A tailing pond heavy metal pollution risk assessment method based on pair-wise ranking learning

Abstract

The present application relates to a kind of tailing pond heavy metal pollution risk assessment method based on pair-wise ranking learning, belong to the branch technical field of geophysics under ecological hydrology.The tailing pond data for tailing pond heavy metal pollution risk assessment, meteorological hydrology data, underlying surface condition data are collected;Based on "source-sink-receptor" framework, the evaluation index system including target layer, criterion layer, index layer is established, and tailing pond heavy metal pollution risk determination logic rule is established;The tailing pond heavy metal pollution risk determination logic rule established, using pair-wise ranking learning algorithm, constructs tailing pond heavy metal pollution risk ranking model, and calculates tailing pond heavy metal pollution risk comprehensive index.The present application solves the defects of subjective weight interference, data distribution influence and inaccurate evaluation results in the existing tailing pond heavy metal pollution risk assessment technology, and provides technical support for tailing pond environmental risk prevention and control and classification and grading supervision.

Description

Technical Field

[0001] This invention relates to a method for assessing the risk of heavy metal pollution in tailings ponds based on pairwise ranking learning, belonging to the field of eco-hydrology under geophysics. Background Technology

[0002] Tailings ponds are sites used to store tailings or other industrial waste generated during the mineral processing of metal or non-metal mines. my country has a large number of tailings ponds of various types, and the amount of mining solid waste is enormous and increasing. Tailings ponds pose a risk of heavy metal pollution to water bodies due to safety hazards such as leaks and dam failures. Risk assessment of heavy metal pollution from tailings ponds and identification of key source areas are crucial for ensuring water quality safety in river basins.

[0003] Existing risk assessments for heavy metal pollution in tailings ponds mainly include subjective assessment methods such as expert scoring and manual weighting, as well as objective assessment methods such as entropy weighting, TOPSIS (Technique for Order Preference by Similarity to Ideal Solution), MEREC (Method based on the Removal Effects of Criteria), and CRITIC (Criteria Importance Through Inter-criteria Correlation).

[0004] With the rapid development of artificial intelligence, machine learning algorithms are increasingly being applied to heavy metal pollution level analysis, such as the training method and device for a contaminated site risk level prediction model (Application / Patent No.: 202211186720.2). This method (Application / Patent No.: 202211186720.2) divides one or more contaminated sites into multiple plots; determines the risk level of each plot based on its pollutant content; constructs an output sample set based on the risk levels of multiple plots; constructs a feature index set for the contaminated site, including multiple feature indicators that indicate the site's environmental attributes, pollutant migration paths, and inherent characteristics; acquires multiple feature data corresponding to these feature indicators for each plot; constructs an input sample set based on the feature data of multiple plots; constructs a training set based on the input and output sample sets; and trains the contaminated site risk level prediction model based on the training set. Specifically, this method (application / patent number: 202211186720.2) calculates the pollution index based on the sampling data (i.e., pollutant content data) of each soil sampling point. For heavy metal pollutants, methods such as the single-factor index method, geoaccumulation index method, pollution load index method, Nemerow index method, and potential ecological risk index method can be selected to calculate the pollution index. The risk level of each plot is determined based on the pollution index. Machine learning algorithms such as CatBoost, XGBoost, and LightGBM are used to construct a risk level prediction model for contaminated sites. That is, machine learning algorithms are used to establish a mapping relationship between the characteristic index set of contaminated sites and the heavy metal pollution level, rather than directly using machine learning algorithms to evaluate the heavy metal pollution level when the heavy metal pollution level is unknown.

[0005] In summary, methods such as expert scoring and manual weighting rely on the professional knowledge and experience of evaluators, and their evaluation results may have a certain degree of subjective bias. Objective evaluation methods such as TOPSIS and MEREC determine the weight of indicators based on the characteristics of data distribution. Although this reduces the interference of human subjective factors, the evaluation results are affected by the dispersion of data distribution, and to some extent, it is difficult to accurately reflect the heavy metal pollution risk of tailings ponds. Machine learning prediction models have been gradually applied to the classification of heavy metal pollution levels, but they are all traditional classification or regression modeling, and there is still a lack of research on evaluating the risk of heavy metal pollution using ranking learning algorithms and risk judgment logic rules.

[0006] Therefore, there is an urgent need to develop a tailings dam heavy metal pollution risk assessment method that does not require subjective weighting, is not affected by data distribution, is based on risk judgment logic rules learning, and has objective and accurate evaluation results. Based on the risk assessment results, key source areas of heavy metal pollution should be identified, and tailings dams that need to be prioritized for treatment and control should be determined. Summary of the Invention

[0007] This invention addresses the shortcomings of existing tailings dam heavy metal pollution risk assessment technologies, such as subjective weighting interference, data distribution influence, and inaccurate assessment results, by providing a tailings dam heavy metal pollution risk assessment method based on pairwise ranking learning.

[0008] Provide technical support for environmental risk prevention and control and classified and graded supervision of tailings ponds.

