An isotopic-based method, system, and apparatus for wastewater heavy metal tracing monitoring

Abstract

The application belongs to the technical field of water quality monitoring, and specifically discloses a wastewater heavy metal tracing monitoring method, system and device based on isotopes, which comprises the following steps: S1, collecting historical water quality data of samples and preprocessing to generate a historical data model; extracting the pollutant data, performing data fitting thereon, and generating a water quality parameter model; S2, inputting to-be-detected water quality data into the water quality parameter model, performing water quality quality analysis, and outputting processed water quality data; S3, constructing a water quality parameter prediction model; and S4, generating a water chemical parameter report of the to-be-detected water quality and completing prediction. The wastewater heavy metal tracing monitoring method, system and device based on isotopes can not only identify and track complex composite heavy metal pollution, but also accurately judge and classify the source of pollutants through data fitting and the classification rules of the water quality filter, thereby providing comprehensive and accurate scientific basis for wastewater treatment and environmental protection.

Description

Technical Field

[0001] This application belongs to the field of water quality monitoring technology, specifically relating to a method, system and device for tracing and monitoring heavy metals in wastewater based on isotopes. Background Technology

[0002] With the rapid development of my country's economy and the increasing industrial production activities, wastewater discharge has become a growing problem. Wastewater contains large amounts of heavy metal ions, such as lead (Pb), cadmium (Cd), mercury (Hg), and arsenic (As), which pose a significant threat to the environment and human health. The cumulative, long-term, and irreversible nature of heavy metal pollution makes the problem more complex, especially when multiple heavy metals exist in the form of compound pollution, further increasing the difficulty of source tracing. Therefore, accurately tracing the sources of heavy metals in wastewater to provide a scientific basis for pollution control and environmental protection has become an urgent and crucial issue.

[0003] Existing methods for monitoring and tracing heavy metal pollution in wastewater primarily employ on-site sampling, laboratory analysis, and data source tracing. This approach relies on manual operation, is complex and time-consuming, typically requiring several days to weeks to complete the entire process from sampling to data analysis, thus failing to meet the demands of real-time monitoring. Furthermore, traditional methods largely depend on the spatial distribution characteristics of heavy metal concentrations to trace pollution sources, neglecting to adequately consider the isotopic characteristics of heavy metals. When multiple heavy metals coexist, traditional concentration distribution methods struggle to effectively separate and identify pollutants from different sources, resulting in low accuracy and resolution in source tracing and an inability to precisely determine the origin of heavy metals.

[0004] Another key problem with traditional source tracing methods is their insufficient robustness. Because changes in the aquatic environment (such as pH and redox potential) can affect the occurrence state of heavy metals, traditional methods fail to fully consider these physical and chemical factors, leading to reduced accuracy in tracing pollution sources. Furthermore, the determination of heavy metal pollutant sources relies on concentration threshold methods, neglecting the isotopic composition information of pollutants, making it difficult to distinguish the contribution rates of heavy metals from different sources in multi-source pollution scenarios. These shortcomings limit the effectiveness of existing technologies in practical monitoring, especially in multi-source tracing analysis of complex pollution, making it difficult to provide comprehensive and accurate scientific evidence for environmental governance. Summary of the Invention

[0005] To address at least one technical problem existing in the prior art, this application provides a method, system, and apparatus for tracing and monitoring heavy metals in wastewater based on isotopes.

[0006] In a first aspect, this application provides an isotope-based method for tracing and monitoring heavy metal sources in wastewater, comprising the following steps: S1. Collect historical water quality data of samples and preprocess them to generate historical data models and establish water quality indicators; The water quality data includes pollutant data such as heavy metals in the water and their isotopic composition and relative content parameters; Extract pollutant data from the historical data model, fit the data to it, and generate a water quality parameter model; The historical data models include a heavy metal isotope pollution source tracing model, an isotope ratio analysis model, and a heavy metal synergistic effect model. The pollutant data is a historical data fluctuation curve established by classifying and processing the heavy metals in water and their isotopic composition and relative content parameters. The data fitting method involves curve fitting the fluctuation curve to generate a water quality parameter model, specifically: (10); In formula (10), Ci ( t () represents the concentration of the i-th heavy metal or isotope at time t. Ci 0 represents the initial background concentration. λi It is the decay constant of a radioactive isotope. Aik , tk It represents the input intensity and occurrence time of the k-th man-made pollution event, Θ( t tk ) is a step function. αi , T It refers to the amplitude and cycle of seasonal fluctuations. βi It is a long-term linear trend. ( t ) is the random noise term. The time tt after the occurrence of the kth pollution event is... k The remaining percentage after; The water quality parameter model also includes a water quality screener, which is a water quality screening rule generated by cluster analysis based on the composition and relative content of heavy metals and their isotopes in the water quality parameter model. The water quality screener includes multiple water quality categories. The water quality category is defined as follows: pollution levels and corresponding pollutant data are determined by comparing the composition and relative content of heavy metals and their isotopes with background values. Specifically: (11); In formula (11), CPI is the comprehensive pollution index, and its calculation method is as follows: (12); In formula (12), SiIt is the first i Safety standards for heavy metals It is a single pollutant. wi It is the first i The toxicity weighting coefficients for each heavy metal, where m is the number of heavy metal species; When the overall pollution index or any single pollution factor exceeds the safety standard, a pollution risk is determined to exist; In one optional implementation, the heavy metal isotope pollution source tracing model is as follows: Concentration of heavy metals / isotopes = k1 Turbidity + k2 Conductivity +k3 Industrial emissions + (7); In formula (7), k1, k2, and k3 are all pollution source contribution coefficients. This is the error term; The isotope ratio analysis model is as follows: Isotope ratio = (8); In formula (8), when the isotope ratio is greater than the threshold, it is determined to be human-caused pollution; when the isotope ratio is less than or equal to the threshold, it is determined to be caused by natural activities. The threshold values ​​are: lead 5~20 μg / L, mercury 0.5~5 μg / L, uranium 10~100 μg / L, and cadmium 1~10 μg / L; The heavy metal synergistic effect model uses: Toxicity index = α Hg-202+β Cd-114+γ Pb-210(9); In formula (9), α, β, γ are toxicity weighting coefficients; S2. Input the water quality data to be tested into the water quality parameter model, perform water quality analysis, and output the processed water quality data; specifically: S21. Preprocess the raw data of the water quality to be tested to obtain preprocessed data. The preprocessing includes: unifying the data format, classifying and normalizing, removing noise and filling missing values. S22. Compare the preprocessed data with historical data, find and mark outliers, and generate a report. S23. First, perform background value elimination and K-means clustering on the normal data, and then normalize the processed data as a whole to ensure that the dimensions of different data are consistent. S24. Input the normal data into the water quality parameter model, calculate the weights of different pollutant data using the entropy weight method, and then perform weighted average processing to determine the water quality category and output the processed water quality data. The entropy weight method is specifically: (1); In formula (1), X' ij It is the standardized value, X ij It is the original value, min(X) j ) and max(X j These are the minimum and maximum values ​​of the j-th indicator, respectively. (2); k = 1 / ln(n)(3; In formulas (2) and (3), Ej is the entropy value of the j-th index, k is the entropy normalization constant, n is the number of samples, and ln(n) is the natural logarithm of the number of samples. (4); In formula (4), Here, m represents the weights, and m represents the number of indicators. The weighted average processing specifically involves: (5); In formula (5), It is the weighted average of each sample; S3. Based on the processed water quality data and the water quality data to be tested, calculate the prediction error of the water quality parameter model, generate a loss value, and use it as the model training value to construct a water quality prediction model. S4. Generate a report on the hydrochemical parameters of the water to be tested, and complete the prediction.

