are the Cryptococcus hypocastreus in the water sample to be measured. The water sample to be measured contains multiple pollutants. The complex pollutants are perfluorooctanesulfonic acid, perfluorooctane, Each exercise video in the multiple exercise videos contains Includes a video of nematode postures of the Putococcus hypocastanum nematode. Includes the curvature angle of each part of the Tococcus hypocastanum nematode. Each part of the Cryptococcus hypocastanum nematode is equally Includes each body part obtained by dividing. Optionally, the above complex contaminants may include salts related to perfluorooctane sulfonic acid and perfluorooctane sulfonyl fluoride (i.e., PFOSs), salts and compounds related to perfluorooctanoic acid (i.e., PFOAs), antibiotics, dioxins, methoxychlor, UV-328, and dichloran, etc., but in the examples of this application, the specific components of the above complex contaminants are not limited. In one embodiment, in combination with FIG. 1, as shown in FIG. 2, S1 includes S101 to S102. For each movement video in a plurality of movement videos of Cryptococcus neoformans nematodes in the water sample to be measured in S101, perform image analysis for each frame of each movement video to obtain an angle matrix including the temporal and spatial change characteristics of the nematode posture. Optionally, the above S101 includes S1011 to S1012. S1011, taking Cryptococcus neoformans nematodes as the identification target, perform identification for each frame of each movement video to obtain the identification target in the video screen of each frame of each movement video, and evenly divide the identification target into k (10 < k ≦ 21, k ∈ Z) segments to obtain k target variables; S1012, measure the bending angle θ (0 ≦ θ ≦ π, rad) of the k target variables at consecutive t (t ≧ 7200) time points, arrange the bending angles θ of the k target variables in each movement video in time series to obtain an angle matrix. Here, the angle matrix is a t × k matrix, the rows of the angle matrix represent target variables, the columns represent time points, and the elements included in the angle matrix are the bending angles θ of the k target variables at consecutive t time points. S102, use the posture feature extraction model to extract the posture features of Cryptococcus neoformans nematodes in the water sample to be measured, and obtain a posture feature matrix of Cryptococcus neoformans nematodes in the water sample to be measured. Optionally, the above S102 includes S1021 to S1023. S1021, input the angle matrix into a neural network model, and obtain the posture feature matrix of Cryptococcus neoformans nematodes in the water sample to be measured through the neural network model. Here, the posture feature matrix is a matrix related to the nematode posture, and the elements included in the posture feature matrix are related to the posture change characteristics of Cryptococcus neoformans nematodes. S1022, perform principal component analysis on the posture feature matrix to obtain a first principal component matrix. Here, the first principal component matrix is a matrix composed of the first principal component of the posture feature matrix, and the elements included in the first principal component matrix are related to the first principal component of the posture change characteristics of Cryptococcus neoformans nematodes. S1023, use the first principal component matrix to The angle matrix of each motion video in multiple motion videos of the castanum nematode is identified and the water to be measured is Obtain the characterization variables of the sample. Specifically, the above S102 includes S1021 to S1022. S1021, multiple Cryptococcus hypocastanum nematodes in water samples to be measured For each exercise video in the exercise video: Assemble the angle matrix of the motion video in chronological order into a two-dimensional array dataset with k columns and m rows. Here, the rows of the two-dimensional array dataset represent time points, the columns represent the target variable, and The elements in the sequence dataset are determined by the bending angles of k target variables at m consecutive time points. θ, The posture feature extraction model is based on k eyes corresponding to the first time point to the m-1th time point. The bending angles of the k target variables corresponding to the mth time point are predicted from the bending angles of the target variables, and The characterization variables of the exercise videos are calculated, and each characterization variable of the exercise video satisfies the following formula (1): Tashi: JPEG0007774204000013.jpg13101 formula (1) where: JPEG0007774204000014.jpg915 represents the characterization variables of each motion video, JPEG0007774204000015.jpg915 is a 1×(m-1) matrix, JPEG0007774204000016.jpg69 represents k target variables, JPEG0007774204000017.jpg730 indicates that x and y are two target variables that interact with each other among k target variables, and 1 ≦x≦k, 1≦y≦k, x,y∈Z, JPEG0007774204000018.jpg1321 is the first m-1 points represents the effect of the curvature angle of JPEG0007774204000019.jpg55 on the curvature angle of the first m-1 time points y, JPEG0007774204000020.jpg1321 is an (m-1) × 1 matrix, JPEG0007774204000021.jpg1012 is the predicted bending angle of the ith target variable at the mth time point, where 1≦i≦k, i∈Z, f (·) is the function of the posture feature extraction model. S1022, multiple Cryptococcus hypocastanum nematodes in the water sample to be measured Characterization variables for each exercise video are collected and used to characterize the water sample to be measured. variable{ JPEG0007774204000022.jpg915, 1 ≤ x ≤ k, 1 ≤ y ≤ k, x, y ∈ Z}. The training process of the posture feature extraction model used in S1 above will now be described in detail. Step 1. Multiple migrations of Cryptococcus hypocastanum nematodes in multiple wastewater samples a plurality of motion videos are collected, the plurality of motion videos are measured, and an angle of each motion video among the plurality of motion videos is measured; The degree matrix is ​​obtained and used as the wastewater sample behavior data set. Specifically, we collected video data of nematode movements exposed to 373 wastewater samples. Analyze the video: select k=11 to divide the worm into 11 equal segments from head to tail. The bending angle of each segment was measured at 7,200 consecutive time points. 00 matrix data, i.e., the angle matrix, is obtained. The data of the angle matrix is ​​standardized and normalized. After that, the 11 time points were collected in chronological order as a behavioral data set of 7190 11 × 1 Split the training samples into 1 matrix. Step 2: Build a graph neural network model and analyze the wastewater sample behavior data set. We train a graph neural network model using the dataset to obtain a posture feature extraction model. .