[0009] The present invention adopts the following technical solution:

[0010] This invention employs the following technical solution: a tailings dam heavy metal pollution risk assessment method based on pairwise ranking learning. Basic data on tailings dams, meteorological and hydrological conditions, and underlying surface conditions within the watershed are collected. A tailings dam heavy metal pollution risk assessment index system is established based on a "source-sink-receptor" framework. Logical rules for determining tailings dam heavy metal pollution risk are established, comprehensively considering source hazard, migration and diffusion potential, and receptor vulnerability. A tailings dam heavy metal pollution risk ranking model based on a pairwise ranking learning algorithm is constructed. A comprehensive heavy metal pollution risk index for tailings dams is calculated. The risk level is determined based on the comprehensive heavy metal pollution risk index, and key source areas of heavy metal pollution in tailings dams are identified. Figure 1 This is a flowchart of the tailings dam heavy metal pollution risk assessment method based on pairwise ranking learning, as described in this invention. The specific steps are as follows:

[0011] (1) Collection of basic data for risk assessment of heavy metal pollution in tailings ponds.

[0012] Collect basic data on tailings ponds, meteorological and hydrological conditions, and underlying surface conditions within the watershed to establish a basic dataset for heavy metal pollution risk assessment of tailings ponds. Specifically:

[0013] ① Collect basic information about the tailings dam, including the name of the tailings dam, the grade of the tailings dam, the type of mineral in the tailings dam, and the production status of the tailings dam.

[0014] ② Collect meteorological and hydrological data, including annual precipitation, distance from river confluence, and the classification of the rivers into which they flow.

[0015] ③ Collect data on underlying surface conditions, including slope and vegetation cover.

[0016] (2) Construction of a risk assessment index system for heavy metal pollution in tailings ponds.

[0017] Based on the "source-sink-receptor" framework, a risk assessment index system for heavy metal pollution in tailings ponds is established, encompassing the target layer, criterion layer, and indicator layer. Specifically:

[0018] ① Target layer: The comprehensive risk index of heavy metal pollution in tailings ponds is used as the target layer.

[0019] ② Criterion layer: The criterion layer consists of source hazard factor, migration and diffusion factor, and receptor vulnerability factor.

[0020] ③ Indicator layer: Source hazard factors mainly consider the tailings dam grade, tailings dam mineral type, and tailings dam production status; migration and diffusion factors mainly consider annual precipitation, slope, and vegetation coverage; receptor vulnerability factors mainly consider distance from river confluence and the grade of the inflowing river.

[0021] (3) Construction of logic rules for determining the risk of heavy metal pollution in tailings ponds.

[0022] Taking into account source hazard, migration and diffusion, and receptor vulnerability, a logical rule for determining the heavy metal pollution risk of tailings ponds is established. Specifically:

[0023] ① Logical rules for determining the heavy metal pollution risk of tailings ponds considering source hazards:

[0024] [a] Tailings ponds, etc.

[0025] The higher the tailings dam height, the larger the capacity, and the higher the tailings dam classification, the higher the risk of heavy metal pollution. The logical rule for determining the heavy metal pollution risk of tailings dams based on their classification is: Class 1 > Class 2 > Class 3 > Class 4 > Class 5. The tailings dam classification is denoted as... .

[0026] Based on the logic rules for determining the heavy metal pollution risk of tailings ponds according to their grade, tailings ponds of grades five, four, three, two, and one are mapped at equal intervals to... ,For example The risk score is based on heavy metal pollution from tailings ponds and other sources. ,Right now , The rest are different. Values ​​between 0 and The values ​​are taken at equal intervals. It quantitatively reflects the logical rules for determining the heavy metal pollution risk of tailings ponds and other types of tailings ponds; unlike subjective weighting or weighting based on data distribution characteristics, it provides a more comprehensive understanding. The numerical values ​​are only used to reflect the relative order of heavy metal pollution risk of tailings ponds of different grades, rather than focusing on the numerical values ​​themselves.

[0027] [b] Tailings Dam Mineral Types

[0028] Based on the biotoxicity and environmental hazards of heavy metals, a logical rule for determining the heavy metal pollution risk of tailings ponds based on the type of mineral in the tailings pond is established: {antimony ore} > {lead ore, lead-zinc ore, zinc ore} > {copper ore, silver ore} > {molybdenum ore, iron ore} > {gold ore, flotation gold ore, pyrite} > {native sulfur, graphite ore}. The type of mineral in the tailings pond is denoted as... .

[0029] Based on the logic rules for determining the heavy metal pollution risk of tailings ponds according to the mineral types in the tailings pond, the main mineral types {natural sulfur, graphite ore}, {gold ore, flotation gold ore, pyrite}, {molybdenum ore, iron ore}, {copper ore, silver ore}, {lead ore, lead-zinc ore, zinc ore}, and {antimony ore} are mapped at equal intervals to... The heavy metal pollution risk score is obtained based on the type of mineral in the tailings dam. ,Right now , Other mineral types Values ​​between 0 and The values ​​are taken at equal intervals. This quantitatively reflects the logical rules for determining the heavy metal pollution risk of tailings ponds based on the mineral types within the tailings pond; unlike methods such as subjective weighting or weighting based on data distribution characteristics. The numerical values ​​are only used to reflect the relative order of heavy metal pollution risk of tailings ponds of different mineral types, rather than focusing on the numerical values ​​themselves.

[0030] [c] Tailings Dam Production Status

[0031] Considering the environmental monitoring and management levels of tailings ponds under different production conditions, a logical rule for determining the heavy metal pollution risk of tailings ponds based on their production status is established: Discontinued > Closed > In Use. The production status of the tailings pond is denoted as... .