[0007] In an optional implementation, step S22, finding and marking outliers, includes the following steps: S221. By cleaning, transforming and normalizing historical data, the effective range of heavy metal isotope pollution concentration data is determined. The specific method for normalization is as follows: (6); In formula (6), X is the original data value, X min It is the minimum value of this feature, X max It is the maximum value of this feature, X norm It is the normalized value; S222. Compare the concentration of heavy metal isotopes in the preprocessed data with the effective range to determine whether it exceeds the effective range. S223. When the data exceeds the valid range, mark the data that exceeds the range, generate a report, and send the normal data to execute S23. If the valid range is not exceeded, execute S23 directly.

[0008] In one optional implementation, in step S3, the water quality parameter prediction model is: (13); In formula (13), Ci ( t +Δ t ) is the first i Heavy metals / isotopes in ( t +Δ t The predicted concentration at time ) αi It is the natural regression coefficient. wij It is the weight of the synergistic effect of metal j on metal i. βi These are multi-parameter coupling coefficients. γi It is the correction factor for natural decay of isotopes. δi It is a meteorological and environmental influencing factor. f (meteorology, t) is a meteorological function. It is a random error term; λi The decay constant of a radioactive isotope is calculated using the following formula: λ =ln(2) / T 1 / 2, of which, T Half of the life is 1 / 2.

[0009] In one optional implementation, step S3, model training includes the following steps: S31. According to the water quality screening rules, multiple sets of historical water quality data are screened sequentially to obtain the corresponding water pollutant data. S32. Based on the water pollutant data, perform data cleaning and data verification on the data to be tested to obtain cleaned data, and then write it into the water quality parameter model to obtain predicted values. Data validation methods include: parallel sample analysis, spiked recovery experiments, and the use of quality control samples. S33. Calculate the loss value between the measured data and the predicted value through the mean absolute error, train the water quality prediction model based on the loss value, and construct the water quality parameter prediction model. When the MAE of the water quality parameter model is ≤0.1mg / L, the output of the water quality parameter model is directly used as the water quality parameter prediction model. When the MAE of the water quality parameter model is ≥0.1mg / L, the historical data is collected, cleaned, transformed and normalized again. The water quality parameter model is optimized by combining machine learning algorithms through multi-model parallel training and model ensemble until the MAE meets the condition. Then, its output is used as the water quality parameter prediction model.

[0010] In one optional implementation, step S4, generating the water chemistry parameter report, includes the following steps: S41. Extract pollutant data from the sample data to be tested and match it with the water quality screening rules. S42. Determine the water quality category according to the water quality screening rules; S43. Output water quality category and generate a water chemical parameter report; The water chemistry parameter report includes: heavy metal and its isotopic composition and content, pH value, dissolved oxygen concentration, redox potential, temperature, pollution level, pollution distribution map, pollution migration map, and pollution source prediction results.

[0011] Secondly, this application also provides a monitoring system for implementing the above-mentioned wastewater heavy metal source tracing and monitoring method, comprising: The acquisition module is used to acquire heavy metal data of the water sample to be tested, and to upload the heavy metal data of the water sample to the analysis module. The analysis module stores historical water quality data of the sample and its corresponding historical data model. Then, through curve fitting, it extracts the historical data fluctuation curves of different heavy metals and their isotopes from the historical data model to obtain the water quality parameter model. Furthermore, the raw data of the water quality to be tested is preprocessed, outlier detection is performed, and normal data processing is performed. The preprocessing includes unifying the data format, classifying and normalizing, removing noise and filling missing values. The normal data processing includes background value elimination, K-means clustering, overall normalization, and weighted average processing based on entropy weight method, so as to determine the water quality category and output the processed water quality data. The historical data model includes a heavy metal isotope pollution source tracing model, an isotope ratio analysis model, and a heavy metal synergistic effect model. The water quality parameter model also includes a water quality screener, which is a water quality screening rule generated by cluster analysis based on the composition and relative content parameters of heavy metals and their isotopes in the water quality parameter model. This water quality screener includes multiple water quality categories. The water quality category is a pollution level and its corresponding pollutant data that are classified by comparing the composition and relative content parameters of heavy metals and their isotopes with background values. The training module is used to receive the test data and the predicted value, calculate the loss value, train the water quality prediction model according to the error feedback training method, and construct the water quality parameter prediction model. The output module is used to input the data to be tested into the water quality parameter prediction model, extract the water quality parameters, and generate a water chemical parameter report; The water chemistry parameter report includes: heavy metal and its isotopic composition, content, pH value, dissolved oxygen concentration, and pollution level.

[0012] Thirdly, this application also provides a wastewater heavy metal source tracing monitoring device using the monitoring system described in the second aspect above, comprising: Storage medium for storing the monitoring system, the cleaning data generated by the monitoring system, and the operating system; A processor, connected to the storage medium, is used to process water chemistry parameter report data emitted by the storage medium and convert it into a visual file; A memory, connected to the storage medium, is used to store historical data, data to be tested, and water chemistry parameter report data; External interfaces are connected to the processor, memory, and storage medium, respectively, for transmitting data; The external interfaces include: a network interface, a historical data input interface, and a test data input interface; The power supply is connected to the external interface, processor, memory, and storage medium respectively, and is used to provide electrical energy.

[0013] Fourthly, this application also provides a computer-readable medium storing a computer-readable program that can be executed by a processor to implement the isotope-based wastewater heavy metal source tracing and monitoring method described in any one of the first aspects.

[0014] This application has at least the following beneficial technical effects: To address the problems of complex operation, long monitoring cycles, insufficient data, inaccurate source tracing, and insufficient robustness in traditional wastewater heavy metal source tracing and monitoring methods, this application proposes an isotope-based wastewater heavy metal source tracing and monitoring method and system. Through historical data modeling, water quality analysis of sample data, iterative model training, and generation of water chemistry parameter reports, it achieves end-to-end monitoring from data collection to result output. Compared with traditional heavy metal pollution source tracing methods, this application can not only identify and track complex compound heavy metal pollution, but also accurately determine and classify the source of pollutants through data fitting and water quality screening rules, providing comprehensive and accurate scientific basis for wastewater treatment and environmental protection.

[0015] The isotope analysis technology employed in this application can measure the composition and relative content of heavy metals and their isotopes in water samples with high sensitivity. A water quality parameter model is generated through curve fitting, serving as the basis for analysis of the test data. Compared to traditional source tracing methods that rely on the distribution characteristics of heavy metal concentrations, this application achieves accurate differentiation and source tracing of multi-source pollution by using the composition and relative content parameters of heavy metals and their isotopes, combined with water quality category and pollution level classification rules in the water quality screener. Its advantages are particularly evident in complex pollution scenarios. Furthermore, the content and proportion of isotopes are unaffected by environmental factors, maintaining long-term monitoring stability even under changes in physicochemical conditions such as pH and redox potential in the water environment, thus ensuring the repeatability and consistency of water quality data.

[0016] In terms of detection speed and information output, the water quality parameter model, data preprocessing, and outlier labeling methods in this application, combined with techniques such as data cleaning, background value elimination, normalization, and classification, significantly improve the processing efficiency of the data to be tested. Through dynamic training of the water quality filter and water quality parameter prediction model, error feedback-based optimization of the water quality model is achieved, thereby accelerating the analysis speed of water quality data. Compared with the traditional sampling-laboratory testing-manual reporting process, this application significantly reduces manual intervention and testing cycles through automated data cleaning and classification analysis, making the processing and report generation of water quality data more efficient and faster. Furthermore, the water chemical parameter report includes multiple key water quality parameters such as heavy metal and its isotopic composition, content, pH value, dissolved oxygen concentration, and pollution level, providing comprehensive data support and decision-making basis for rapid decision-making in wastewater discharge monitoring and pollution control.

