Non-point source pollution treatment and rain flood storage method and system

Abstract

The application relates to the technical field of environmental data processing, and in particular to a non-point source pollution treatment and rainwater flood storage method and system. The method comprises the following steps: preprocessing pollutant concentration, pollutant accessibility and runoff of a target pixel to obtain target feature data; inputting the target feature data into a trained accessibility classification model for classification to obtain a pollutant accessibility category; and determining a corresponding rainwater storage and pollutant treatment strategy according to a land use type corresponding to the target pixel, the pollutant accessibility category and rainwater quantity and rainwater pollution degree detected by an intelligent interception device. Therefore, the problems of low rainwater collection efficiency, poor treatment effect and low utilization rate of existing measures can be effectively solved.

Description

Technical Field

[0001] This application relates to the field of environmental data processing technology, and in particular to a method and system for non-point source pollution control and stormwater retention. Background Technology

[0002] Due to rapid urban expansion, the areas surrounding drinking water sources are largely comprised of built-up areas, as well as diverse land use types such as woodlands and grasslands, leading to an increasingly complex environment. Current technologies for watershed non-point source pollution prediction, rainwater retention, and water resource utilization in this environment largely employ traditional, piecemeal approaches, lacking comprehensive solutions that consider rainwater harvesting, purification, and utilization. This results in poor rainwater collection and treatment effectiveness and low utilization efficiency. Summary of the Invention

[0003] In view of this, the embodiments of this application provide a method and system for non-point source pollution control and stormwater retention, which can effectively solve the problems of low rainwater collection efficiency, poor treatment effect and low utilization rate of existing measures.

[0004] In a first aspect, embodiments of this application provide a method for non-point source pollution control and stormwater retention, including:

[0005] The pollutant concentration, pollutant accessibility, and runoff of the acquired target pixels are preprocessed to obtain target feature data;

[0006] The target feature data is input into a trained accessibility classification model for classification to obtain pollutant accessibility categories.

[0007] Based on the land use type corresponding to the target pixel, the pollutant accessibility category, and the rainfall and rainwater pollution level detected by the intelligent interception device, the corresponding rainwater retention and pollutant treatment strategies are determined.

[0008] In some embodiments, determining the corresponding rainwater retention and pollutant treatment strategy based on the land use type corresponding to the target pixel, the pollutant accessibility category, and the rainfall and rainwater pollution level detected by the intelligent interception device includes:

[0009] For the target area corresponding to the land use type, the catchment area is divided into multiple target catchment areas based on a preset division standard;

[0010] Based on the pollutant accessibility category corresponding to each target catchment area, and the rainfall and rainwater pollution level of the target catchment area predicted by the intelligent interception device using a regression tree model, a rainwater retention and pollutant treatment strategy is determined.

[0011] Secondly, embodiments of this application provide a non-point source pollution control and stormwater retention control system, wherein the system implements the non-point source pollution control and stormwater retention method provided in the first aspect of this application by controlling various electrically controlled gates.

[0012] In some embodiments, the system includes an intelligent interception device; the intelligent interception device includes: an inlet, a high-pollution rainwater outlet, a low-pollution rainwater outlet, a first well chamber, a second well chamber, a slag screen, an anti-backflow plate, and multiple electrically controlled gates;

[0013] A high-pollution rainwater outlet and a low-pollution rainwater outlet are installed below the first well chamber, and each outlet is equipped with an electrically controlled gate.

[0014] The water inlet is connected to the second well chamber; a slag screen is installed between the second well chamber and the first well chamber; an anti-backflow plate is installed below the slag screen;

[0015] The intelligent interception device controls the corresponding electrically controlled gates according to the rainfall and the degree of rainwater pollution, diverting highly polluted rainwater through the highly polluted rainwater outlet and diverting low-pollution rainwater through the low-pollution rainwater outlet.

[0016] The embodiments of this application have the following beneficial effects:

[0017] This application discloses a method for non-point source pollution control and stormwater retention, comprising: preprocessing the pollutant concentration, pollutant accessibility, and runoff of acquired target pixels to obtain target feature data; inputting the target feature data into a trained accessibility classification model for classification to obtain pollutant accessibility categories; and determining corresponding stormwater retention and pollutant control strategies based on the land use type corresponding to the target pixel, the pollutant accessibility category, and the rainfall and stormwater pollution level detected by the intelligent interception device. This application allows for tailored and categorized measures to address different human activities and accessibility conditions within a watershed. This effectively solves the problems of low stormwater collection efficiency, poor treatment effect, and low utilization rate of existing measures. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This illustration shows an application scenario diagram of the non-point source pollution control and stormwater retention control system according to an embodiment of this application;

[0020] Figure 2 This diagram illustrates another application scenario of the non-point source pollution control and stormwater retention control system according to an embodiment of this application.

[0021] Figure 3 A flowchart illustrating a method for non-point source pollution control and stormwater retention according to an embodiment of this application is shown;

[0022] Figure 4 This illustration shows a schematic diagram of a classification tree in the non-point source pollution control and stormwater retention method according to an embodiment of this application;

[0023] Figure 5 This illustration shows a schematic diagram of the first type of watershed division involved in the river non-point source pollution source identification method according to an embodiment of this application;

[0024] Figure 6 This illustration shows a second type of watershed division involved in the river non-point source pollution source identification method according to an embodiment of this application;

[0025] Figure 7 This illustration shows a schematic diagram of a regression tree in the non-point source pollution control and stormwater retention method according to an embodiment of this application;

[0026] Figure 8 This paper shows a first side view of the intelligent interception well in the non-point source pollution control and stormwater retention method according to an embodiment of this application;

[0027] Figure 9 This paper shows a second side view of the intelligent interception well in the non-point source pollution control and stormwater retention method according to an embodiment of this application;

[0028] Figure 10 The third side view of the intelligent interception well in the non-point source pollution control and stormwater retention method of this application embodiment is shown.

[0029] Explanation of key component symbols:

[0030] 10-Intelligent interception well; 20-Ecological reservoir; 30-Small lake / pond reservoir; 40-Reservoir; 50-Sluice gate; 60-Water replenishment point; 70-Water plant; 80-Drainage area; 90-Clear water ditch; 100-Ecological ditch; 110-Reservoir sluice gate; 120-Pipeline network; 130-Water replenishment pipeline network; 101-Inlet; 102-High-pollution rainwater outlet; 103-Low-pollution rainwater outlet; 104-Sludge trap; 105-Anti-backflow plate; 106-Solar cell; 107-Control room; 201-Manhole cover; 202-First well chamber; 203-Second well chamber; 204-Electrically controlled gate; 2041-Fixed frame; 2042-Robotic arm; 2043-Gate; 205-Automatic water quality monitoring meter; 206-Rain gauge; 207-Level gauge. Detailed Implementation

[0031] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0032] The components of the embodiments of this application described and illustrated in the accompanying drawings can be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of this application provided in the drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0033] In the following text, the terms "comprising," "having," and their cognates, which may be used in various embodiments of this application, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as primarily excluding the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more combinations thereof. Furthermore, the terms "first," "second," "third," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0034] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of this application pertain. Terms (such as those defined in commonly used dictionaries) shall be interpreted as having the same meaning as in their contextual meaning in the relevant technical field and shall not be construed as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of this application.