[0032] Based on the logic rules for determining the heavy metal pollution risk of tailings ponds according to their production status, the intervals between the states of use, closure, and shutdown will be mapped to... A heavy metal pollution risk score is obtained based on the production status of the tailings dam. ,Right now , , . This quantitatively reflects the logical rules for determining the heavy metal pollution risk of tailings ponds based on their production status; unlike methods such as subjective weighting or weighting based on data distribution characteristics. The numerical values ​​are only used to reflect the relative order of heavy metal pollution risk of tailings ponds under different production conditions, rather than focusing on the numerical values ​​themselves.

[0033] ② Logic rules for determining the risk of heavy metal pollution in tailings ponds considering migration and diffusion:

[0034] [a] Annual precipitation

[0035] The greater the annual rainfall in the area where the tailings dam is located, the stronger the potential erosion and the higher the risk of heavy metal leakage and pollution. Therefore, the logical rule for determining the heavy metal pollution risk of tailings dams based on annual rainfall is: the greater the annual rainfall, the greater the risk of heavy metal pollution from the tailings dam. Annual rainfall is denoted as... .

[0036] Following the logic rules for determining the heavy metal pollution risk of tailings ponds based on annual precipitation, the annual precipitation is normalized using the min-max forward normalization method before being multiplied by [the threshold value]. The annual precipitation will be positively mapped to its numerical value. A heavy metal pollution risk score is obtained based on annual precipitation. . This quantitatively reflects the logical rules for determining the heavy metal pollution risk of tailings ponds based on annual precipitation; unlike methods such as subjective weighting or weighting based on data distribution characteristics. The numerical values ​​are only used to reflect the relative order of heavy metal pollution risk of tailings ponds in areas with different annual precipitation, rather than focusing on the numerical values ​​themselves.

[0037] [b] Slope

[0038] The steeper the terrain and the greater the slope of the tailings dam area, the stronger the precipitation erosion and the higher the risk of heavy metal migration. Therefore, the logical rule for determining the heavy metal pollution risk of tailings dams based on slope is: the greater the slope, the greater the risk of heavy metal pollution from the tailings dam. The slope is denoted as... .

[0039] Following the logic rules for determining the heavy metal pollution risk of tailings ponds based on slope, the slope is normalized using the min-max forward normalization method before being multiplied by [the appropriate factor]. That is, the slope is positively mapped to its numerical value. A heavy metal pollution risk score based on slope is obtained. . This quantitatively reflects the logical rules for determining the heavy metal pollution risk of tailings ponds based on slope; unlike methods such as subjective weighting or weighting based on data distribution characteristics, The numerical values ​​are only used to reflect the relative order of heavy metal pollution risk of tailings ponds in different slope areas, rather than focusing on the numerical values ​​themselves.

[0040] [c] Vegetation coverage

[0041] The lower the vegetation cover in the area where a tailings dam is located, the weaker the ability of the underlying surface to stabilize the soil against erosion and intercept pollutants, and the higher the risk of heavy metal migration. Therefore, the logical rule for determining the heavy metal pollution risk of tailings dams based on vegetation cover is: the lower the vegetation cover, the greater the risk of heavy metal pollution from the tailings dam. Vegetation cover is denoted as... .

[0042] Following the logic rules for determining the heavy metal pollution risk of tailings ponds based on vegetation coverage, the vegetation coverage is normalized using the max-min inverse normalization method before being multiplied by [the factor]. That is, to reverse map vegetation cover to its numerical value. A heavy metal pollution risk score is obtained based on vegetation coverage. . This quantitatively reflects the logical rules for determining the heavy metal pollution risk of tailings ponds based on vegetation cover; unlike methods such as subjective weighting or weighting based on data distribution characteristics. The numerical values ​​are only used to reflect the relative order of heavy metal pollution risk of tailings ponds in areas with different vegetation cover, rather than focusing on the numerical values ​​themselves.

[0043] ③ Logic rules for determining the heavy metal pollution risk of tailings ponds considering receptor vulnerability:

[0044] [a] Distance from the confluence of the river channel

[0045] The smaller the distance between a tailings dam and its river confluence, the easier it is for heavy metals to enter the water body, and the greater the risk of pollution. Therefore, the logical rule for determining the heavy metal pollution risk of tailings dams based on their distance from river confluence is: the smaller the distance, the greater the risk of heavy metal pollution. The distance from river confluence is denoted as... .

[0046] Based on the logic rule for determining the heavy metal pollution risk of tailings ponds according to the distance from the river confluence, the max-min inverse normalization method is used to normalize the distance from the river confluence before multiplying it by [the appropriate factor]. This means that the distance from the river confluence will be inversely mapped to its numerical value. A heavy metal pollution risk score is obtained based on the distance from the river confluence. . This quantitatively reflects the logical rules for determining the heavy metal pollution risk of tailings ponds based on their distance from river confluence; unlike methods such as subjective weighting or weighting based on data distribution characteristics. The numerical values ​​are only used to reflect the relative order of heavy metal pollution risk of tailings ponds at different distances from the river confluence, rather than focusing on the numerical values ​​themselves.

[0047] [b] Classification of rivers flowing into the river

[0048] The higher the classification of the river flowing downstream of a tailings dam, the more important the river and the greater the risk of pollution. For example, if the river flowing downstream of a tailings dam is a Class I river, the risk of heavy metal pollution is relatively high; if the river flowing downstream is a Class III river, the risk of heavy metal pollution is relatively low. The logical rule for determining the heavy metal pollution risk of tailings dams based on the classification of the flowing river is: the higher the classification of the flowing river, the greater the risk of heavy metal pollution from the tailings dam. The classification of the flowing river is denoted as... .