[0017] This application features a smooth overall process and clear module division of labor, with significant advantages such as high accuracy, good stability, comprehensive information, and fast detection speed. It can be widely used in the fields of wastewater discharge supervision, heavy metal pollution control, and water resource management.

[0018] The monitoring system provided in this application utilizes the high sensitivity and specificity of isotope technology to identify and track trace amounts of pollutants or specific chemical components in water, thereby achieving precise pollution source tracing. Because stable isotopes are not easily altered under natural conditions, the system provides long-term stable water quality tracking capabilities. Thanks to the high efficiency of isotope analysis technology, the system can complete water quality monitoring and source tracing in a short time, significantly improving the real-time performance and efficiency of water quality monitoring. Through the collaborative work of modules and automated database integration, a complete process is achieved, from historical data modeling, analysis of test data, data integration and verification to the automatic generation of technical reports. With its advantages of high sensitivity, long-term stability, and rapid detection, this system has broad application value in the fields of water pollution control and environmental protection. With the continuous development and improvement of isotope analysis technology, isotope-based water quality source tracing and monitoring methods will further mature and be widely applied in the fields of water resource protection and utilization. Attached Figure Description

[0019] Figure 1 This is a flowchart of the isotope-based wastewater heavy metal source tracing and monitoring method of this application; Figure 2 This is a flowchart of the isotopic composition and relative abundance analysis and iterative training in this application; Figure 3 This is a flowchart of the process for finding and marking outliers in this application; Figure 4 This is the flowchart for training the BenShen model; Figure 5 This is a structural diagram of the wastewater heavy metal tracing and monitoring system based on isotopes, as described in this application. Figure 6 This is a structural diagram of the wastewater heavy metal source tracing and monitoring device of this application; Among them: 401-Acquisition module, 402-Analysis module, 403-Training module, 404-Output module, 500-Device casing, 510-Processor, 520-Memory, 530-Storage medium, 540-Power supply, 550-Network interface, 560-Historical data input interface, 570-Test data input interface, 531-Operating system, 532-Clean data, 533-Monitoring system. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be described in more detail below with reference to the accompanying drawings. The described embodiments are only some, not all, of the embodiments of this application, and are intended to explain this application, not to limit it. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0021] The following is combined Figure 1-4 This application provides a more detailed description of the isotope-based wastewater heavy metal source tracing and monitoring method.

[0022] This application discloses an isotope-based method for tracing and monitoring heavy metals in wastewater, comprising the following steps: S1. Collect historical water quality data for the samples. This data includes pollutant data such as the composition and relative concentration of heavy metals and their isotopes in the water. Preprocess the collected data, including data cleaning and normalization, to establish a historical data model. Specifically, in this embodiment, the historical data model is shown in Tables 1 and 2 below.

[0023]

[0024]

[0025] The aforementioned historical data model also includes three sub-models: a heavy metal isotope pollution source tracing model, an isotope ratio analysis model, and a heavy metal synergistic effect model.

[0026] Specifically, the heavy metal isotope pollution source tracing model is used to quantify the contribution of different pollution sources to heavy metal / isotope pollution, as follows: Concentration of heavy metals / isotopes = k1·turbidity + k2·conductivity + k3·industrial emissions + …(7) In formula (7), the preferred concentration of heavy metal / isotope is Pb-210 (Note: Users can replace it with other heavy metals / isotopes according to actual conditions), and k1, k2 and k3 are all pollution source contribution coefficients. This is the error term.

[0027] Isotope ratio analysis models are used to determine the sources of heavy metal pollution, specifically: Isotope ratio = (8); In formula (8), when the isotope ratio is greater than the threshold, it may be due to anthropogenic pollution, such as industrial wastewater; when the isotope ratio is less than or equal to the threshold, it may be due to natural geological activity.

[0028] Specifically, the thresholds are based on lead 5~20μg / L, mercury 0.5~5μg / L, uranium 10~100μg / L, and cadmium 1~10μg / L. Users can also make fine adjustments according to the specific sample type (environment, biological, geological) and instrument performance.

[0029] The heavy metal synergistic effect model is used to calculate the comprehensive toxicity index when multiple heavy metals act together, specifically: Toxicity index = α·Hg-202 + β·Cd-114 + γ·Pb-210 (9) In formula (9), α, β, and γ are toxicity weighting coefficients, which can be determined based on field measurements or ecotoxicological data. Users can also introduce factors such as radioactive contamination (cesium-137 (Cs-137) or strontium-90 (Sr-90)), sediment accumulation, and bioaccumulation as supplementary parameters or correction factors to further improve the risk assessment.

[0030] The three models work together to identify pollution sources, determine pollution attributes, and assess toxicity risks, respectively, thus constructing an integrated judgment system for pollution cause analysis and ecological risk assessment.

[0031] Subsequently, preprocessed pollutant data was extracted from the historical data model, classified, and historical fluctuation curves of different heavy metals and their isotopes were established. The fluctuation curves were then fitted to generate a water quality parameter model, as detailed below: (10); In formula (10), Ci ( t ) is the concentration of the i-th heavy metal or isotope at time t (e.g., Pb-210, Hg-202). Ci 0 represents the initial background concentration (generally from natural geological sources); λi It is the decay constant of a radioactive isotope (if it is a non-radioactive metal, then λi=0); Aik , tk It is the first k The intensity and timing of secondary anthropogenic pollution events (e.g., industrial leaks). Θ( t tk () is a step function, which generally indicates that it takes effect after a pollution event occurs; αi , T It refers to the amplitude and cycle of seasonal fluctuations (such as the scouring effect of the rainy season). βi It is a long-term linear trend (e.g., continuous accumulation of agricultural pollution). ( t ) is the random noise term, which is generally the measurement error or unknown disturbance; The radioactive isotope in the kth pollution event after time t has elapsed since the event. t k The remaining proportion reflects the natural decay process of isotopes over time.

[0032] In this embodiment, the water quality parameter model also includes a water quality screener. Based on the composition and relative abundance parameters of heavy metals and their isotopes in the water quality parameter model, the water quality screener generates water quality screening rules through cluster analysis. These screening rules are used to classify water quality into multiple pollution levels, including lightly polluted, moderately polluted, heavily polluted, and unpolluted. The classification of water quality is based on comparing the composition and relative abundance parameters of heavy metals and their isotopes with background values ​​to determine the pollution level and its corresponding pollutant data.

[0033] The water quality screener is as follows: (11); In formula (11), CPI is the comprehensive pollution index, which is used to reflect the overall pollution level under the combined effect of multiple heavy metals. The specific calculation method is as follows: (12); In formula (12), Si It is the safety standard for the i-th heavy metal (for example, Pb-210 is generally 0.05 μg / L). This is a single pollutant factor, used to represent the multiple by which the concentration of a single heavy metal or isotope exceeds its safety standard, reflecting the extent to which an individual pollutant exceeds the standard. When this ratio is greater than 1, it indicates that there is a risk of exceeding the standard for that heavy metal; wi It is the first i The toxicity weighting coefficient of each heavy metal (e.g., the weight of Hg-202 is greater than that of Cd-114); m is the number of heavy metals considered.

[0034] Furthermore, the water quality screener employs a dual constraint principle and a "one-vote veto" principle to comprehensively assess the overall pollution level and the risk of individual exceedances. The dual constraint principle requires that both CPI and individual pollutants be considered in the pollution level assessment. Only when the CPI is below the threshold and all individual pollutants are within acceptable limits can the system be deemed safe, thus achieving simultaneous control over both overall risk and local exceedances.

[0035] The "one-vote veto" principle further strengthens the ability to identify single-item exceedances. It stipulates that if any single pollutant factor, such as heavy metal or isotope, exceeds the safety standard, it is considered to pose a pollution risk and will directly trigger an alarm. Whether the CPI exceeds the standard or not will not affect the judgment result.