[0035] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0036] This application first provides a non-point source pollution control and stormwater retention control system. The system implements the non-point source pollution control and stormwater retention method of this application by controlling various electrically controlled gates. In other words, the non-point source pollution control and stormwater retention control system is used to realize rainwater collection, retention, and reuse.

[0037] Demonstratively, scenarios where non-point source pollution control and stormwater retention control systems are applicable include... Figure 1As shown, the scenario includes an ecological reservoir 20, a small lake / pond 30, a reservoir 40, a sluice gate 50, a water replenishment point 60, a water plant 70, a drainage area 80, a clear water ditch 90, an ecological ditch 100, a pipeline network 120, and a water replenishment pipeline network 130. It also includes multiple intelligent interception wells 10, multiple sluice gates 50, and multiple reservoir sluice gates 110. The scenario is divided into a near-natural area, an upstream area with strong human activity, and a downstream area with strong human activity. The non-point source pollution control and stormwater retention control system implements the non-point source pollution control and stormwater retention method of this application embodiment by controlling multiple intelligent interception wells 10, multiple sluice gates 50, and multiple reservoir sluice gates 110.

[0038] The areas are categorized into near-natural zones, upstream areas with high human activity, and downstream areas with high human activity based on different land use types. Near-natural zones are areas with minimal human disturbance and a high degree of naturalness; areas with high human activity are highly urbanized or industrialized areas; upstream areas are water conservation and ecologically sensitive areas; and downstream areas are areas that receive pollutants, are densely populated, and are economically active.

[0039] like Figure 2 As shown, the scene is rasterized to obtain multiple pixels based on a preset resolution; the pixels include information such as elevation and land use type.

[0040] The following examples illustrate the methods for controlling non-point source pollution and storing rainwater.

[0041] Figure 3 A flowchart illustrating a method for non-point source pollution control and stormwater retention according to an embodiment of this application is shown. Exemplarily, this method includes the following steps:

[0042] S100 involves preprocessing the acquired pollutant concentration, pollutant accessibility, and runoff of the target pixels to obtain target feature data. Exemplarily, the target feature data includes graded pollutant accessibility, runoff, and pollutant data.

[0043] Understandably, pretreatment includes classifying runoff, pollutant accessibility, and pollutant accessibility.

[0044] S200: Input the target feature data into the trained accessibility classification model for classification to obtain the pollutant accessibility category.

[0045] Furthermore, reachability classification models include classification tree models; methods for constructing classification tree models include:

[0046] S210, acquire training data; the training data includes indicator attributes of multiple pixels; the indicator attributes include: pollutant concentration, runoff, and pollutant accessibility, etc.

[0047] S220, each indicator attribute is divided into levels according to the corresponding rules to obtain multiple sets of indicator feature data.

[0048] As an example, runoff and pollutant accessibility are classified:

[0049] Runoff and pollutant accessibility are classified into four levels using quartiles: 0-0.25 (Level I), 0.25-0.5 (Level II), 0.5-0.75 (Level III), and 0.75-1 (Level IV). The classification is based on percentages, for example, 0.25 represents 25%.

[0050] Pollutant concentrations are classified into three levels according to Table 1:

[0051] COD: <100 mg / L (Grade I), 100-800 mg / L (Grade II), >800 mg / L (Grade III); TSS: <100 mg / L (Grade I), 100-1000 mg / L (Grade II), >1000 mg / L (Grade III); TP: <0.2 mg / L (Grade I), 0.2-1 mg / L (Grade II), >1 mg / L (Grade III);

[0052] Wherein, COD stands for Chemical Oxygen Demand; TSS stands for Total Suspended Solids, which is the mass concentration of suspended matter in water that cannot pass through a filter membrane (usually with a pore size of 0.45 micrometers), and the unit is generally mg / L; TP stands for Total Phosphorus, which represents the total amount of phosphorus in all forms in water.

[0053] Table 1. Average values ​​and pollution levels of pollutant indicators for different land use types.

[0054]

[0055] Multiple sets of indicator characteristic data, such as those shown in Table 2, include pixel number, pollutant concentration arranged by level, runoff, and pollutant accessibility.

[0056] Table 2 Indicator Feature Data

[0057]

[0058] S230 uses a classification tree algorithm to generate a classification tree model by utilizing multiple sets of indicator feature data.

[0059] Methods for generating classification tree models include:

[0060] S231, Input Data Preparation. The input data consists of the initially selected indicator attributes.

[0061] The indicator attributes include pollutant concentration, runoff, and pollutant accessibility for each pixel. Each pixel serves as a sample point, forming a dataset containing multiple features.

[0062] S232, Feature hierarchical processing.

[0063] Pollutant concentration: Classified into 3 levels according to the type of pollutant (such as COD, TSS, TP).

[0064] Runoff volume: Divided into 4 levels (I ~ IV) using quartiles.

[0065] Pollutant accessibility: also divided into 4 levels (I ~ IV), and used as a target variable or decision-making basis.

[0066] S233, Select the optimal segmentation feature and threshold.

[0067] The Gini index is used to measure the purity of a feature under a given partitioning scheme. A smaller Gini index indicates a stronger ability of the partitioning scheme to distinguish between categories. At each node split, the feature with the smallest Gini index and its optimal split point are selected for the split. That is, the Gini index is calculated for each candidate feature (e.g., pollutant concentration A1, runoff A2). The feature with the smallest Gini index is selected as the current node.

[0068] The Gini index is:

[0069]

[0070] Where K represents the existence of K classes, Let be the probability that a sample point belongs to the k-th class.

[0071] For a given sample set D, its Gini index is:

[0072]

[0073] in, It is the subset of samples in D that belong to the k-th class, where K is the number of classes.

[0074] In this embodiment, taking pollutant concentration as an example, the sample set D of regional pixels is divided into A1 (pollutant concentration) and A2 (runoff) according to different features. A1 is further divided into three parts: I, II, and III. Then, under the pollutant concentration condition, the Gini index of feature A1 is:

[0075]

[0076]

[0077]

[0078] The node with the smallest Gini index is selected as the optimal split point. Similarly, the Gini index of feature A2 is calculated in the same way.

[0079] S234 recursively constructs a tree structure. Starting from the root node, it recursively partitions the current sample set until a stopping condition is met (such as reaching the maximum depth, the number of samples being less than a threshold, or the Gini index being lower than a certain value).

[0080] The root node includes the entire sample set. The first level of splitting involves selecting the optimal feature (e.g., pollutant concentration) and its optimal split point (e.g., the boundary between levels I and II). Sub-nodes continue splitting: the above process is repeated for each subset, selecting new optimal features and split points. The stopping conditions are: all samples in the current node belong to the same class; the preset maximum depth is reached; the number of remaining samples is less than a set threshold; or the Gini index change is less than a certain precision.

[0081] S235, output the classification result.