[0049] Following the logic rules for determining the heavy metal pollution risk of tailings ponds based on the class of the inflowing river, the max-min inverse normalization method is used to normalize the class of the inflowing river before multiplying it by a factor. The river level that will soon flow into the river is inversely mapped to its numerical value. The heavy metal pollution risk score is based on the level of the river into which it flows. . This quantitatively reflects the logical rules for determining the heavy metal pollution risk of tailings ponds based on the level of the rivers into which they flow; unlike methods such as subjective weighting or weighting based on data distribution characteristics. The numerical values ​​are used only to reflect the relative order of heavy metal pollution risk of tailings ponds of different grades flowing into rivers, rather than focusing on the numerical values ​​themselves.

[0050] (4) Analysis of the comprehensive index of heavy metal pollution risk in tailings ponds.

[0051] A pairwise ranking learning algorithm is used to construct a ranking model of heavy metal pollution risk in tailings ponds, and a comprehensive index of heavy metal pollution risk in tailings ponds is calculated. Specifically:

[0052] ① Based on the logic rules for determining the heavy metal pollution risk of tailings ponds in step (3), the indicators are... The values ​​are added together to obtain the heavy metal pollution risk ranking labels for each tailings dam. Heavy metal pollution risk ranking labels for each tailings pond. The calculation formula is:

[0053]

[0054] Will The pairwise ranking learning algorithm serves as the learning objective. It's important to note that, unlike general regression models, the core of the pairwise ranking learning algorithm is to learn the order of heavy metal pollution risks among different tailings ponds, rather than absolute relationships. Numerical value.

[0055] ② Construct a tailings dam heavy metal pollution risk ranking model based on the pairwise ranking learning algorithm. The tailings dam heavy metal pollution risk assessment indicators from step (2) are used as model inputs. Specifically, the model input variables are: tailings dam grade, tailings dam mineral type, tailings dam production status, annual precipitation, slope, vegetation cover, distance from river confluence, and the grade of the inflowing river. It should be noted that the input to the tailings dam heavy metal pollution risk ranking model based on the pairwise ranking learning algorithm is the original value of each indicator, not the ranked values. Value. The risk ranking model for heavy metal pollution in tailings ponds based on the pairwise ranking learning algorithm can be expressed as:

[0056]

[0057] In the formula: Learning for pairwise sorting algorithms The output result after sorting. The relative magnitude of the values ​​reflects the relative magnitude of the heavy metal pollution risk from tailings ponds; This refers to pairwise ranking learning algorithms. Available gradient boosting decision tree-based algorithm tools mainly include XGBoost, LightGBM, and CatBoost. The key parameters commonly used by these algorithm tools mainly include the number of iterations, tree depth, learning rate, and early stopping strategy.

[0058] ③ Using the basic data for heavy metal pollution risk assessment of tailings ponds from step (1) and the logical rules for determining heavy metal pollution risk of tailings ponds from step (3), train the heavy metal pollution risk ranking model of tailings ponds. Based on the pairwise ranking learning algorithm, the ranking model automatically learns the relative magnitude change law of heavy metal pollution risk of each tailings pond and outputs the heavy metal pollution risk score of each tailings pond. The min-max forward normalization method was used to score the heavy metal pollution risk of tailings ponds. Mapping to the interval [0, 1] yields the comprehensive index of heavy metal pollution risk in tailings ponds, denoted as... , .

[0059] (5) Risk level of heavy metal pollution in tailings ponds and delineation of key source areas.

[0060] The risk level of heavy metal pollution in tailings ponds is determined based on a comprehensive risk index, and key source areas of heavy metal pollution in tailings ponds are identified. Specifically:

[0061] ① The heavy metal pollution risk of tailings ponds is classified into five levels: low, relatively low, medium, relatively high, and high, with low risk being the lowest. Lower risk Medium risk Higher risk High risk .

[0062] ② Based on the analysis results of the comprehensive index of heavy metal pollution risk of tailings ponds in step (4), determine the risk level of heavy metal pollution of each tailings pond, and classify the tailings ponds listed as high risk as key source areas of heavy metal pollution, and carry out priority treatment and control.

[0063] Beneficial effects

[0064] This invention is a method for assessing the heavy metal pollution risk of tailings ponds based on pairwise ranking learning. It is mainly applicable to the batch ranking of heavy metal pollution risks and identification of key source areas for multiple tailings ponds at the watershed scale. It constructs a heavy metal pollution risk assessment index system for tailings ponds that covers source hazard, migration and diffusion, and receptor vulnerability, establishes logical rules for determining the heavy metal pollution risk of tailings ponds, and uses a pairwise ranking learning algorithm to automatically learn the relative magnitude variation law of heavy metal pollution risk of tailings ponds and output a comprehensive index of heavy metal pollution risk of tailings ponds to determine the heavy metal pollution risk level and key source areas.