[0036] The following is a specific demonstration of the above principles in this embodiment (taking Pb-210 and Hg-202 as examples): ; In this demonstration, Pb-210 is treated as a stable isotope with a radioactive decay constant set to 0, meaning the influence of natural decay processes on concentration changes is not considered. Periodic anthropogenic pollution inputs are introduced into the pollution process simulation, assuming a pollution event occurs every two years (2024, 2026, 2028), with each event causing an increase of 0.1 μg / L in Pb-210 concentration; Hg-202 is then calculated using the same method.

[0037] In this scenario, the water quality screener outputs the following results: Taking time point t=2025 as an example, the model simulation yields concentrations of 0.23 μg / L for Pb-210 and 0.0015 μg / L for Hg-202. According to the established safety standards, the limit for Pb-210 is 0.05 μg / L, and the limit for Hg-202 is 0.001 μg / L. Based on the CPI calculation formula, the calculation result is: CPI=0.23 / 0.05+0.0015 / 0.001=4.6+1.5=6.1; It can be seen that the CPI exceeds the judgment threshold (refer to Formula 11, greater than 2), and the single pollutant factor Pb-210 is 4.6, exceeding the safety standard by 4.6 times. According to the established dual constraint principle and the "one-vote veto" principle, the comprehensive index exceeds the standard and there is a single heavy metal exceeding the standard, so the water quality is judged to be severely polluted.

[0038] In practice, before collecting water samples, the sampling equipment must be cleaned and disinfected to avoid sample contamination. During sampling, appropriate sampling points and methods must be selected, such as surface sampling, well water sampling, or river sampling, and the latitude, longitude, and depth of the sampling location must be recorded in detail.

[0039] After water samples are collected, they need to be pre-treated and purified in a timely manner to remove suspended solids, organic matter, and impurities to improve the accuracy of subsequent analyses. Subsequently, heavy metal isotopes are separated and extracted from the purified water samples, usually using physicochemical methods (such as ion exchange, extraction, and distillation) to separate the isotopes from the water samples.

[0040] After isotope separation, the content and proportion of heavy metal isotopes are determined using instruments such as mass spectrometry to ascertain the isotopic composition and relative content of heavy metals in the water sample. To assess the degree of pollution, the composition and relative content of heavy metal isotopes are compared with the background values ​​of the sample. If the isotopic composition and relative content in the water sample are the same as or similar to the background values, it indicates that the water sample has not been polluted externally, or that the source of the heavy metals is the water system itself. If the isotopic composition and relative content in the water sample differ from the background values, it indicates that the water sample may have been affected by external pollution sources, or may have originated from other water systems.

[0041] After collecting historical data from a large number of samples, cluster analysis is performed on the water quality data and isotope data. The aim is to cluster similar water quality data into one class and separate dissimilar data into different categories. Various methods can be used in cluster analysis, such as hierarchical clustering and K-means clustering. Through cluster analysis, multiple water quality categories can be identified, such as slightly polluted, moderately polluted, heavily polluted, and unpolluted. Each water quality category includes the corresponding pollutant type and its concentration data.

[0042] The relative content parameters of the above-mentioned heavy metals and isotopes, as well as the classification standards of water quality categories, can be classified according to the source of the sample. The corresponding indicators and classification requirements in the "Surface Water Environmental Quality Standard" (GB 3838-2002) and the "Groundwater Quality Standard" (GB / T14848-2017) are often referenced to classify and evaluate the water quality category of the water sample.

[0043] Cluster analysis of historical water quality data can reveal the evolution and patterns of water quality changes. By determining the quantity and proportion of water quality categories in different periods and analyzing the changing trends of these categories, a scientific basis can be provided for water quality management and water environment protection.

[0044] S2. Input the water quality data to be tested into the water quality parameter model, perform water quality analysis, and output the processed water quality data; Because the water quality data to be tested may contain outliers, background noise, and missing values ​​before being input into the model, preprocessing operations such as outlier removal, background value elimination, and missing value imputation are necessary. Since these operations improve data integrity and consistency, the processed data generally needs to be normalized before being input into the water quality parameter model to ensure consistency in dimensions between different data points and avoid the model being affected by dimensional differences. The specific steps are as follows: S21. Preprocess the water quality data to be tested to obtain preprocessed data.

[0045] The main purpose of preprocessing is to remove noise from the data to ensure the accuracy of factor analysis and cluster analysis. Key steps in preprocessing include data cleaning and noise removal. After data cleaning, the goal of factor analysis is to reduce the dimensionality of the original data, extracting multiple datasets of similar indicator types, thus reducing the dimensionality and redundant information. Through these steps, the preprocessed data is finally obtained.

[0046] S22. Input the preprocessed data into the historical data model, compare it with the historical data, mark outliers and generate an outlier report, while retaining normal data.

[0047] In step S21, outliers may be present in the data due to environmental interference, equipment malfunction, or pollution source leakage during the sampling process. Therefore, the preprocessed data needs to be input into the historical data model and compared with historical data to select, mark, and confirm outliers. For confirmed outliers, the model will record their location, degree of anomaly, and possible causes in an outlier report. This report information can be used for manual review and further processing of outlier data.

[0048] Furthermore, in S22 above, finding and marking outliers includes the following steps: S221. By cleaning, transforming and normalizing historical water quality data, cleaned and transformed historical water quality data are obtained.

[0049] Data cleaning involves removing errors, anomalies, and duplicate information to improve data quality and ensure the accuracy and consistency of analysis.

[0050] Data transformation involves adjusting the format and scope of data to optimize its suitability for analytical models, enhance data interpretability, and improve model performance.

[0051] Normalization of water quality data involves separating and scaling the data to a fixed range (e.g., between 0 and 1) to eliminate differences in the magnitude of the data and ensure consistent dimensions in subsequent comparative analysis, facilitating data comparison and discrimination.

[0052] Specifically, in this embodiment, linear normalization, also known as min-max scaling, is used. It is a simple data standardization technique that scales data to a specified range, typically [0,1]. The purpose of this method is to transform each feature (attribute) value in the original dataset to this new range in order to eliminate dimensional differences between different features, or to make certain algorithms (such as neural networks) more stable and efficient during training.

[0053] The specific formula for linear normalization is as follows: (6) In formula (6), X is the original data value, X min It is the minimum value of this feature, X max It is the maximum value of this feature, X norm It is the normalized value.

[0054] Statistical analysis was performed on the historical water quality data after the above treatment to obtain the effective range of isotope pollution concentration data.

[0055] The method for determining the effective range includes, firstly, calculating descriptive statistics such as the maximum, minimum, mean, and standard deviation of pollutants from historical water quality data after cleaning and conversion. Then, based on relevant environmental standards and water quality benchmarks (GB3838-2002 and GB / T 14848-2017), confidence intervals and tolerance intervals are set to determine the normal and abnormal concentration ranges of pollutants, in order to identify potential water quality anomalies. This process provides data for subsequent comparisons and labeling, ensuring the scientific rigor and traceability of environmental monitoring and pollutant identification.

[0056] S222. Compare the heavy metal isotope concentration data in the preprocessed data with the effective range to determine whether it exceeds the effective range of isotope pollution concentration data.

[0057] S223. When the concentration data of heavy metal isotopes exceeds the valid range, mark the data as abnormal data and generate an abnormal data report, while sending the normal data to execute S23; when it does not exceed the valid range, execute S23 directly.

[0058] The purpose of data tagging is to record the data source, processing procedure, and usage history by assigning specific tagging information to water quality monitoring data. Data tagging ensures the traceability and transparency of water quality data, enabling effective tracking of data flow and source in environmental management, pollution source identification, and water quality change analysis, thereby guaranteeing the accuracy and reliability of water quality monitoring results.