[0082] The classification results are divided into four pollutant accessibility categories (1-4). For example, accessibility levels include Level 1 accessibility (accessibility 1), Level 2 accessibility (accessibility 2), Level 2 accessibility (accessibility 3), and Level 2 accessibility (accessibility 4). Each level corresponds to different rainwater retention and pollutant treatment strategies. For example, the final classification tree is constructed as follows: Figure 4 As shown. Figure 4 The pollutant concentration, runoff, and accessibility data used in the classification tree are all limited classification data, such as levels I, II, and III. This application uses a classification tree to classify accessibility based on the pollutant concentration, runoff, and accessibility classification data of pixels throughout the watershed, and then constructs rainwater retention and pollutant control strategies based on the pollutant concentration and runoff characteristics corresponding to the four accessibility categories.

[0083] For example, Level 1 accessibility corresponds to low pollutant concentration and small runoff; Level 2 accessibility corresponds to low pollutant concentration and large runoff; Level 3 accessibility corresponds to high pollutant concentration and large runoff; and Level 4 accessibility corresponds to high pollutant concentration and small runoff.

[0084] S300 determines the corresponding rainwater retention and pollutant treatment strategies based on the land use type, pollutant accessibility category, and rainfall and rainwater pollution level detected by the intelligent interception device corresponding to the target pixel.

[0085] This application takes into account various factors, such as land use type, pollutant accessibility category, and rainwater pollution level, and can provide reasonable rainwater retention and pollutant treatment strategies.

[0086] In one implementation, to improve accuracy, based on the land use type, pollutant accessibility category, and rainfall and rainwater pollution level detected by the intelligent interception device corresponding to the target pixel, a corresponding rainwater retention and pollutant treatment strategy is determined, including:

[0087] S310, for the target area corresponding to the land use type, the water catchment area is divided according to the preset division standard to obtain multiple target water catchment areas;

[0088] S320 determines rainwater retention and pollutant treatment strategies based on the pollutant accessibility categories corresponding to each target catchment area and the rainfall and rainwater pollution levels predicted by the intelligent interception device using a regression tree model. The pollutant accessibility categories corresponding to the target catchment areas are the pollutant accessibility categories of the corresponding pixels within those catchment areas.

[0089] Furthermore, for the target areas corresponding to land use types, watershed zones are divided based on preset zoning standards, resulting in multiple target watersheds, including:

[0090] S311, if the land use type is a near-natural area near an upstream drinking water source, multiple target catchment areas are obtained by dividing the near-natural area into pre-defined unit watershed catchment zones; wherein, the pre-defined unit watershed is obtained by the following division method:

[0091] The line connecting the highest points of two adjacent rivers is used as the watershed for each unit watershed. The target watershed (the watershed containing the pixel) is then manually vectorized to divide it into unit watersheds based on rivers. For example... Figure 5 The diagram shown is a schematic representation of watershed division in the prior art. Figure 6 The diagram shown is a schematic representation of the unit watershed division in this application. Figure 5 , Figure 6 The serial numbers in the code represent the codes for each watershed.

[0092] S312 If the land use type is an upstream area with strong human activity near the upstream drinking water source, then multiple target catchment areas are obtained for the upstream area with strong human activity based on the preset unit watershed catchment area.

[0093] S313 If the land use type is a downstream area with strong human activity, then the downstream area with strong human activity is divided into multiple target catchment areas according to the drainage unit zoning; wherein, the drainage unit zoning is an urban drainage unit divided based on human facilities, which is applicable to urban drainage system planning and management.

[0094] Furthermore, based on the pollutant accessibility categories corresponding to each target catchment area, and the rainfall and rainwater pollution levels predicted by the intelligent interception device using a regression tree model for the target catchment area, rainwater retention and pollutant treatment strategies are determined, including:

[0095] Based on the accessibility category of pollutants corresponding to the target catchment area, at least two target nodes for rainwater flow direction are determined, along with the treatment methods at each target node; wherein, if at least two target nodes include intelligent interception devices, the intelligent interception devices determine the rainwater flow direction within the devices based on rainfall amount and rainwater pollution level. Specifically, this includes the following steps:

[0096] S321, if the land use type is a near-natural area near an upstream drinking water source, the target nodes include clear water ditches, ecological ditches, intelligent interception devices, and ecological reservoirs; the treatment methods include natural purification.

[0097] Exemplary, such as Figure 1 As shown, in the case of a near-natural area with land use type corresponding to an upstream drinking water source:

[0098] (1) If the pollutant accessibility category corresponding to the target catchment area is Level 1 accessibility (accessibility 1), then the rainwater is controlled by the intelligent interception well 10 to directly enter the clear water ditch 90 via the ecological ditch 100.

[0099] (2) If the pollutant accessibility category corresponding to the target catchment area is accessibility 2 (corresponding to low pollutant concentration and large runoff), the rainwater is controlled by the intelligent interception well 10 to directly enter the clear water ditch 90 through the ecological ditch 100.

[0100] (3) If the pollutant accessibility category corresponding to the target catchment area is accessibility 3 (corresponding to high pollutant concentration and large runoff), the rainwater is controlled to enter the intelligent interception well 10 through the ecological ditch 100, and the high-pollution rainwater in the intelligent interception well 10 is controlled to enter the ecological reservoir 2 for natural purification, while the low-pollution rainwater enters the clear water ditch 90.

[0101] (4) If the pollutant accessibility category corresponding to the target catchment area is accessibility 4 (corresponding to high pollutant concentration and low runoff), the rainwater is controlled by the intelligent interception well 10 to directly enter the ecological reservoir 20 through the ecological ditch 100 for natural purification.

[0102] To ensure reliability, in near-natural areas near upstream drinking water sources, if the water quality in the ecological reservoir meets standards but the reservoir capacity is less than the preset capacity, purified water from the ecological reservoir will be released into the main reservoir via a controlled sluice gate. If downstream rivers require ecological replenishment, water from the ecological reservoir will be released into the downstream rivers via a controlled sluice gate and clear water ditch. Specifically, if the water quality in ecological reservoir 20 meets standards but the main reservoir capacity is insufficient, purified rainwater from the ecological reservoir can enter the main reservoir; if downstream rivers require ecological replenishment, water can be released from ecological reservoir 20 and released downstream via clear water ditch 90.

[0103] During flood discharge, water from upstream is diverted to the ecological reservoir in the near-natural area by controlling the sluice gate until the preset maximum storage capacity is reached. Then, the water is discharged downstream through the clear water ditch. For example, in extreme weather conditions where rainfall runoff is excessive and flood discharge is necessary, the upstream area can prioritize filling the ecological reservoir 20 in the near-natural area using the sluice gate, thus retaining rainwater. The subsequent rainwater is then discharged through the clear water ditch 90.

[0104] S322, if the land use type is an upstream area with strong human activity near the upstream drinking water source, the target nodes include small lakes, ponds, reservoirs, sewage treatment plants, clear water ditches, and intelligent interception devices; the treatment methods include natural purification, artificial purification, sedimentation, and retention.