[0065] This invention presents a tailings dam heavy metal pollution risk assessment method based on pairwise learning to rank. Within the "source-sink-receptor" index framework, it establishes logical rules for determining the heavy metal pollution risk of tailings dams. It employs a pairwise learning to rank algorithm to analyze the heavy metal pollution risk index of tailings dams, classifies risk levels based on this index, and identifies key heavy metal pollution source areas. This addresses the shortcomings of existing tailings dam heavy metal pollution risk assessment technologies, such as subjective weighting interference, data distribution influence, and inaccurate assessment results. It enriches and improves the methodological system for heavy metal pollution risk assessment of tailings dams, enhances the utilization efficiency of data on tailings dams, meteorological and hydrological conditions, and underlying surface conditions, and provides technical support for environmental risk prevention and control and classified and graded supervision of tailings dams. Attached Figure Description

[0066] Figure 1 This is a flowchart of the tailings dam heavy metal pollution risk assessment method based on pairwise ranking learning of the present invention.

[0067] Figure 2 This is the basic information of the tailings dam in the Laoguan River Basin, which is an example of the present invention.

[0068] Figure 3 The annual precipitation is the area where the tailings dam of the Laoguan River Basin, an example of this invention, is located.

[0069] Figure 4 This refers to the distance between the tailings dam in the Laoguan River basin and the confluence of the river channel, as exemplified in this invention.

[0070] Figure 5 This is an example of the river level into which the tailings dam in the Laoguan River basin flows.

[0071] Figure 6 This refers to the slope of the area where the tailings dam in the Laoguan River Basin, an example of this invention, is located.

[0072] Figure 7 This refers to the vegetation coverage of the area where the tailings dam in the Laoguan River Basin, an example of this invention, is located.

[0073] Figure 8This invention relates to an index system for assessing the risk of heavy metal pollution from tailings ponds in the Laoguan River basin.

[0074] Figure 9 This is the comprehensive risk index of heavy metal pollution from tailings ponds in the Laoguan River Basin, as exemplified by this invention. Detailed Implementation

[0075] To make the objectives and technical solutions of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0076] Figure 1 This is a flowchart of the tailings dam heavy metal pollution risk assessment method based on pairwise ranking learning, as described in this invention. The application research area of ​​this invention is the Laoguan River Basin, a typical small watershed in the water source area of ​​the South-to-North Water Diversion Project. The Laoguan River Basin is located in southwestern Henan Province, at the junction of Hubei, Shaanxi, and Henan provinces, with geographical coordinates between 110°17′–111°50′E and 32°55′–34°01′N. As a typical mountain river, the Laoguan River originates in the area of ​​Nannihu Village, Lengshui Town, Luanchuan County. Its main stream flows from northwest to southeast through Lushi County, Xixia County, Neixiang County, and Xichuan County, finally flowing into the Danjiangkou Reservoir at Shanghe Village, Xichuan County. It is a first-order tributary of the Danjiang River, a tributary of the Han River in the Yangtze River Basin. The Laoguan River is approximately 254 km long, with a total drainage area of ​​4219 km². 2 It accounts for approximately 2.5% of the total area of ​​Henan Province, of which 3266 km² 2 Located within Nanyang City, the river basin has a well-developed water system, with major tributaries including the Ding River, Shewei River, Yanzhen River, Guzhuang River, Wangou River, and Jiuligou River. The drainage area of ​​each major tributary is over 50 km². 2 The above branches flow into the main stream of the Laoguan River in a tree-like pattern.

[0077] Step 1: Collection of basic data for heavy metal pollution risk assessment of tailings ponds in the Laoguan River Basin

[0078] Basic data on tailings ponds, meteorological and hydrological conditions, and underlying surface conditions within the Laoguan River basin were collected to establish a basic dataset for heavy metal pollution risk assessment of tailings ponds in the Laoguan River basin. Specifically:

[0079] Collect basic information on tailings ponds in the Laoguan River Basin, including the name of the tailings pond, its grade, the main minerals in the tailings pond, and its production status. Figure 2 This is the basic information of the tailings dam in the Laoguan River Basin, which is an example of the present invention.

[0080] Collect meteorological and hydrological data for the area where the tailings dam in the Laoguan River basin is located, including annual precipitation, distance from the river confluence, and the classification of the rivers into which it flows. Figure 3 The annual precipitation in the area where the tailings dam of the Laoguan River Basin, an example of this invention, is located. Figure 4 This is an example of the distance between the tailings dam in the Laoguan River basin and the river confluence. Figure 5 This is an example of the river level into which the tailings dam in the Laoguan River basin flows.

[0081] Collect data on underlying surface conditions in the area where the tailings dams are located in the Laoguan River Basin, including slope and vegetation cover. Figure 6 The slope of the area where the tailings dam in the Laoguan River basin, an example of this invention, is located. Figure 7 This refers to the vegetation coverage of the area where the tailings dam in the Laoguan River Basin, an example of this invention, is located.

[0082] Step 2: Construction of a Risk Assessment Index System for Heavy Metal Pollution from Tailings Ponds in the Laoguan River Basin

[0083] Based on the "source-sink-receptor" framework, a risk assessment index system for heavy metal pollution in tailings ponds in the Laoguan River Basin was established, covering the target layer, criterion layer, and indicator layer. Figure 8 This invention provides an example of a heavy metal pollution risk assessment index system for tailings ponds in the Laoguan River basin. Specifically:

[0084] Target layer: The comprehensive risk index of heavy metal pollution in tailings ponds in the Laoguan River Basin is used as the target layer.

[0085] Criterion layer: The criteria layer consists of source hazard factor, migration and diffusion factor, and receptor vulnerability factor.