[0059] The abnormal data report includes the numerical value of the abnormal data, the reason for the labeling, the criteria for determining whether it exceeds the range, and the timestamp of the abnormal label. This information will be automatically recorded and monitored to form an abnormal data report for reference by environmental management departments. This report can be used for tracing the source of pollution incidents, identifying the source of pollutants, and retrospective analysis of water quality safety monitoring, thereby providing data support and scientific basis for environmental governance and management decisions.

[0060] S23. After removing background values ​​and clustering the labeled normal data, perform overall normalization.

[0061] Background elimination is achieved by separating and removing background noise signals from normal data, highlighting the characteristics of heavy metal isotopes, thereby improving the signal-to-noise ratio of the data and ensuring the accuracy of subsequent analysis.

[0062] Simultaneously, K-means clustering is performed on multiple datasets with similar indicator types to obtain a set of candidate cluster centers. K-means clustering is a basic machine learning algorithm that divides a dataset into K mutually exclusive clusters through an iterative process. The goal is to minimize the sum of squared distances between each data point and its cluster center. The algorithm starts by randomly selecting K data points as initial cluster centers and then repeats two main steps: First, each data point is assigned to the cluster represented by the nearest cluster center; second, the center of each cluster is updated to the mean position of all data points within that cluster. This process continues until the change in cluster centers is less than a preset threshold or the maximum number of iterations is reached, thus achieving effective clustering of the data points.

[0063] After background value elimination and clustering, the data is input into the overall normalization process, resulting in processed, normal data. The purpose of overall normalization is to ensure that all data have consistent dimensions, avoiding biases in the model's classification and clustering results caused by differences in data units and numerical values, thus providing consistent input data for the model's classification and clustering analysis.

[0064] S24. Input the processed normal data into the water quality parameter model, calculate the weights of different pollutant data using the entropy weight method, then perform weighted average processing to determine the water quality category, and output the processed water quality data.

[0065] Entropy weighting is an objective weighting method that determines weights based on the information entropy of the data. Furthermore, data standardization is required to eliminate the influence of different units of measurement. The specific method is as follows: (1) In formula (1), X'ij is the standardized value, Xij is the original value, and min(Xj) and max(Xj) are the minimum and maximum values ​​of the j-th index, respectively.

[0066] (2) k = 1 / ln(n)(3) In formulas (2) and (3), Ej is the entropy value of the j-th index, k is the entropy normalization constant, which is used to standardize the calculation result of Ej so that its value range is limited to [0, 1], n is the number of samples, and ln(n) is the natural logarithm of the number of samples.

[0067] (4) In formula (4), 'm' represents the weight, and 'm' represents the number of indicators.

[0068] The weighted average processing method is as follows: (5) In formula (5), It is the weighted average of each sample.

[0069] S3. Based on the processed water quality data and the water quality data to be tested, calculate the prediction error of the water quality parameter model, generate the loss value, and use it as the model training value to construct the water quality parameter prediction model. At the same time, the processed water quality data and the water quality data to be tested are repeatedly compared during the process to verify the reliability of the water quality parameter prediction model.

[0070] Specifically, the water quality parameter prediction model is as follows: ; In formula (13), Ci ( t +Δ t ) is the first i Heavy metals / isotopes in t +Δ t Predicted concentration at any given time; αi It is the natural regression coefficient, used to reflect the inertial influence of the current concentration on future trends, and is often obtained by fitting historical data collected; wij It is the synergistic effect weight of metal j on metal i, for example: the superimposed toxicity effect of Hg and Cd; βi These are multi-parameter coupling coefficients; γi This is the isotope natural decay correction factor, used to correct for concentration changes in radioactive isotopes caused by natural decay (e.g., U-238). This factor is not calculated for non-radioactive metals. λi The decay constant of a radioactive isotope is calculated using the following formula: λ=ln(2) / T 1 / 2, of which, T Half of the life is 1 / 2; δi These are meteorological and environmental influencing factors, used to characterize the effects of meteorological conditions such as rainfall and temperature on changes in the concentration of heavy metals or isotopes. f (meteorological, t) is a meteorological function that represents the effect of environmental factors such as rainfall and temperature on the concentration change over time t. The random error term follows a normal distribution N(0,σ). 2 ), where the mean is 0 and the variance is σ², is used to describe measurement error and other unpredictable random disturbances.

[0071] The training of a water quality parameter prediction model includes the following steps: S31. According to the water quality screening rules, the historical water quality data are screened sequentially to obtain the water pollutant data corresponding to each pollution level, and further obtain the water pollution type.

[0072] The purpose of water quality screening is to classify and filter historical water quality data to identify different types of water pollution and specific pollutants. The screening process is based on water quality assessment rules. It involves extracting features from water quality element data (such as the composition and relative content of heavy metal isotopes) and comparing them with pollution level classification standards to divide the water quality data into multiple categories. Within each category, the pollution type is further matched based on its pollution characteristics.

[0073] Water pollution types typically include indicators such as chemical oxygen demand (COD), biochemical oxygen demand (BOD), and ammonia nitrogen (NH3-N), which are directly related to the pollution characteristics of water. For each water quality category, the model categorizes it into a specific pollutant dataset based on its pollution type. This pollution type matching operation not only provides targeted datasets for subsequent data cleaning and model training but also provides training samples for feature optimization of the water quality prediction model.

[0074] S32. Based on the water pollutant data, perform data cleaning and data verification on the water quality data to be tested, obtain the cleaned data, and write it into the water quality parameter model to obtain the predicted value.

[0075] The goal of data cleaning is to remove invalid, anomalous, and inconsistently formatted data to ensure the accuracy and reliability of model training and data analysis.

[0076] The data cleaning process specifically includes removing null values, extreme values, and duplicate data, using deduplication algorithms to ensure data uniqueness and consistency. Furthermore, the data format and unit dimensions are standardized, unifying the concentration values ​​of pollutant indicators (COD, BOD, NH3-N, etc.) for different pollution types into standard mg / L or ppm formats to ensure data comparability during classification and model training.

[0077] Based on data cleaning, the water quality parameter model validated the data using parallel sample analysis, spiked recovery experiments, and quality control sample methods. In the parallel sample analysis, the results of multiple parallel samples were compared to ensure data stability and consistency. In the spiked recovery experiment, known concentrations of pollutants were added to the samples, and the recovery rate was measured to confirm data accuracy. In the quality control sample method, multiple measurements of standard samples ensured data comparability and consistency, which is crucial for large-scale water quality sample testing.

[0078] After cleaning and validating the data, the cleaned data is written into the water quality parameter model, the historical dataset in the model is updated, and the corresponding predicted values ​​are output. The purpose of this operation is to merge the high-quality cleaned data with the historical data to form a larger and higher-quality training dataset.

[0079] S33. Calculate the loss value between the water quality data to be tested and the predicted value to obtain the water quality parameter model loss value. Train the water quality parameter model based on the loss value to obtain the water quality parameter prediction model.

[0080] Specifically, the water quality data to be tested is compared with the predicted values ​​in the water quality parameter model, and the prediction error of the water quality parameter model is calculated using a loss function (such as mean squared error (MSE), mean absolute error (MAE), or cross-entropy). The core objective of the loss function is to measure the difference between the model's predicted values ​​and the actual values. By dynamically changing the loss value, the parameters and weights of the water quality parameter model are dynamically adjusted.

[0081] After the loss value is calculated, an error feedback mechanism is used to feed the loss value back into the model. Backpropagation (BP) and gradient descent (SGD) are then used to adjust the model's weights and biases to minimize the loss value. To further optimize model performance, various optimization strategies can be employed during the optimization process, including genetic algorithms (GA), simulated annealing (SA), and particle swarm optimization (PSO). These algorithms improve the model's generalization ability and prediction accuracy through global search and stochastic optimization.