[0105] Exemplary, such as Figure 1 As shown, in the case of an upstream area with strong human activity near an upstream drinking water source:

[0106] (1) If the pollutant accessibility category corresponding to the target catchment area is Accessibility 1, control the rainwater to first enter the intelligent interception well, and control the high-pollution rainwater to be transported to the sewage treatment plant through the sewage pipe network through the intelligent interception well, and control the low-pollution rainwater to be discharged into the clear water ditch through the rainwater pipe network. (2) If the pollutant accessibility category corresponding to the target catchment area is Accessibility 2, control the rainwater to first enter the intelligent interception well, and control the high-pollution rainwater to be transported to the sewage treatment plant through the sewage pipe network after sedimentation in the sedimentation tank through the intelligent interception well, and control the low-pollution rainwater to be stored in the small lake pond reservoir before flowing into the clear water ditch. (3) If the pollutant accessibility category corresponding to the target catchment area is Accessibility 3, control the rainwater to first enter the intelligent interception well, and control the high-pollution rainwater to be transported to the sewage treatment plant through the sewage pipe network through the intelligent interception well, and control the low-pollution rainwater to be stored and purified in the small lake pond reservoir before flowing into the clear water ditch. (4) If the pollutant accessibility category corresponding to the target catchment area is accessibility 4, control the rainwater to enter the intelligent interception well first, control the high-pollution rainwater to be transported to the sewage treatment plant through the sewage pipe network through the intelligent interception well, and control the low-pollution rainwater to enter the small lake pond reservoir for purification before flowing into the clear water ditch.

[0107] Furthermore, if the land use type is an upstream area of ​​heavy human activity near an upstream drinking water source, small lakes, ponds, or reservoirs can be converted into waterways between the upstream area of ​​heavy human activity and the downstream river channel, based on natural conditions. The elevation difference between the upstream and downstream areas can be used to divert wastewater from the upstream area of ​​heavy human activity to a downstream wastewater treatment plant for treatment. Specifically, small lakes, ponds, or reservoirs can be converted into waterways between the upstream area of ​​heavy human activity and the downstream river channel, based on natural conditions, to enhance rainwater retention and purification. Since this upstream area of ​​heavy human activity is close to a drinking water source, wastewater treatment plants should be avoided in the area as much as possible. The elevation difference will be used to divert wastewater and highly polluted rainwater to a downstream wastewater treatment plant for treatment.

[0108] S323, if the land use type is a downstream area with strong human activity, the target nodes include rivers, sewage treatment plants, ecological lakes / ponds, intelligent interception devices and rivers; the treatment methods include natural purification, artificial purification and retention.

[0109] In the case where the land use type is a downstream area with strong human activity:

[0110] (1) If the pollutant accessibility category corresponding to the target catchment area is Accessibility 1, high-pollution rainwater is first collected into the sewage network through intelligent interception wells, and low-pollution rainwater is discharged into the river. (2) If the pollutant accessibility category corresponding to the target catchment area is Accessibility 2, rainwater is controlled to first enter the intelligent interception well. The intelligent interception well controls the high-pollution rainwater to be transported to the sewage treatment plant through the sewage network and controls the low-pollution rainwater to enter the ecological lake / pond for rainwater retention. Then, the rainwater is discharged into the downstream river through the ecological lake / pond. (3) If the pollutant accessibility category corresponding to the target catchment area is Accessibility 3, if conditions permit, high-pollution rainwater is first collected into the sewage network through intelligent interception wells, and low-pollution rainwater enters the ecological lake / pond. Then, the ecological lake / pond connects to the downstream river. If an ecological lake / pond cannot be built in the area, the low-pollution rainwater is directly connected to the downstream river. (4) If the pollutant accessibility category corresponding to the target catchment area is accessibility 4, high-pollution rainwater is collected into the sewage pipe network through intelligent interception wells, and low-pollution rainwater is discharged into the river.

[0111] Furthermore, if the land use type is a downstream area of ​​heavy human activity, and there are small lakes / ponds / reservoirs between the upstream area of ​​heavy human activity and the downstream river channel, the treated reclaimed water from the sewage treatment plant is replenished into these small lakes / ponds / reservoirs by controlling the sluice gates. The water is then purified in these small lakes / ponds / reservoirs before being replenished to the downstream river channel. If there are no small lakes / ponds / reservoirs between the upstream area of ​​heavy human activity and the downstream river channel, the reclaimed water is sent to the nearest replenishment point within the downstream area of ​​heavy human activity via a replenishment pipe. Specifically, the small lakes / ponds / reservoirs constructed in the above process can purify the reclaimed water on non-rainy days. The treated reclaimed water from the sewage treatment plant is sent to the nearest replenishment point via a replenishment pipe for ecological replenishment of the river channel. Where there is an ecological lake / pond / reservoir, the reclaimed water is preferentially added to the ecological lake / pond / reservoir for further purification before being replenished to the downstream river channel; where there is no ecological lake / pond / reservoir, replenishment is directly carried out through the replenishment point beside the river channel.

[0112] In one implementation, the intelligent interception device uses a regression tree model to predict rainfall and rainwater pollution levels in the target catchment area to determine rainwater retention and pollutant treatment strategies, including:

[0113] The intelligent interception device uses a generated regression tree model to predict the amount of rainwater and the degree of rainwater pollution based on the rainwater retention time, rainfall intensity, rainfall amount and rainwater pollution level in the target catchment area, and diverts the rainwater according to the degree of rainwater pollution.

[0114] In this embodiment, multivariate regression analysis is used to predict the changes in water quality over time under different rainfall amounts, based on rainfall and water quality data measured by rain gauges and automatic water quality monitoring devices.

[0115] Exemplary methods for generating regression tree models include:

[0116] S410, sort each target variable in the training dataset according to a preset rule to obtain a continuous variable training set; the training dataset includes rainwater retention time, rainfall, rainfall intensity and rainwater pollution level (rainwater pollution level is represented by pollutant concentration); the target variables include rainfall and rainwater pollution level.

[0117] First, prepare the training dataset: obtain a dataset containing multiple samples, where each sample includes an input feature vector (multiple rainwater retention time points and corresponding rainfall intensity) and a corresponding set of target variables (rainfall and rainwater pollution level).

[0118] Table 3 Training Dataset

[0119]

[0120] For each target variable (e.g., rainfall and rainwater pollution level), the values ​​of that target variable in the training set are sorted in ascending or descending order according to their numerical values ​​to obtain an ordered sequence. The ordered sequences corresponding to all target variables constitute the continuous variable training set.

[0121] S420: For each target variable, the corresponding candidate split point is calculated based on the target variables that are adjacent to it in the continuous variable training set.

[0122] For each ordered sequence of target variables, the median value between two adjacent samples is calculated as a candidate split point. For example, candidate thresholds for pollutant concentrations are determined as (A1+A2) / 2, (A2+A3) / 2, ..., (A(n-1)+An) / 2.

[0123] Save all candidate split points as a candidate split set.

[0124] S430: Construct regression tree nodes based on the continuous variable training set, and traverse all candidate split points to calculate the squared difference loss corresponding to each candidate split point. Select the candidate split point corresponding to the minimum squared difference loss as the optimal split point.

[0125] Specifically, initialize the root node, which contains the entire training dataset. Determine the optimal split point for each node. For example, for pollutant concentration, use (C1+C2) / 2, (C2+C3) / 2, ... (C(n-1)+Cn) / 2 as node values, calculate the loss function, and select the node value with the smallest loss function as the optimal split point. Similarly, for runoff, use (A1+A2) / 2, (A2+A3) / 2, ... (A(n-1)+An) / 2 as node values, calculate the loss function, and select the node value with the smallest loss function as the optimal split point.