[0086] Indicator layer: Source hazard factors mainly consider tailings dam grade, tailings dam mineral type, tailings dam production status, etc.; migration and diffusion factors mainly consider annual precipitation, slope, vegetation cover, etc.; receptor vulnerability factors mainly consider distance from river confluence, class of inflowing river, etc.

[0087] Step 3: Constructing logical rules for assessing heavy metal pollution risks in tailings ponds in the Laoguan River Basin.

[0088] Taking into account source hazard, migration and diffusion, and receptor vulnerability, a logical rule for determining the heavy metal pollution risk of tailings ponds in the Laoguan River basin is established. Specifically:

[0089] Tailings ponds, etc., are not recorded as The logical rule for determining the heavy metal pollution risk of tailings ponds based on their grade is: Grade 1 > Grade 2 > Grade 3 > Grade 4 > Grade 5. The grades of tailings ponds (Grade 5, Grade 4, Grade 3, Grade 2, and Grade 1) are mapped at equal intervals to... In this embodiment The risk score is based on heavy metal pollution from tailings ponds and other sources. .

[0090] The mineral type of the tailings dam is denoted as The logical rule for determining the heavy metal pollution risk of tailings ponds based on the mineral types in the tailings ponds is as follows: {antimony ore} > {lead ore, lead-zinc ore, zinc ore} > {copper ore, silver ore} > {molybdenum ore, iron ore} > {gold ore, flotation gold ore, pyrite} > {native sulfur, graphite ore}. The major mineral types {native sulfur, graphite ore}, {gold ore, flotation gold ore, pyrite}, {molybdenum ore, iron ore}, {copper ore, silver ore}, {lead ore, lead-zinc ore, zinc ore}, and {antimony ore} are mapped at equal intervals to... The heavy metal pollution risk score is obtained based on the type of mineral in the tailings dam. .

[0091] The production status of tailings ponds is recorded as follows: The logical rule for determining the heavy metal pollution risk of tailings ponds based on their production status is: Discontinued > Closed > In Use. The states of in use, closed, and discontinued will be mapped at equal intervals to... A heavy metal pollution risk score is obtained based on the production status of the tailings dam. .

[0092] Logical rules for determining the risk of heavy metal pollution in tailings ponds that take into account migration and diffusion.

[0093] Annual precipitation is denoted as The logical rule for determining the heavy metal pollution risk of tailings ponds based on annual precipitation is: the greater the annual precipitation, the greater the risk of heavy metal pollution in the tailings pond. The annual precipitation is normalized using the min-max forward normalization method and then multiplied by 5, that is, the annual precipitation is positively mapped to its numerical value. A heavy metal pollution risk score is obtained based on annual precipitation. .

[0094] Slope is denoted as The logical rule for determining the heavy metal pollution risk of tailings ponds based on slope is: the steeper the slope, the greater the risk of heavy metal pollution in the tailings pond. The slope is normalized using the min-max forward normalization method and then multiplied by 5, that is, the slope is positively mapped to its numerical value. A heavy metal pollution risk score based on slope is obtained. .

[0095] Vegetation coverage is denoted as The logical rule for determining the heavy metal pollution risk of tailings ponds based on vegetation cover is: the lower the vegetation cover, the greater the risk of heavy metal pollution in the tailings pond. The vegetation cover is normalized using the max-min inverse normalization method and then multiplied by 5, that is, the vegetation cover is inversely mapped to its numerical value. A heavy metal pollution risk score is obtained based on vegetation coverage. .

[0096] Logical rules for determining the risk of heavy metal pollution in tailings ponds considering receptor vulnerability.

[0097] The distance from the confluence of the river is denoted as The logical rule for determining the heavy metal pollution risk of tailings ponds based on their distance from river confluence is: the smaller the distance from river confluence, the greater the risk of heavy metal pollution from the tailings pond. The max-min inverse normalization method is used to normalize the distance from river confluence and then multiply it by 5, that is, the distance from river confluence is inversely mapped to its numerical value. A heavy metal pollution risk score is obtained based on the distance from the river confluence. .

[0098] The class of the rivers into which they flow is denoted as The logical rule for determining the heavy metal pollution risk of tailings ponds based on the inflow river level is: the higher the inflow river level, the greater the heavy metal pollution risk of the tailings pond. The max-min inverse normalization method is used to normalize the inflow river level and then multiply it by 5, that is, the inflow river level is inversely mapped to the inflow river level according to its numerical value. The heavy metal pollution risk score is based on the level of the river into which it flows. .

[0099] Step 4: Comprehensive analysis of the heavy metal pollution risk index of tailings ponds in the Laoguan River Basin.

[0100] A pairwise ranking learning algorithm was used to construct a risk ranking model for heavy metal pollution in tailings ponds of the Laoguan River Basin, and the comprehensive risk index of heavy metal pollution in tailings ponds of the Laoguan River Basin was calculated. Specifically:

[0101] Based on the logical rules for determining the heavy metal pollution risk of tailings ponds in the Laoguan River basin in the third step, the various indicators will be... The values ​​are added together to obtain the heavy metal pollution risk ranking labels for each tailings dam. ,Will The target of the pairwise ranking learning algorithm is the heavy metal pollution risk ranking label of each tailings dam. The calculation formula is:

[0102]

[0103] A risk ranking model for heavy metal pollution in tailings ponds based on the pairwise ranking learning algorithm (CatBoostRanker) is constructed, using the risk assessment indicators for heavy metal pollution in tailings ponds in the Laoguan River basin from the second step as model input. The risk ranking model for heavy metal pollution in tailings ponds based on the pairwise ranking learning algorithm (CatBoostRanker) can be expressed as follows:

[0104]

[0105] In the formula: Learning for pairwise sorting algorithms The output result after sorting. The relative magnitude of the values ​​reflects the relative magnitude of the heavy metal pollution risk from tailings ponds; This represents a pairwise ranking learning algorithm, using CatBoostRanker as the algorithm tool, with iterations=10000, tree depth=4, learning rate=0.03, and early stopping policy=50.