[0082] Taking MAE as an example: if MAE≤0.1mg / L, it means that the prediction results of the water quality parameter model are consistent with the actual data, indicating that the model has good prediction ability and can be directly output as a water quality parameter prediction model.

[0083] If MAE ≥ 0.1 mg / L, it indicates a significant difference between the two, requiring further optimization of the water quality parameter model. This generally includes: Feature selection: Through correlation analysis and feature importance assessment, prioritize the use of parameters that are significant.

[0084] Data cleaning: For some data points with outliers, which may be caused by measurement errors, these outliers are deleted.

[0085] Error analysis: If the model is found to perform poorly under certain conditions, then the data will be resampled or a different model will be used.

[0086] During the optimization process, users can introduce various machine learning algorithms (such as Support Vector Machine (SVM), Random Forest (RF), and Neural Network (NN)) for modeling as needed. Through parallel training of multiple models and model ensemble, the predictive power and stability of the water quality parameter model are further improved. By repeatedly inputting the water quality data to be tested and the optimized processed data, the loss value is dynamically calculated, and the model parameters are continuously adjusted until the loss value meets the standard (e.g., MAE ≤ 0.1 mg / L), ultimately generating a water quality parameter prediction model.

[0087] The water quality parameter prediction model can take the water quality data to be tested as input, predict the heavy metal composition, pollution level, and water quality assessment results, and output this information as a water chemistry parameter report. In actual monitoring, the output results of the water quality parameter prediction model will be used for water quality safety assessment, pollution early warning, and pollution source tracing, providing decision support for water quality management departments and environmental protection organizations.

[0088] S4. Generate a report on the hydrochemical parameters of the water to be tested, completing the prediction. The process of generating the hydrochemical parameter report includes the following steps: S41. Extract pollutant data from the water quality data to be tested and match it with water quality screening rules. Determine the water quality information that needs to be extracted. Data extraction refers to the process of retrieving and collecting specific data from a database or data source. The key is to accurately screen the required fields and records according to the purpose and scope of the report. In this process, the water quality parameter prediction model will focus on extracting key water quality information such as pH value, dissolved oxygen concentration, heavy metal composition and its isotope content from the water quality parameters and heavy metal isotope data of the sample.

[0089] Furthermore, the data extraction process begins by defining the purpose and scope of the report to ensure a clear data extraction objective and that the extracted water quality parameters support the analytical needs of the hydrochemical parameter report. Subsequently, data fields from the water quality parameter prediction model and real-time data sources are retrieved, and data related to the report, such as pH, dissolved oxygen, heavy metal concentration, and isotopic composition, are selected and compared with water quality screening rules to ensure data consistency and completeness. If data is inconsistent in format or units, the water quality parameter prediction model automatically converts it to standardized units, such as unifying pollutant concentration data to mg / L or ppm format, ensuring data comparability and consistency. Through this data extraction process, users can accurately obtain key data related to the hydrochemical parameter report from large-scale historical and real-time data, providing high-quality input data for subsequent water quality classification and report generation.

[0090] S42. Based on the water quality screening rules, determine the water quality category and further identify the pollution type and parameters.

[0091] In the process of determining water quality categories, the extracted water quality data to be tested is input into a water quality parameter prediction model. The model automatically classifies and labels the data according to water quality screening rules. The classification is mainly based on pollutant content thresholds and water quality characteristic classification standards. The model will determine the composition, content, pH value, and dissolved oxygen concentration of heavy metal isotopes, generate a water quality category label for each sample, and perform feature annotation on each category of data, such as labeling the sample as "excessive heavy metals," "excessive ammonia nitrogen," or "excessive chemical oxygen demand," etc.

[0092] To ensure data traceability and consistency, the model generates labeled records and timestamps for each category of water quality data and stores the classification results in the model log for subsequent data tracing and report generation. This classification process accurately determines water quality categories and pollution types, providing data support for report visualizations and charts, thereby improving report readability and interpretability.

[0093] S43. The water quality parameter prediction model outputs the classification results of water quality categories as a water chemical parameter report. This report includes the composition, content, pH value, dissolved oxygen concentration, and pollution level of heavy metals and their isotopes, and presents the key results of water quality monitoring in the form of visualizations and charts. The generation of the water chemical parameter report is a key output operation of the model, and its purpose is to provide complete water quality monitoring results and pollution source tracing analysis reports, which are convenient for use by water quality management departments, environmental monitoring agencies, or pollution control units.

[0094] The main contents of a water chemistry parameter report include the following parts.

[0095] The general overview section summarizes the report's purpose, scope, and main conclusions, and provides the key objectives and core findings of this water quality monitoring. The work area overview section describes the geographical location of the sampling points, the characteristics of the groundwater system, and the sampling environment, and explains the underlying hydrogeochemical theory. The work deployment section describes the sample collection methods and testing deployment, such as the time, quantity, location, and testing equipment used for sample collection. The sample collection and testing section details the testing methods, equipment, and results, and compares and interprets key water quality indicators such as pH, dissolved oxygen concentration, heavy metal content, and isotopic composition.

[0096] The pollution source analysis section is one of the core parts of the report. Through a conceptual model of groundwater pollution, combined with data simulation and source tracing calculations, it analyzes the sources and contribution rates of groundwater pollution. By simulating groundwater flow fields and pollutant migration paths, it generates pollution distribution maps and pollution migration diagrams, thus providing a visual representation of the distribution characteristics of pollution sources and pollution diffusion trends. The digital generation of these maps is a significant highlight of the report. The model uses GIS (Geographic Information System) and CAD software to graphically display the distribution of pollution sources, groundwater pollution distribution characteristics, and groundwater flow field data, creating clear, focused, and well-organized maps. The scale of these maps is dynamically adjusted according to the size of the work area, and standardized legends and annotation symbols are used to ensure readability and professionalism.

[0097] In the Problems and Recommendations section, the report summarizes the main distribution areas of pollution sources and risk factors for water pollution, and proposes improvement suggestions for pollution control and response strategies for environmental management. These recommendations include not only specific methods for pollution control measures but also directions for optimizing data monitoring, such as increasing sampling frequency or expanding the detection range to improve monitoring accuracy and timeliness. The report output includes text reports and visualizations, and computer-aided map generation and digitization ensure the report's visualization effectiveness and the accuracy of its digital archiving. The report's visualizations consist of pollution distribution maps, groundwater flow field maps, and pollutant diffusion path maps, etc. Dynamic adjustments and digital generation of the maps greatly enhance the report's readability and operability.

[0098] Based on the above monitoring methods, such as Figure 5 As shown, this application also provides a wastewater heavy metal source tracing and monitoring device, including a data acquisition module 401, an analysis module 402, a training module 403, and an output module 404.

[0099] Specifically, the acquisition module 401 connects to the real-time water quality detection module, acquires heavy metal data of the water sample to be tested, and uploads the heavy metal data of the water sample to the analysis module. The analysis module 402 stores historical water quality data of the sample and its corresponding historical data model. It then extracts the historical data fluctuation curves of different heavy metals and their isotopes from the historical data model through curve fitting to obtain a water quality parameter model. Furthermore, it inputs the water quality data to be tested into the water quality parameter model for cleaning and verification, obtaining cleaned data. The training module 403 receives the water quality data to be tested and the cleaned data, calculates the loss value, trains the water quality prediction model using the error feedback training method, and constructs the water quality parameter prediction model. The output module 404 inputs the water quality data to be tested into the water quality parameter prediction model, extracts water quality parameters, and generates a water chemistry parameter report.