[0126] In the specific calculation process, for the current node, all candidate split points are traversed, and the squared error loss for each split point is calculated using the following steps:

[0127] (1) Divide the current node dataset into two subsets, left and right, based on the split point; (2) Calculate the average target value of the left and right subsets; (3) Calculate the total squared error loss; (4) Select the split point that minimizes the loss as the optimal split point for the current node.

[0128] S440 recursively divides the optimal split point until the termination condition is met, and outputs the final regression tree model.

[0129] Divide the current node into left and right child nodes based on the optimal split point. Repeat steps (1)-(4) for each child node until any of the following termination conditions are met:

[0130] If the number of samples in the current node is less than a set threshold, the loss decrease is less than a set threshold, the tree height reaches the maximum depth limit, or the variance of all target variables is lower than a certain threshold, it indicates that the data has been sufficiently purified.

[0131] like Figure 7 As shown, the final regression tree model is output, resulting in a complete regression tree. Each leaf node stores the mean (or other statistic) of the target variable for all samples under that node, which serves as the prediction output.

[0132] The mathematical expression for the regression tree model is as follows:

[0133]

[0134] x: Pollutant concentration / rainfall; R m : The time corresponding to the leaf node; m: the index of the leaf node, i.e., which leaf node it is; M is the total number of indexes; c m : The predicted value corresponding to the leaf node after regression analysis; I: When When this condition is met, the corresponding set is unique.

[0135] c m The method for determining the loss function is to minimize it. The loss function is as follows:

[0136]

[0137]

[0138]

[0139] Optimization goal:

[0140]

[0141] At this point, the loss function contains only one unknown parameter. Take the derivative of the above equation directly and set the derivative to 0 to solve. :

[0142]

[0143]

[0144]

[0145] Set leaf nodes Includes samples If there are 1, then the above formula =

[0146] Setting the above expression to 0, we have:

[0147]

[0148] but:

[0149]

[0150] That is, when each leaf node The value is taken from all samples of this node. The loss is minimized when the average value is reached, which is the optimal regression tree.

[0151] Exemplary, Figure 7 Based on real-time monitoring data from a single node—the intercepting well—and continuous data on rainfall intensity, retention time, rainfall amount, and pollutant concentration, multiple regression analysis is used to categorize rainfall amount and pollutant concentration under different rainfall intensities and retention times. This categorization helps the intelligent electronic control device of the intercepting well predict the opening and closing of low-concentration and high-concentration rainwater inlets based on measured data and future rainfall conditions (rainfall intensity and time). When this prediction is accurate enough, future efforts could reduce labor costs and allow the water quality monitor to be removed, with the intelligent controller taking complete control of the intelligent electronic control device's opening and closing based on rainfall intensity and time.

[0152] In one embodiment, the intelligent interception device includes: an inlet, a high-pollution rainwater outlet, a low-pollution rainwater outlet, a first well chamber, a second well chamber, multiple electrically controlled gates, a slag-blocking net, an anti-backflow plate, solar cells, a control room, and a well cover.

[0153] A high-pollution rainwater outlet and a low-pollution rainwater outlet are installed below the first well chamber, and each outlet is equipped with an electrically controlled gate.

[0154] The inlet is connected to the second well chamber; a slag screen is installed between the second well chamber and the first well chamber; an anti-backflow plate is installed below the slag screen;

[0155] The intelligent interception device controls the corresponding electronically controlled gates based on the amount of rainfall and the degree of rainwater pollution. High-pollution rainwater is diverted through the high-pollution rainwater outlet, and low-pollution rainwater is diverted through the low-pollution rainwater outlet.

[0156] Furthermore, the intelligent interception device includes: a rain gauge, an automatic water quality monitoring meter, and a level gauge; the rain gauge is installed on one side above the first well chamber. The automatic water quality monitoring meter is used to measure the measured water quality data required for training the regression tree.

[0157] The intelligent interception device in this application embodiment is not limited to an intelligent interception well.

[0158] like Figure 8 , Figure 9 , Figure 10As shown, the first well chamber 202 is a rectangular brick-built, cement-plastered structure. Electrically controlled gates 204 are installed at the bottom of the well chamber for both high-pollution rainwater outlet 102 and low-pollution rainwater outlet 103, controlling their opening and closing. An automatic water quality monitoring meter 205, a rain gauge 206, and a level gauge 207 are installed on one side above the first well chamber 202. The inlet 101 is connected to the second well chamber 203 within the well chamber.

[0159] After rainwater enters the second well chamber 203 through inlet 101, it first passes through a slag trap 104 to remove large particles of impurities. These impurities are periodically removed and cleaned by maintenance personnel to prevent the accumulation of large amounts of garbage in the well chamber. A backflow prevention plate 105 is installed below the slag trap 104 to prevent rainwater from the first well chamber 202 from flowing back into the second well chamber 203 when the water volume is large, thus avoiding secondary pollution or backflow from the rainwater inlet pipe.

[0160] The electrically controlled gate 204 is powered by solar cells 106 and consists of a fixed frame 2041, a robotic arm 2042, and a gate 2043. The fixed frame 2041 is installed on the well wall below the first well chamber 202, directly above the two rainwater outlets. The robotic arm 2042 is electrically controlled by commands from the control room 107 and the level gauge 207, thereby driving the gate to rise and fall. The gate is connected to the robotic arm 2042, and the rise and fall are controlled by the robotic arm 2042, thereby realizing the opening and closing of the two rainwater outlets.

[0161] The control room 107 is powered by solar cells 106 and consists of a water quality control module and a rainfall monitoring module. The water quality control module monitors the water quality below the low-pollution rainwater drain using an automatic water quality monitor 205, and then determines the water quality control conditions. The rainfall monitoring module monitors the rainfall and rainfall intensity using a rain gauge, and then determines the rainfall control conditions. The control room 107 uses the water quality control module and the rainfall monitoring module to jointly determine the opening and closing of the electrically controlled gate 204.

[0162] The level gauge 207 is powered by solar cell 106 and is located on one side above the first well chamber 202. When the first well chamber 202 reaches the warning water level due to poor drainage during sudden extreme heavy rainfall, it is used to issue a command to open all gates, thereby opening all outlet pipes to achieve rapid drainage and avoid flooding.

[0163] Solar cells 106 are installed on the ground surface, above the manhole cover 201, to power the control room 107, automatic water quality monitor 205, rain gauge 206, level gauge 207 and electric control gate 204 in the intelligent interception well.