[0106] Using the basic data from the first step of the heavy metal pollution risk assessment of tailings ponds in the Laoguan River Basin and the logical rules for determining heavy metal pollution risk in the third step, a heavy metal pollution risk ranking model for tailings ponds in the Laoguan River Basin was trained. Based on the pairwise ranking learning algorithm, the ranking model automatically learned the relative magnitude variation pattern of heavy metal pollution risk for each tailings pond and output the heavy metal pollution risk score for each tailings pond in the Laoguan River Basin. The min-max forward normalization method was used to score the heavy metal pollution risk of tailings ponds in the Laoguan River basin. Mapping to the interval [0, 1] yields the comprehensive index of heavy metal pollution risk in the tailings ponds of the Laoguan River basin, denoted as... , . Figure 9 This is the comprehensive risk index of heavy metal pollution from tailings ponds in the Laoguan River Basin, as exemplified by this invention.

[0107] Step 5: Risk level of heavy metal pollution in tailings ponds in the Laoguan River Basin and delineation of key source areas.

[0108] Risk levels were determined based on the comprehensive risk index of heavy metal pollution from tailings ponds in the Laoguan River basin, and key source areas of heavy metal pollution from tailings ponds in the Laoguan River basin were identified. Specifically:

[0109] The risk of heavy metal pollution in tailings ponds is classified into five levels: low, relatively low, medium, relatively high, and high. Low risk... Lower risk Medium risk Higher risk High risk .

[0110] Based on the analysis results of the comprehensive risk index of heavy metal pollution in tailings ponds in the fourth step of the Laoguan River Basin, the risk level of heavy metal pollution in each tailings pond was determined. Tailings ponds W, A, Q, K, E, and U were classified as high-risk tailings ponds and designated as key source areas of heavy metal pollution, and priority was given to their treatment and control.

[0111] The application of the tailings dam heavy metal pollution risk assessment method based on pairwise ranking learning in this invention is as follows: A risk assessment index system for heavy metal pollution in tailings dams, encompassing source hazard, migration and diffusion, and receptor vulnerability, is constructed. Logical rules for determining the heavy metal pollution risk of tailings dams are established. A pairwise ranking learning algorithm is used to automatically learn the relative magnitude variation patterns of heavy metal pollution risk in tailings dams and output a comprehensive index of heavy metal pollution risk, thereby determining the heavy metal pollution risk level and key source areas. This tailings dam heavy metal pollution risk assessment method based on pairwise ranking learning is mainly applicable to batch heavy metal pollution risk ranking and key source area identification for multiple tailings dams at the watershed scale. It addresses the shortcomings of existing tailings dam heavy metal pollution risk assessment technologies, such as subjective weight interference and inaccurate evaluation results. It enriches and improves the methodological system for heavy metal pollution risk assessment in tailings dams, enhances the utilization efficiency of data on tailings dams, meteorological and hydrological conditions, and underlying surface conditions, and provides technical support for environmental risk prevention and control and classified and graded supervision of tailings dams.

[0112] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention 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 the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for assessing the risk of heavy metal pollution in tailings ponds based on pairwise ranking learning, characterized in that, The evaluation method is as follows: S1. Collect tailings dam data, meteorological and hydrological data, and underlying surface condition data for the risk assessment of heavy metal pollution in tailings dams. S2. Based on the "source-sink-receptor" framework, establish an evaluation index system that includes a target layer, a criterion layer, and an indicator layer; S3. Based on the evaluation index system in step S2, establish logical rules for determining the heavy metal pollution risk of tailings ponds by combining the tailings dam data, meteorological and hydrological data, and underlying surface condition data from step S1. S4. Based on the logical rules for determining the heavy metal pollution risk of tailings ponds established in step S3, a pairwise sorting learning algorithm is used to construct a heavy metal pollution risk ranking model for tailings ponds and calculate the comprehensive index of heavy metal pollution risk of tailings ponds.

2. The method for assessing the heavy metal pollution risk of tailings ponds based on pairwise ranking learning according to claim 1, characterized in that, The tailings dam data in step S1 includes: tailings dam name, tailings dam grade, tailings dam mineral type, and tailings dam production status. The meteorological and hydrological data include: annual precipitation, distance from river confluence, and the class of the rivers into which they flow. The underlying surface condition data includes: slope and vegetation coverage.

3. The method for assessing the heavy metal pollution risk of tailings dams based on pairwise ranking learning according to claim 1, characterized in that, In step S2, the comprehensive risk index of heavy metal pollution in tailings ponds is used as the target layer. The source hazard factor, migration and diffusion factor, and receptor vulnerability factor are used as the criteria layer; The indicator layer includes the tailings dam classification, tailings dam mineral type, and tailings dam production status related to the source hazard factors. Migration and diffusion factors include annual precipitation, slope, and vegetation cover; The receptor vulnerability factors involve the distance from the river confluence and the class of the river into which it flows.