[0100] like Figure 6 As shown, this application also provides a wastewater heavy metal source tracing and monitoring device using the above-mentioned monitoring system, including: a housing 500, one or more processors (CPUs) 510, a memory 520, a storage medium 530, and a power supply 540.

[0101] Specifically, storage medium 530 is used to store monitoring system 533, cleaning data 532 generated by monitoring system 533, and operating system 531. Storage medium 530 can be a storage device for short-term or long-term storage, such as one or more mass storage devices.

[0102] The processor 510 is connected to the storage medium 530 and is used to process the water chemistry parameter report data emitted by the storage medium 530 and convert it into a visual file. Simultaneously, the processor 510 can also execute operation instructions stored in the storage medium 530 to control the device and process data. These operation instructions can include one or more functions, each of which can contain a series of operation instructions related to the isotope water quality monitoring method, such as core operations like data acquisition, data cleaning, model training, error feedback, and report generation.

[0103] The memory 520 is connected to the storage medium 530 and is used to store historical data, water quality data to be tested, and water chemical parameter report data. The memory 520 can be short-term storage (such as RAM) or long-term storage (such as ROM) to ensure fast access and long-term preservation of data.

[0104] External interfaces are connected to the processor 510, memory 520, and storage medium 530 respectively for data transmission and interaction. These external interfaces include a network interface 550, a historical data input interface 560, and a water quality data input interface 570. The network interface 550 can be used to connect to an external network, supporting remote data transmission and remote operation of the monitoring system, such as retrieving historical water quality data from a remote server or remotely updating the operating system. The historical data input interface 560 is used to input external historical water quality data as input data for equipment data modeling and system optimization, while the water quality data input interface 570 is specifically used to receive the water quality data to be detected from the real-time water quality detection module. This data will be transmitted to the processor 510 and memory 520 for analysis and processing.

[0105] The power supply 540 is connected to the external interface, processor 510, memory 520, and storage medium 530, respectively, providing power support to ensure a stable power supply for the device during long-term monitoring and data processing tasks. The power supply 540 can be a built-in rechargeable power supply or an external power adapter to ensure the stability and reliability of the device during continuous monitoring operations.

[0106] To improve the scalability and compatibility of the device, the device may also include one or more operating systems 531, such as Windows Server, Mac OS X, Unix, Linux, or FreeBSD, so that the device can support remote operation and system updates across multiple platforms. Those skilled in the art will understand that the device architecture is not limited to the illustrated structure, and the number, type, and arrangement of its components can be added, deleted, reorganized, and replaced according to actual applications. For example, the functions of certain components can be combined, modules can be merged, or functional optimization can be achieved through hardware or software means.

[0107] Finally, this application also discloses a computer-readable medium, which is either a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, storing a computer-readable program. This computer-readable program can be executed by a processor to implement any of the above-described isotope-based wastewater heavy metal source tracing and monitoring methods. The storage method of the medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0108] Application example: Source tracing analysis of pollution in rivers downstream of industrial zones Fish deaths have recently occurred in a river downstream of an industrial zone (monitoring point ID: R2023-09), initially suspected to be caused by heavy metal pollution. To clarify the source of pollution and assess ecological risks, pollution monitoring and source tracing analysis were conducted on this water body. The main objectives of this monitoring included: determining the concentration of heavy metals and their isotopes in the water, identifying the pollution source (industrial emissions or natural geological background), and proposing remediation recommendations based on the assessment results.

[0109] During the implementation process, on-site sampling and rapid testing were carried out first. Sampling points were set up in the main river channel (R2023-09), tributary inlets (R2023-09-01), and near sewage outlets (R2023-09-02). On-site, pH, dissolved oxygen (DO), conductivity, and turbidity were monitored in real time using a portable water quality analyzer. At the same time, water samples were collected and sent to the laboratory for ICP-MS analysis to measure the concentrations of heavy metals and isotopes such as Pb-210, Hg-202, U-238, and Cd-114.

[0110] In the data analysis phase, based on the historical average water quality values ​​for the same period (September) over the past five years and the geological background data of the region (the natural background value of U-238 ranges from 0.1 to 0.3 Bq / L), the pollution source contribution rate model and the isotope ratio analysis model were used. The pollution source contribution rate model was used to calculate the contribution ratio of each pollution source to the heavy metal concentration; the isotope ratio analysis used the Pb-210 / U-238 ratio as the criterion, and when the ratio was greater than 3, it was considered to be dominated by anthropogenic pollution.

[0111] Subsequently, pollution source tracing and risk assessment were conducted based on the monitoring results. Through the calculation of the Comprehensive Pollution Index (CPI) and the determination of individual exceeding factors, the exceeding items and their main sources were identified. The results showed that Pb-210, Hg-202, and Cd-114 all exceeded the standards to varying degrees. Pb-210 exceeded the standard by 2.4 times, mainly from electroplating plant wastewater; Hg-202 exceeded the standard by 2.3 times, mainly from chemical plant emissions; and Cd-114 exceeded the standard by 1.8 times, mainly from the battery manufacturing process. U-238 concentration was within the safe range, originating from natural background levels. An example of the predicted water chemistry parameter monitoring report is shown in Table 3 below. Monitoring Report Number: WR-202309-RIVER Test date: September 15, 2023

[0112] The comprehensive analysis based on the above reports indicates that heavy metal pollution in this river section is primarily caused by anthropogenic emissions, mainly concentrated in industries such as electroplating, chemicals, and battery manufacturing. It is recommended that priority be given to treating the emission channels of electroplating plants and chemical enterprises, and that the water quality of tributaries and sewage outlets be continuously monitored to prevent further spread of heavy metal pollution and safeguard regional ecological security.