[0008] For example, the behavioral dataset obtained in step 1 is divided into 86% training set and 14% test set. We choose m=10 and select the first 10 time points of each training sample. The 11th point (current point) is used as the target variable, and the graph neural network model The goal variable output from the graph neural network model is input to the model and trained. and the corresponding 11th time point data in the behavioral dataset. The target variable output from the neural network model meets the preset accuracy rate. We adjusted the parameters of the graph neural network model and analyzed the trained graph neural network. The neural network model is used as the posture feature extraction model. S2, based on the characterization variables of the water sample to be measured, Determine the nematode behavior score for Putococcus hypocastanum nematodes. Here, we will measure the behavioral status of Cryptococcus hypocastanum nematodes in water samples. The core is a water sample containing Cryptococcus hypocastanum and nematodes. It is used to describe the effect of on nematode posture. In one embodiment, the characterization variables of the water sample to be measured { JPEG0007774204000023.jpg915, 1≦x≦k, 1≦y≦k, x, y∈Z} is acquired as an example. ~Includes S202. S201, Characterization variables of the water sample to be measured { JPEG0007774204000024.jpg915, 1≦x≦k, 1≦y≦k, x, y∈Z}, where Treat the variable with x > y in it as a forward characterization variable, which is a characterization variable of the water sample to be measured. Treat the variable with x < y in it as a backward characterization variable, which is a characterization variable of the water sample to be measured. Treat the variable with x = y in it as a self-characterization variable. S202, the nematode behavior of Cryptococcus hippostanamus nematodes in the water sample to be measured Calculate the score, and for the nematode behavior of Cryptococcus hippostanamus nematodes in the water sample to be measured The nematode behavior score satisfies the following formula: (|F(Ψ)| = |0.26a + 0.26b + 0.48c|) Formula (2) Here, |F(Ψ)| represents the nematode behavior score of Cryptococcus hippostanamus nematodes in the water sample to be measured, a represents the average value of the forward characterization variables, b represents the average value of the backward characterization variables , and c represents the average value of the self-characterization variables. S3. Based on the characterization variables of the water sample to be measured and the nematode behavior score of Cryptococcus hippostanamus nematodes in the water sample to be measured, determine the comprehensive toxicity of the composite pollutants in the water sample to be measured. Specifically, taking the case of obtaining the nematode behavior score |F(Ψ)| of Cryptococcus hippostanamus nematodes in the water sample to be measured as an example, determine the magnitude of the nematode behavior score |F(Ψ)| of Cryptococcus hippostanamus nematodes in the water sample to be measured: When |F(Ψ)| ≤ 1.1, the composite pollutants in the water sample to be measured do not affect the nematode behavior of Cryptococcus hippostanamus nematodes, indicating that the composite pollutants in the water sample to be measured have no environmental risk. When |F(Ψ)| > 1.1, the composite pollutants in the water sample to be measured affect the nematode behavior of Cryptococcus hippostanamus nematodes, and the composite pollutants in the water sample to be measured ​ The toxicity of the water sample is shown to be present and the characterization variables to be measured are integrated into the comprehensive toxicity model. Input and integrated toxicity model for Cryptococcus hypocastanum in the water sample to be measured The acute toxicity index of the nematode is output. num nematode and 10% increase in Cryptococcus hypocastanum nematode Next, the 10% body length inhibitory concentration of Cryptococcus hypocastanum nematodes was measured. The mean 10% growth inhibitory concentration of Cryptococcus hypocastanum nematodes was measured. The results should be used to evaluate the overall toxicity of the multiple pollutants in the water sample. The correlation between the above-mentioned characterization variables of the water sample to be measured and the acute toxicity index is based on the overall toxicity It should be understood that the input data for the comprehensive toxicity model can be measured. The output data of the comprehensive toxicity model are acute toxicity indicators. The comprehensive toxicity model is based on the support vector machine model and the gradient boosting tree model. , Bayesian network model, k-nearest neighbor model, random forest model, decision tree model, Lasso model or XGBoost model. In one application scenario, the training and selection process for the integrated toxicity model is as follows. Step 1: Conduct comprehensive toxicity modeling. Specifically, the nematode acute toxicity index (Cryptococcus hypocastreus) of 373 wastewater samples was num nematode and 10% increase in Cryptococcus hypocastanum nematode The growth inhibitory concentration (Growth Inhibitory Concentration) was measured, and the 10% body length inhibitory concentration and The mean 10% growth inhibitory concentrations of the nematodes Cryptococcus hypocastanum and the combined pollutants were calculated. EC: Evaluation of the overall toxicity of substances 10Let's say. Data normalization, data standardization, and characterization of the 373 wastewater samples obtained in S1 After the data downscaling and the data downscaling, the data was downscaled to 31-dimensional features. The 373 wastewater samples were divided into a 75% training set and a 25% test set. , C. elegans acute toxicity EC 10 The target variable is set as the feature, and the feature variables after processing are constructed as features. This allows for multiple toxicity models to be obtained. Optionally, the data normalization adopts the Max-Min method, and the data standardization is Z- The score method is used. The above data downscaling uses the PLS-DA method, The range after data normalization can be [-5, 5]. The above multiple toxicity models are supported by support vectors. machine model, gradient boosted tree model, Bayesian network model, k-nearest neighbor Neighborhood model, Random Forest model, Decision Tree model, Lasso model and XGBo It may include a ost model. After training, multiple toxicity models are obtained, and the post-treatment characterization variables in the test set are used as features. Each of these was input into multiple toxicity models, and a comparison was made based on the predicted values ​​output from the multiple toxicity models. The coefficient of determination R 2 Calculate the coefficient of determination R 2 The toxicity model with the highest value was used as the overall toxicity model. The coefficient of determination R 2 The formula for calculating is as follows: JPEG0007774204000025.jpg2175 formula (2) where R 2 is the coefficient of determination, and y i is the acute toxicity value of the i-th test in the test set, The mean acute toxicity value in the set, and the i-th output based on the features in the test set is the predicted value of acute toxicity, and n is the number of acute toxicity values. In one real-world application scenario, wastewater samples were collected from 62 full-flow process sections of a wastewater plant. The number of wastewater samples was 373, and the actual influent water of a city wastewater plant was was selected as the water sample to be measured, and the pH value of this water sample was 7.1. The OD measurement was 192.45 mg / L and the total organic carbon (TOC) measurement was 30.97 m The wild-type Cryptococcus hypocastanum nematode N2 used was from the United States. The strain was purchased from the Caenorhabditis elegans Genetics Center at the University of Nescota. The temperature is constantly controlled at about 20°C, and the experiment is started once the culture has stabilized after N2 synchronization treatment. When the method provided by the embodiment of the present application is applied to the above application scenario, the The characterization variables of the water samples to be measured are shown in Table 1.