[0164] This application comprehensively considers non-point source pollution, rainwater retention, and utilization within the watershed, avoiding the pitfalls of traditional piecemeal approaches that fail to address the overall issues of rainwater collection, purification, and utilization. This application allows for tailored, categorized measures based on varying levels of human activity and accessibility within the watershed, resulting in effective treatment and high water resource utilization. Specifically, regarding the upstream near-natural area: this application not only considers the collection and treatment of non-point source pollution under different accessibility conditions in the upstream near-natural area but also interconnects different treatment units, treatment units with reservoirs, and the entire upstream near-natural area with the downstream river channel, forming an organic whole. This not only effectively collects and purifies non-point source pollution but also serves as a rainwater retention measure and can provide ecological water replenishment to the downstream area through coordinated scheduling. Regarding the upstream area with high human activity: although this area is characterized by high human activity, its geographical location is a transitional zone between the upstream and downstream. Rainwater collection and utilization must be comprehensive and accurate while avoiding additional pollution loads on the near-natural area. This application utilizes the elevation difference to collect highly polluted rainwater for centralized treatment in downstream areas with high human activity, and to further retain and purify low-pollution rainwater in small lakes and ponds near the clear water channel to meet the needs of this area. From the perspective of the downstream area with high human activity: this area has a high pollution load, and the river's water volume depends on upstream reservoir discharge. Therefore, the proposed solution for this area not only treats rainwater in different ways based on accessibility, but also utilizes reclaimed water collected from the wastewater treatment plant for replenishment, thus saving upstream water resources.

[0165] Furthermore, this application uses the intelligent interception well as a key component of the rainwater harvesting system, its function being to accurately identify and separate highly polluted and low-polluted rainwater. Typically, interception wells have two outlets at different elevations for initial and later rainwater runoff, making it impossible to adjust the timing and flow rate of high-polluted and low-polluted rainwater according to actual conditions. The intelligent interception well in this application, based on the initial identification of rainwater with different pollution concentrations using rain gauges and automatic water quality monitors, controls the opening and closing of electrically controlled gates. Later, based on extensive measured data and multiple regression analysis, it accurately predicts rainfall and rainwater quality under different rainfall intensities and durations, thus achieving intelligent opening and closing of the electrically controlled gates. This approach offers the advantage of flexible control over rainwater diversion. It also incorporates structural advantages such as a dual-separation chamber, a sludge net, and an anti-backflow plate, ensuring the intelligent interception well's rainwater separation efficiency under different operating conditions.

[0166] The following specific embodiments illustrate the pollutant concentration (corresponding to the average pollutant concentration below), pollutant accessibility, and runoff of each pixel in the embodiments of this application.

[0167] As an example, the pollutant concentration, pollutant accessibility (pollutant accessibility index), and runoff for each pixel are calculated using the following method:

[0168] S510, based on the geographic information data package corresponding to the target watershed, calculate multiple pollution impact factors of the target watershed raster (corresponding to the above-mentioned pixels) on the target river.

[0169] For example, the geographic information data package includes the following data for the target watershed: 1) Digital elevation model (DEM) data; 2) Watershed raster data; 3) River network raster data; 4) Land use type data; and 5) Runoff data.

[0170] 1) Digital Elevation Model (DEM) data includes: (1) Regular grid point elevations, which are regular grid point plane coordinates (X, Y) and corresponding elevation values ​​(Z) stored in raster form, forming the basic dataset of topographic relief; (2) Topographic factors, such as slope, aspect, and curvature; (3) Hydrological analysis parameters, such as flow direction, runoff accumulation, and watershed boundaries. 2) Watershed raster data: Watershed raster data is a dataset of watershed geographic information stored in regular grid form. It divides the study area into regular small grids (watershed grids), each grid containing specific geographic and attribute information. The following are the geographic and attribute information typically included in watershed raster data: (1) Basic elevation raster, which stores the elevation values ​​(Z) of regular grid units in two-dimensional matrix form. Each grid unit corresponds to plane coordinates (X, Y), which is a direct expression of the spatial distribution of topography; (2) Hydrological feature raster, flow direction, runoff accumulation; (3) Watershed boundaries. 3) River network raster data: River network raster data represents rivers in the form of regular grids, which is convenient for overlay analysis with other raster data (such as watershed raster data, DEM data, etc.). The main contents include: (1) river location information, (2) river grade information, (3) the higher the grade, (4) river width and depth information, (5) flow velocity information, (6) river connectivity information, (7) river elevation information, (8) river buffer zone information, (9) hydrological characteristic information. 4) Land use type data: The use and cover type of land in the study area. The main contents of land use type data usually include:

[0171] (1) Land use classification: According to certain standards (such as the China Land Use Status Classification, Corine LandCover, USGS Land Use Classification, etc.), the study area is divided into different land use types. Common classifications include:

[0172] Urban land: such as residential areas, commercial areas, and industrial areas. Farmland: such as paddy fields and dry land. Forests: such as coniferous forests, broad-leaved forests, and mixed forests. Grasslands: such as natural grasslands and artificial grasslands. Water bodies: such as rivers, lakes, and reservoirs. Bare land: such as deserts, Gobi deserts, and sandy areas. Wetlands: such as swamps and tidal flats. Other special-use land: such as airports, ports, and mines.

[0173] (2) Land use proportion; In the rasterized study area, each river network grid may contain a single or mixed land use type. If it is a mixed type, the proportion (percentage) of each land use type is recorded. For example, a grid may contain 60% farmland, 30% grassland, and 10% water. (3) Land use change information; (4) Soil characteristic information; (5) Human activity intensity. The intensity of human activity is different for each land use type. For example: Urban land: dense buildings, heavy traffic, and pollutants mainly come from domestic sewage and industrial emissions. Farmland: agricultural fertilization and pesticide use may lead to nitrogen and phosphorus pollution. Industrial area: may emit pollutants such as heavy metals and organic matter. (6) Pollutant emission characteristics. (7) Rainwater runoff characteristics.

[0174] 5) Runoff data, which describes the flow of surface water and groundwater in the study area and directly affects the transport and diffusion of pollutants. The following are the main contents that runoff data usually includes: (1) total runoff; (2) surface runoff and groundwater runoff; (3) runoff depth; (4) runoff coefficient; (5) runoff temporal distribution; (6) runoff spatial distribution; (7) runoff path; (8) runoff pollutant concentration; (9) runoff model simulation data, runoff data simulated using hydrological models (such as SWAT, HSPF, HEC-HMS, etc.).

[0175] S520 calculates the pollutant accessibility index of the target watershed grid to the target river based on multiple pollution impact factors.

[0176] For example, pollution influencing factors include, but are not limited to, distance, relative elevation difference, runoff volume, and average pollutant concentration.

[0177] Among them, pollutant accessibility refers to the potential ability of pollutants to enter a sensitive environment or organism through physical migration or chemical diffusion. In this application, the sensitive environment refers to the target river.

[0178] The geographic information data packet in this application can also be a geographic information data packet for the current time period.

[0179] The Pollutant Accessibility Index (PAI) is used to quantify the likelihood of a pollutant reaching a target river from a location such as a watershed grid.

[0180] For example, a higher Pollutant Accessibility Index (PAI) indicates a greater likelihood that a target watershed grid is a potential non-point source pollution source for the target river. For instance, a threshold can be set, and a critical PAI threshold can be determined based on experience or statistical analysis. If the PAI value of a target watershed grid exceeds the PAI threshold, the grid is considered a potential non-point source pollution source for the target river. The location distribution of target watershed grids corresponding to each potential non-point source pollution source is then statistically analyzed to obtain the distribution of non-point source pollution.

[0181] In one implementation, based on the geographic information data package corresponding to the target watershed, multiple pollution impact factors of the target watershed grid on the target river are calculated, including: based on the geographic information data package corresponding to the target watershed, calculating the Euclidean distance, relative elevation difference, runoff and average pollutant concentration of the target watershed grid to the target river.