[0113] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for tracing and monitoring heavy metals in wastewater based on isotopes, characterized in that, Includes the following steps: S1. Collect historical water quality data of samples and preprocess them to generate historical data models and establish water quality indicators; The water quality data includes pollutant data such as heavy metals in the water and their isotopic composition and relative content parameters; Extract pollutant data from the historical data model, fit the data to it, and generate a water quality parameter model; The historical data models include a heavy metal isotope pollution source tracing model, an isotope ratio analysis model, and a heavy metal synergistic effect model. The pollutant data is a historical data fluctuation curve established by classifying and processing the heavy metals in water and their isotopic composition and relative content parameters. The data fitting method involves curve fitting the fluctuation curve to generate a water quality parameter model, specifically: （10）； In formula (10), Ci ( t () represents the concentration of the i-th heavy metal or isotope at time t. Ci 0 represents the initial background concentration. λi It is the decay constant of a radioactive isotope. Aik , tk It is the first k The input intensity and timing of secondary anthropogenic pollution events, Θ( t tk ) is a step function. αi , T It refers to the amplitude and cycle of seasonal fluctuations. βi It is a long-term linear trend. ( t ) is the random noise term. It is the first k In this secondary pollution incident, the radioactive isotopes were observed after a time tt. k The remaining percentage after; The water quality parameter model also includes a water quality screener, which is a water quality screening rule generated by cluster analysis based on the composition and relative content of heavy metals and their isotopes in the water quality parameter model. The water quality screener includes multiple water quality categories. The water quality category is defined as follows: pollution levels and corresponding pollutant data are determined by comparing the composition and relative content of heavy metals and their isotopes with background values. Specifically: （11） In formula (11), CPI is the comprehensive pollution index, and its calculation method is as follows: （12）； In formula (12), Si It is the first i Safety standards for heavy metals It is a single pollutant. wi It is the first i The toxicity weighting coefficients for each heavy metal, where m is the number of heavy metal species; When the overall pollution index or any single pollution factor exceeds the safety standard, a pollution risk is determined to exist; The heavy metal isotope pollution source tracing model is as follows: Concentration of heavy metals / isotopes = k1 Turbidity + k2 Conductivity +k3 Industrial emissions + …(7) In formula (7), k1, k2, and k3 are all pollution source contribution coefficients. This is the error term; The isotope ratio analysis model is as follows: Isotope ratio = (8); In formula (8), when the isotope ratio is greater than the threshold, it is determined to be human-caused pollution; when the isotope ratio is less than or equal to the threshold, it is determined to be caused by natural activities. The threshold values ​​are: lead 5~20 μg / L, mercury 0.5~5 μg / L, uranium 10~100 μg / L, and cadmium 1~10 μg / L; The heavy metal synergistic effect model uses: Toxicity index = α Hg-202+β Cd-114+γ Pb-210(9); In formula (9), α, β, γ are toxicity weighting coefficients; S2. Input the water quality data to be tested into the water quality parameter model, perform water quality analysis, and output processed water quality data, specifically including: S21. Preprocess the raw data of the water quality to be tested to obtain preprocessed data. The preprocessing includes: unifying the data format, classifying and normalizing, removing noise and filling missing values. S22. Compare the preprocessed data with historical data, find and mark outliers, and generate a report. S23. First, perform background value elimination and K-means clustering on the normal data, and then normalize the processed data as a whole to ensure that the dimensions of different data are consistent. S24. Input the normal data into the water quality parameter model, calculate the weights of different pollutant data using the entropy weight method, and then perform weighted average processing to determine the water quality category and output the processed water quality data. The entropy weight method is specifically: （1）； In formula (1), X' ij It is the standardized value, X ij It is the original value, min(X) j ) and max(X j ) are the minimum and maximum values ​​of the j-th indicator, respectively; （2）； k = 1 / ln(n)(3; In formulas (2) and (3), Ej is the entropy value of the j-th index, k is the entropy normalization constant, n is the number of samples, and ln(n) is the natural logarithm of the number of samples. （4）； In formula (4), Here, m represents the weights, and m represents the number of indicators. The weighted average processing specifically involves: （5）； In formula (5), It is the weighted average of each sample; S3. Based on the processed water quality data and the water quality data to be tested, calculate the prediction error of the water quality parameter model, generate the loss value, and use it as the model training value to construct a water quality prediction model. S4. Generate a report on the hydrochemical parameters of the water to be tested, and complete the prediction.

2. The monitoring method according to claim 1, characterized in that, In step S22, finding and marking outliers includes the following steps: S221. By cleaning, transforming and normalizing historical data, the effective range of heavy metal isotope pollution concentration data is determined. The specific method for normalization is as follows: （6）； In formula (6), X is the original data value, X min It is the minimum value of this feature, X max It is the maximum value of this feature, X norm It is the normalized value; S222. Compare the concentration of heavy metal isotopes in the preprocessed data with the effective range to determine whether it exceeds the effective range. S223. When the data exceeds the valid range, mark the data that exceeds the range, generate a report, and send the normal data to execute S23. If the valid range is not exceeded, execute S23 directly.

3. The monitoring method according to claim 2, characterized in that, In step S3, the water quality parameter prediction model is: （13）； In formula (13), Ci ( t +Δ t ) is the first i Heavy metals / isotopes in ( t +Δ t The predicted concentration at time ) αi It is the natural regression coefficient. wij It is the weight of the synergistic effect of metal j on metal i. βi These are multi-parameter coupling coefficients. γi It is the correction factor for natural decay of isotopes. δi It is a meteorological and environmental influencing factor. f (meteorology, t) is a meteorological function. It is a random error term; λi The decay constant of a radioactive isotope is calculated using the following formula: λ =ln(2) / T 1 / 2, of which, T Half of the life is 1 / 2.

4. The monitoring method according to claim 3, characterized in that, In step S3, model training includes the following steps: S31. According to the water quality screening rules, multiple sets of historical water quality data are screened sequentially to obtain the corresponding water pollutant data. S32. Based on the water pollutant data, perform data cleaning and data verification on the data to be tested to obtain cleaned data, and then write it into the water quality parameter model to obtain predicted values. Data validation methods include: parallel sample analysis, spike recovery experiments, and the use of quality control samples. S33. Calculate the loss value between the measured data and the predicted value through the mean absolute error, train the water quality prediction model based on the loss value, and construct the water quality parameter prediction model. When the MAE of the water quality parameter model is ≤0.1mg / L, the output of the water quality parameter model is directly used as the water quality parameter prediction model. When the MAE of the water quality parameter model is ≥0.1mg / L, the historical data is collected, cleaned, transformed and normalized again. The water quality parameter model is optimized by combining machine learning algorithms through multi-model parallel training and model ensemble until the MAE meets the condition. Then, its output is used as the water quality parameter prediction model.

5. The monitoring method according to claim 3, characterized in that, In step S4, generating the water chemistry parameter report includes the following steps: S41. Extract pollutant data from the sample data to be tested and match it with the water quality screening rules. S42. Determine the water quality category according to the water quality screening rules; S43. Output water quality category and generate a water chemical parameter report; The water chemistry parameter report includes: heavy metal and its isotopic composition and content, pH value, dissolved oxygen concentration, redox potential, temperature, pollution level, pollution distribution map, pollution migration map, and pollution source prediction results.

6. A monitoring system for implementing the wastewater heavy metal source tracing and monitoring method according to any one of claims 4 or 5, characterized in that, include: The acquisition module (401) is used to acquire heavy metal data of the water sample to be tested, and to upload the heavy metal data of the water sample to the analysis module. The analysis module (402) stores historical water quality data of the sample and its corresponding historical data model. Then, it extracts the historical data fluctuation curves of different heavy metals and their isotopes in the historical data model through curve fitting to obtain the water quality parameter model. Furthermore, the raw data of the water quality to be tested is preprocessed, outlier detection is performed, and normal data processing is performed. The preprocessing includes unifying the data format, classifying and normalizing, removing noise and filling missing values. The normal data processing includes background value elimination, K-means clustering, overall normalization, and weighted average processing based on entropy weight method, so as to determine the water quality category and output the processed water quality data. The historical data model includes a heavy metal isotope pollution source tracing model, an isotope ratio analysis model, and a heavy metal synergistic effect model. The water quality parameter model also includes a water quality screener, which is a water quality screening rule generated by cluster analysis based on the composition and relative content parameters of heavy metals and their isotopes in the water quality parameter model. This water quality category includes multiple water quality categories. The water quality category is a pollution level and its corresponding pollutant data that are classified by comparing the composition and relative content parameters of heavy metals and their isotopes with background values. The training module (403) is used to receive the test data and the predicted value, calculate the loss value, train the water quality prediction model according to the error feedback training method, and construct the water quality parameter prediction model. The output module (404) is used to input the data to be measured into the water quality parameter prediction model, extract the water quality parameters, and generate a water chemical parameter report; The water chemistry parameter report includes: heavy metal and its isotopic composition, content, pH value, dissolved oxygen concentration, and pollution level.

7. A wastewater heavy metal source tracing and monitoring device using the monitoring system as described in claim 6, characterized in that, include: Storage medium (530) for storing monitoring system (533), cleaning data (532) generated by monitoring system (533) and operating system (531); A processor (510), connected to the storage medium (530), is used to process water chemistry parameter report data issued by the storage medium (530) and convert it into a visual file; A memory (520), connected to the storage medium (530), is used to store historical data, test data, and water chemistry parameter report data; External interfaces are connected to the processor (510), memory (520) and storage medium (530) respectively, for transmitting data; The external interfaces include: a network interface (550), a historical data input interface (560), and a test data input interface (570). A power supply (540), which is connected to the external interface, processor (510), memory (520) and storage medium (530) respectively, is used to provide electrical energy.

Images