[0009] Table 1: Characterization variables for water samples to be measured JPEG0007774204000026.jpg238168JPEG0007774204000027.jpg238168JPEG0007774204000028.jpg238168JPEG0007774204000029.jpg238168

[0010] By S2 of the above method, the 31 characterization variables in Table 1 are divided into 10 forward characterization variables, 1 0 backward characterization variables and 11 self-characterization variables, and The mean value of a = 1.492 for the backward characterization variables, the mean value of b = 1.138 for the self-characterization variables The average value c = 1.027 was calculated, and the amount of cryptococcal bacteria in the water sample to be measured was calculated using equation (2). A nematode behavior score of S. hypocastanum nematode |F(Ψ)| = 1.177 was calculated. When |F(Ψ)|=1.177, determine the magnitude of |F(Ψ)| using S3 of the above method. However, since |F(Ψ)|>1.1, the composite contaminants in the water sample to be measured are cryptococcal. in water samples that have a significant effect on nematode behavior and should be measured. It was shown that the combined pollutants were toxic. Based on this, the coefficient of determination of the model, R 2 Scree The algorithm supports support vector machine models, gradient boosted tree models, and Bayesian networks. Network model, k-nearest neighbor model, random forest model, decision tree model, Lass Select one model from the o model and the XGBoost model as the overall toxicity model; Among them, the coefficient of determination R of the support vector machine model 2 is the highest (R 2 =0.842) Therefore, the support vector machine model was used as the final comprehensive toxicity model. Selected. Data normalization, data standardization and data downscaling of the characterization variables of the water samples to be measured After scaling, the resulting 31-dimensional features were input into the comprehensive toxicity model. The model is based on the 10% body length inhibitory concentration of Cryptococcus hypocastanum nematodes and the Average EC (10% growth inhibitory concentration) of Coccus hypocastanum nematodes 10 =47.13 The results are used as the overall toxicity assessment results for the complex pollutants in the water sample to be directly measured. In summary, the examples provided in this application evaluate the overall toxicity of multiple pollutants in wastewater. In this method, the posture of the nematode is identified by a posture feature extraction model and counted manually. Compared to Cryptococcus aureus, it saves time and effort, is more efficient and has a smaller error rate, and By aligning the bending angles of each part of the S. hypocastanum nematode with the posture of the nematode, complex structures in the wastewater can be identified. Multiple behavioral indicators (head) of Cryptococcus hypocastanum nematode due to interactions between multiple pollutants This avoids interference with the evaluation process due to factors such as swing frequency and body bending frequency, and the evaluation results becomes more accurate and comprehensive.

Claims

1. A method for assessing the overall toxicity of complex pollutants in wastewater, comprising the steps of: identifying a plurality of movement videos of Cryptococcus hypocastanum nematodes in a water sample to be measured using a posture feature extraction model; and determining characterization variables for the water sample to be measured; wherein the posture feature extraction model is a graph neural network model; the water sample to be measured is wastewater containing complex pollutants, the complex pollutants including perfluorooctane sulfonic acid, perfluorooctanoic acid, and nonylphenol; each of the plurality of movement videos includes a video image of the nematode posture of the Cryptococcus hypocastanum nematode, the nematode posture including the bending angle of each part of the Cryptococcus hypocastanum nematode; and determining the characterization variables for the water sample to be measured based on the characterization variables for the water sample to be measured. and determining the overall toxicity of the composite contaminants in the water sample to be measured based on the characterization variables of the water sample to be measured and the nematode behavior score of the Cryptococcus hippocastanum nematode in the water sample to be measured, wherein the step of identifying a plurality of movement videos of the Cryptococcus hippocastanum nematode in the water sample to be measured using the posture feature extraction model and determining the characterization variables of the water sample to be measured includes: performing frame-by-frame image analysis of each movement video among the plurality of movement videos of the Cryptococcus hippocastanum nematode in the water sample to be measured, and obtaining an angle matrix including temporal