4. The method for assessing the heavy metal pollution risk of tailings ponds based on pairwise ranking learning according to claim 3, characterized in that, The logical rules for determining the heavy metal pollution risk of tailings ponds based on source hazard factors are as follows: The logical rules for determining the heavy metal pollution risk of tailings ponds and other tailings ponds are as follows: (1) Tailings dams are classified according to their height, capacity, and grade: Grade 1 > Grade 2 > Grade 3 > Grade 4 > Grade 5; the grade of tailings dam is recorded as follows: ; According to the logic rules for assessing the heavy metal pollution risk of tailings ponds and other types of tailings ponds, tailings ponds and other types of tailings ponds are classified as follows: Equal spacing mapping to Obtain risk scores for heavy metal pollution from tailings ponds and other sources. ; (2) The logical rule for determining the heavy metal pollution risk of tailings ponds by mineral type is as follows: antimony ore > lead ore, lead-zinc ore, zinc ore > copper ore, silver ore > molybdenum ore, iron ore > gold ore, flotation gold ore, pyrite > native sulfur, graphite ore; the mineral type of tailings pond is denoted as ; According to the logical rules for determining the heavy metal pollution risk of tailings ponds based on the mineral type of the tailings pond, the mineral types of the tailings pond are classified. Equal spacing mapping to The heavy metal pollution risk score of the mineral type in the tailings dam was obtained. ; (3) The logical rule for determining the heavy metal pollution risk of tailings ponds based on their production status is: Closed > Shutdown > In Use; the production status of tailings ponds is recorded as follows: ; According to the logical rules for determining the heavy metal pollution risk of tailings ponds based on their production status, the production status of the tailings pond will be... Equal spacing mapping to The heavy metal pollution risk score of the tailings dam's production status is obtained. ; The logical rules for determining the risk of heavy metal pollution in tailings ponds based on migration and diffusion factors are as follows: (1) The annual precipitation in the area where the tailings dam is located is used as the logical rule for determining the heavy metal pollution risk of the tailings dam. The rule is: the greater the annual precipitation, the greater the risk of heavy metal pollution of the tailings dam. The annual precipitation is denoted as: ; Annual precipitation Map it positively according to its numerical value. A heavy metal pollution risk score is obtained based on annual precipitation. ; (2) The slope of the area where the tailings dam is located is used as the logical rule for determining the heavy metal pollution risk of the tailings dam. The rule is: the greater the slope, the greater the risk of heavy metal pollution of the tailings dam. The slope is denoted as... ; slope Map it positively according to its numerical value. A heavy metal pollution risk score based on slope is obtained. ; (3) The vegetation cover of the area where the tailings dam is located is used as the logical rule for determining the heavy metal pollution risk of the tailings dam. The rule is that the lower the vegetation cover, the greater the risk of heavy metal pollution of the tailings dam. The vegetation cover is denoted as... ; vegetation coverage Reverse mapping according to its numerical value to A heavy metal pollution risk score based on vegetation coverage is obtained. ; The logical rules for determining the risk of heavy metal pollution in tailings ponds based on receptor vulnerability factors are as follows: (1) The distance between the tailings dam and the river confluence is used as the logical rule for determining the heavy metal pollution risk of the tailings dam. The rule is: the smaller the distance from the river confluence, the greater the risk of heavy metal pollution from the tailings dam; the distance from the river confluence is denoted as... ; Distance of river confluence Reverse mapping according to its numerical value to A heavy metal pollution risk score is obtained based on the distance from the river confluence. ; (2) Based on the classification of the downstream rivers flowing into the tailings dam as the logic rule for determining the heavy metal pollution risk of the tailings dam, the rule is: the higher the classification of the flowing river, the greater the risk of heavy metal pollution from the tailings dam. The classification of the flowing river is recorded as follows: ; It will flow into the river level Reverse mapping according to its numerical value to The heavy metal pollution risk score is based on the level of the river into which it flows. .

5. The method for assessing the risk of heavy metal pollution in tailings ponds based on pairwise ranking learning according to claim 1 or 4, characterized in that, The tailings dam heavy metal pollution risk ranking model in step S4 will... As the learning target of the pairwise ranking learning algorithm, the heavy metal pollution risk ranking label of the tailings dam is used. The expression is as follows: ； According to the heavy metal pollution risk ranking labels of tailings ponds The output results are used to establish pairwise sorting learning. The expression for the algorithm's heavy metal pollution risk ranking model for tailings ponds is as follows: ； In the formula This represents a pairwise sorting learning algorithm; Learning through pairwise sorting The heavy metal pollution risk score of the output tailings dam is mapped to the [0, 1] interval using the min-max forward normalization method to obtain the comprehensive heavy metal pollution risk index of the tailings dam, denoted as . , .

6. The method for assessing the heavy metal pollution risk of tailings dams based on pairwise ranking learning according to claim 5, characterized in that, The risk level is determined based on the comprehensive index calculated by the tailings dam heavy metal pollution risk ranking model. The heavy metal pollution risk of tailings dams is divided into five levels: low risk, relatively low risk, medium risk, relatively high risk, and high risk. Among them, low risk Lower risk Medium risk Higher risk High risk .

Images