[0182] The pollutant accessibility index of the target watershed grid to the target river is calculated based on multiple pollution impact factors, including: the distance, relative elevation difference, runoff and average pollutant concentration of the target watershed grid to the target river.

[0183] Furthermore, based on the geographic information data package corresponding to the target watershed, the Euclidean distance, relative elevation difference, runoff, and average pollutant concentration of the target watershed raster to the target river are calculated, including:

[0184] (1) Based on the river network raster data and watershed raster data of the target watershed in the geographic information data package, calculate the Euclidean distance from the center point of the target watershed raster to the center point of the target river network raster; wherein, the target river network raster is the river network raster that is closest to the watershed raster in the target river.

[0185] Exemplary, in this embodiment, the watershed grid is one of the grids obtained by dividing the target watershed according to a set resolution; the river network grid is one of the grids obtained by dividing the target river according to a set resolution. In other words, both the watershed grid and the river network grid are small areas obtained by dividing the target watershed according to a certain resolution.

[0186] Distance: The distance from the river affects the degree of pollution that pollutants cause to the river water.

[0187] ArcGIS is a professional geographic information mapping software that provides users with a scalable and comprehensive GIS platform. It helps users quickly create maps, supports single-user and multi-user editing, and enables complex automated workflows.

[0188] Using the Euclidean distance toolset in ArcGIS, calculate the Euclidean distance between the target watershed raster and the river network raster within the target watershed:

[0189]

[0190] In the formula, , Represents the coordinates of the center point of the i-th river network grid; , This represents the coordinates of the center point of the j-th watershed grid.

[0191] Furthermore, this application also requires the standardization of the Euclidean distance (also known as the proximity factor), specifically using the following formula, where the greater the distance, the lower the accessibility of the pollutant:

[0192]

[0193] in, This represents the pollutant accessibility normalized component of the distance factor in the i-th watershed grid. This represents the Euclidean distance from the i-th watershed grid cell to the center point of the target river network grid cell; This represents the maximum Euclidean distance among all watershed grids within the target watershed.

[0194] In one implementation, the present application embodiment first re-divides the target watershed into multiple unit watersheds, which are obtained using the following method:

[0195] (1) Fill depressions in the digital elevation model data included in the geographic information data package.

[0196] As an example, the "Fill" tool is used to fill depressions in the DEM (Digital Elevation Model) data of the target watershed. Local depressions or depressions in the DEM data can cause flow interruptions or unreasonable flow directions in flow calculations. Open ArcToolbox → Spatial Analyst Tools → Hydrology → Fill. Input the original DEM data, and the process will generate a DEM with filled depressions. This process fills in all unreasonable low-lying points in the DEM data, ensuring the consistency of subsequent flow direction analysis.

[0197] (2) Based on the digital elevation model data after filling depressions, the flow direction of the target watershed grid is calculated using a preset tool;

[0198] The "Flow Direction" tool is used to calculate the flow direction, using the filled DEM data as input to generate flow direction information for each small grid cell. Open ArcToolbox → Spatial Analyst Tools → Hydrology → Flow Direction. Input the filled DEM data. The tool, based on the D8 algorithm, calculates the steepest descent direction of the water flow for each watershed grid cell and uses an encoding (typically 1, 2, 4, 8, 16, 32, 64, 128) to represent one of eight directions. This gives each watershed grid cell a corresponding flow direction value.

[0199] It is important to note that these river networks formed by runoff are not the actual river networks that flow continuously within the city. Therefore, this application overlays the actual river network vector data with the river network generated from DEM data, forcing the water flow to follow the true river direction.

[0200] (3) Connect the points corresponding to the highest elevation between two adjacent rivers to form the watershed of the unit watershed; based on each watershed, use a preset method to divide the target watershed to obtain each unit watershed.

[0201] Specifically, the line connecting the highest points of two adjacent rivers is used as the watershed of a unit watershed. The target watershed is manually vectorized to divide it into unit watersheds based on the rivers. Each grid cell within the divided unit watershed (small watershed) is considered a potential non-point source pollution source for that river. Whether a source is a non-point source pollution source is determined based on the pollutant accessibility index. For example... Figure 5 The diagram shown is a schematic representation of watershed division in the prior art. Figure 6 The diagram shown is a schematic representation of the unit watershed division in this application. Figure 5 , Figure 6 The serial numbers in the code represent the codes for each watershed. Each watershed grid within a unit watershed can be matched with a corresponding watershed grid in each existing watershed, based on its location and size.

[0202] Furthermore, based on the geographic information data package corresponding to the target watershed, the Euclidean distance, relative elevation difference, runoff, and average pollutant concentration of the target watershed grid to the target river are calculated. This includes: calculating the relative elevation difference of the target watershed grid based on the elevation of the target watershed grid within the target unit watershed and the average and minimum elevations of all river network grids of the target river within the target unit watershed.

[0203] As an example, based on DEM data, each watershed unit is coded. Using a zonal statistics tool, the average and minimum elevations of all river network rasters within the target watershed unit are calculated according to the coded fields. Finally, a raster calculator is used to calculate the relative elevation difference between each watershed raster and the river within the target watershed unit. ):

[0204]

[0205] Where i∈1,2,...,n,j∈1,2,...,m, they represent the rows and columns of the watershed raster, respectively; k∈1,2,...,z, represents the coding sequence number of the unit watershed; This represents the relative elevation difference between the watershed grid in the i-th row and j-th column and the rivers within the target watershed unit; This represents the elevation of the watershed raster in the i-th row and j-th column; This represents the average elevation of all river network grids within the target river in the k-th unit watershed; This represents the minimum elevation among all river network grids of the target river within the k-th unit watershed.

[0206] The relative elevation difference is standardized using the following formula; the greater the relative elevation difference, the higher the accessibility of pollutants:

[0207]

[0208] in, This represents the pollutant accessibility standardized component of the relative elevation factor in the k-th unit watershed; This represents the relative elevation difference between the watershed grid in the i-th row and j-th column of the target watershed and the rivers within the target watershed. This represents the maximum relative elevation difference among all unit watersheds within the target watershed.

[0209] In one implementation, based on the geographic information data package corresponding to the target watershed, the average Euclidean distance, relative elevation difference, runoff, and pollutant concentration of the target watershed grid to the target river are calculated, including: estimating the runoff of each watershed grid using a preset model based on the watershed runoff data and land use type data of the target watershed in the geographic information data package.

[0210] The amount of runoff is a crucial factor determining the accessibility of pollutants. To quickly and accurately estimate direct surface runoff under specific rainfall events, especially in the absence of detailed hydrological observation data, this application embodiment uses watershed runoff data and land use type data, employing ArcSWAT on the ArcGIS platform based on the SCS-CN (Soil Conservation Service Curve Number method) model, to estimate direct surface runoff for specific rainfall events. SCS-CN is an empirical model developed for estimating direct surface runoff for specific rainfall events. The runoff curve parameter CN (Curve Number) is the only key parameter of this model, jointly determined by soil type, anterior soil moisture conditions, and land use type. When regional soil type and moisture are relatively stable, changes in CN values ​​are mainly caused by changes in land use type. CN values ​​typically range from 0 to 100, with higher values ​​indicating greater runoff generation. The specific steps for calculating runoff include: inputting rainfall data and watershed characteristic parameters; obtaining CN values ​​from tables based on soil type and land use type; calculating direct surface runoff using formulas; and outputting the results.