and spatial change features of the nematode posture;and identifying an angle matrix of each of a plurality of movement videos of Cryptococcus hippocastanum nematodes in the water sample to be measured using the posture feature extraction model to obtain characterization variables of the water sample to be measured. For each of the plurality of movement videos of Cryptococcus hippocastanum nematodes in the water sample to be measured, performing frame-by-frame image analysis of each of the movement videos to obtain an angle matrix containing temporal and spatial change features of the nematode posture includes: The tanum nematode is used as a discrimination target, and discrimination is performed for each frame of each motion video. The discrimination target is obtained in the video screen of each frame of each motion video. The discrimination target is equally divided into k (10<k≦21, k∈Z) segments, which are used as k target variables. The bending angles θ (0≦θ≦π, rad) of the k target variables at t (t≧7200) consecutive time points are measured. The bending angles θ of the k target variables in each motion video are arranged in time series to obtain the angle matrix, where the angle matrix is ​​a t×k matrix, and the rows of the angle matrix represent the target variables and the columns represent the time. represents a point, and elements included in the angle matrix are bending angles θ of k target variables at t consecutive time points. Identifying the angle matrix of each movement video among the plurality of movement videos of Cryptococcus hypocastanum nematodes in the water sample to be measured using the posture feature extraction model and obtaining characterization variables of the water sample to be measured includes: for each movement video among the plurality of movement videos of Cryptococcus hypocastanum nematodes in the water sample to be measured, in chronological order, dividing the angle matrix of the movement video into k columns, The motion video is assembled into a two-dimensional array dataset including m rows, where the rows of the two-dimensional array dataset represent time points and the columns represent target variables, and the elements included in the two-dimensional array dataset are bending angles θ of k target variables at m consecutive time points. The posture feature extraction model predicts the bending angles of the k target variables corresponding to the m-th time point from the bending angles of the k target variables corresponding to the first time point to the (m-1)th time point, and calculates characterization variables for each motion video, where the characterization variables for each motion video satisfy the following formula: where: represents a characterization variable of each said exercise video; is a 1×(m−1) matrix, represents the k target variables, represents that x and y are two interacting target variables among the k target variables, where 1≦x≦k, 1≦y≦k, x, y∈Z, are the first m-1 time points represents the effect of the bending angle of the first m-1 time points y on the bending angle of the first m-1 time points y, is an (m-1) × 1 matrix, is a predicted bending angle of the i-th target variable at the m-th time point, 1≦i≦k, i∈Z, f(·) is a function of the posture feature extraction model, and characterization variables of each movement video among the plurality of movement videos of Cryptococcus hypocastanum nematodes in the water sample to be measured are collected, and the characterization variables of the water sample to be measured { , 1≦x≦k, 1≦y≦k, x, y∈Z}, and determining a nematode behavior score of Cryptococcus hypocastanum nematodes in the water sample to be measured based on the characterization variables of the water sample to be measured includes: , 1≦x≦k, 1≦y≦k, x, y∈Z}, variables of x>y among the characterization variables of the water sample to be measured are treated as forward characterization variables, variables of x<y among the characterization variables of the water sample to be measured are treated as backward characterization variables, and variables of x=y among the characterization variables of the water sample to be measured are treated as autocharacterization variables, and a nematode behavior score of Cryptococcus hippocastanum nematodes in the water sample to be measured is calculated. A method for evaluating the overall toxicity of multiple pollutants in wastewater, characterized in that the nematode behavior score of Cryptococcus hypocastanum nematodes satisfies (|F(Ψ)| = |0.26a + 0.26b + 0.48c|), where |F(Ψ)| represents the nematode behavior score of Cryptococcus hypocastanum nematodes in the water sample to be measured, a represents the average value of the forward characterization variables, b represents the average value of the backward characterization variables, and c represents the average value of the autocharacterization variables.

2. Determining the overall toxicity of the composite contaminants in the water sample to be measured based on the characterization variables of the water sample to be measured and the nematode behavior score of Cryptococcus hippocastanum nematodes in the water sample to be measured includes: determining the magnitude of the nematode behavior score |F(Ψ)| of Cryptococcus hippocastanum nematodes in the water sample to be measured, where |F(Ψ)|≦1.1 indicates that the composite contaminants in the water sample to be measured do not affect the nematode behavior of Cryptococcus hippocastanum nematodes, and the composite contaminants in the water sample to be measured do not pose an environmental risk; and |F(Ψ)|>1.1 indicates that the composite contaminants in the water sample to be measured affect the nematode behavior of Cryptococcus hippocastanum nematodes, and the composite contaminants in the water sample to be measured pose an environmental risk.

2. The method for evaluating the integrated toxicity of multiple contaminants in wastewater according to claim 1, wherein the method indicates the presence of toxicity of multiple contaminants in a water sample, and characterization variables of the water sample to be measured are input into an integrated toxicity model, and the integrated toxicity model outputs an acute toxicity index of Cryptococcus hippocastanum nematodes in the water sample to be measured, wherein the acute toxicity index includes a 10% body length inhibitory concentration of Cryptococcus hippocastanum nematodes and a 10% growth inhibitory concentration of Cryptococcus hippocastanum nematodes, and the average value of the 10% body length inhibitory concentration of Cryptococcus hippocastanum nematodes and the 10% growth inhibitory concentration of Cryptococcus hippocastanum nematodes is used as the evaluation result of the integrated toxicity of the multiple contaminants in the water sample to be measured.

3. The method for evaluating the integrated toxicity of multiple contaminants in wastewater according to claim 2, wherein the integrated toxicity model is a support vector machine model, a gradient boosting tree model, a Bayesian network model, a k-nearest neighbor model, a random forest model, a decision tree model, a Lasso model, or an XGBoost model.

Images