[0211] To improve accuracy, this application embodiment also introduces machine learning algorithms to optimize the CN value, or dynamically adjusts the CN value by combining remote sensing data. Specifically, the machine learning algorithm optimization of the CN value in the SCS-CN model includes: input features including, but not limited to, the following variables: rainfall, soil type, land use type, terrain features (slope, elevation), vegetation cover, and initial humidity conditions.

[0212] Target variable: Actual observed runoff or historical CN values.

[0213] As an example, regression models such as linear regression, random forest regression, and gradient boosting regression (GBDT) are used to optimize the CN value, which is suitable for predicting continuous CN values. Objective: To establish a mapping relationship between input features and CN values.

[0214] The steps include: collecting historical data (rainfall, runoff, soil type, etc.); using a regression model to fit the relationship between the input features and the actual CN value; validating the model performance and adjusting hyperparameters to improve prediction accuracy.

[0215] Furthermore, the runoff factor is standardized using the following formula: the larger the runoff, the higher the accessibility of pollutants.

[0216]

[0217] in, This represents the pollutant accessibility normalized component of the runoff factor in the i-th watershed grid. This represents the runoff of the i-th watershed grid cell; This represents the maximum runoff value among all watershed grids within the target watershed.

[0218] Furthermore, based on the geographic information data package corresponding to the target watershed, the Euclidean distance, relative elevation difference, runoff, and average pollutant concentration of the target watershed grid to the target river are calculated, including: based on the total pollutant discharge, runoff, area, and land use type or percentage of land use type in the target watershed grid in the geographic information data package, the average pollutant concentration of the target watershed grid is calculated.

[0219] In one implementation, a watershed grid for a single land use type is applied, with only one land use type in each watershed grid. Exemplarily, the average pollutant concentration is calculated using the following formula:

[0220]

[0221] in, M represents the average value of pollutants in a single rainfall runoff; M represents the total discharge of a certain pollutant in a single rainfall runoff, in grams (g); V represents the runoff volume of the i-th watershed grid, obtained as described above in the runoff volume calculation method. This represents the area of ​​the i-th watershed raster; it is generally consistent with the precision in the DEM data. For example, if the precision of the DEM data is 30m, then the watershed raster is 30*30m. 2 . This represents the average pollutant index corresponding to different land use types. The average pollutant index can be, for example, any one of average COD, average TSS, and average TP. The land use type of the i-th watershed grid is determined from the land use type data.

[0222] Average COD (Chemical Oxygen Demand) represents the total amount of organic and inorganic matter in water that can be oxidized by strong oxidants, reflecting the degree of organic pollution in the water body. Average TSS (Total Suspended Solids) refers to the total mass of suspended particulate matter per unit volume of water (usually expressed in mg / L), used to measure the turbidity and particulate pollution of water bodies. Average TP (Total Phosphorus) refers to the total amount of phosphorus in all forms (including dissolved and particulate forms) in water bodies, and is an important indicator for assessing eutrophication, as shown in Table 1.

[0223] In practical applications, many watershed rasters may contain multiple land use types (e.g., half urban impervious surfaces and the other half green spaces). Simply classifying the entire watershed raster into one type could lead to errors. Therefore, in this embodiment, the contribution of each land use type is allocated proportionally.

[0224] In another implementation, to improve identification accuracy and broaden the scope of application, such as to meet the requirements of watershed grids applicable to mixed land use types, the average pollutant concentration is calculated using the following formula:

[0225]

[0226] Where N represents the total number of land use types; M represents the average value of pollutants in a single rainfall runoff; M represents the total discharge of a certain pollutant in a single rainfall runoff, in grams (g); V represents the runoff volume of the i-th watershed grid, obtained as described above in the runoff volume calculation method. This represents the area of ​​the nth land use type in the i-th watershed grid; This represents the average value of pollutant indicators corresponding to the nth land use type.

[0227] Assuming that in the i-th watershed grid, impermeable ground accounts for 50% and green space accounts for 50%, then in the formula... S1 (impermeable surface) × (Impervious surface) + S2 (Green space) × (Green space).

[0228] Furthermore, the method also includes:

[0229] (1) Based on the watershed raster data in the geographic information data package, a grid vector file is created using a preset tool at a set resolution to generate a grid vector file that is identical to each watershed raster (in both location and size). The set resolution is the same as the resolution used when dividing the raster in the target watershed. The grid vector file includes data segmentation and block management, dividing the target watershed into regular small units (a grid cell), which facilitates block processing of large-scale data. Each grid cell is treated as an independent analysis object.

[0230] (2) Based on the grid vector file, use the preset area tabulation tool to count the percentage of different land use types in the grid cells, and determine the area of ​​different land use types according to the percentage of different land use types.

[0231] As an example, create a fishing net in ArcGIS to generate a grid vector file with the same size as the watershed raster described above. Specifically: 1) Open ArcGIS software. 2) Use the "Create Fishnet" tool. 3) Set the cell size of the fishing net to be the same as the existing raster size (e.g., 10 meters × 10 meters). 4) Ensure the fishing net covers the entire study area (target watershed). 5) Output the generated grid vector file.

[0232] Use the area tabulation tool to calculate the percentage of different land use types within a grid. Determine the area of ​​each land use type based on its percentage. Specifically: 1) Load the land use type data (usually in raster format) into ArcGIS. 2) Use the "Tabulate Area" tool. 3) Enter the following parameters:

[0233] Zone Layer: Select the fishnet vector file you just generated (each grid cell represents a watershed raster). Class Layer: Select the land use type raster data. Output Table: Specify the path to the table containing the output statistical results.

[0234] 4) After running the tool, a table file will be generated, listing the area and percentage of different land use types in each watershed raster.

[0235] Results: A detailed statistical table was obtained, recording the proportion of various land use types in each watershed grid.

[0236] The average pollutant concentration was standardized using the following formula: the higher the average pollutant concentration, the higher the pollutant accessibility.

[0237]

[0238] This represents the pollutant accessibility normalized component of the runoff factor in the i-th watershed grid. This represents the average value of pollutants in the i-th watershed grid. This represents the maximum value of the average pollutant concentration across all watershed grids within the target watershed.

[0239] In one implementation, based on the Euclidean distance, relative elevation difference, runoff, and average pollutant concentration of the target watershed grid to the target river, a pollutant accessibility index of the target watershed grid to the target river is calculated, including:

[0240] Based on the maximum values ​​corresponding to each value within the target watershed, the average values ​​of Euclidean distance, relative elevation difference, runoff, and pollutant concentration are standardized.

[0241] The pollutant accessibility is obtained by multiplying the standardized Euclidean distance, relative elevation difference, runoff, and average pollutant concentration.

[0242] The pollutant accessibility index A is used to evaluate the accessibility of pollutants from various potential pollution sources within the watershed. The calculation formula is as follows:

[0243]

[0244] in, Let x be the pollutant accessibility index for the x-th watershed grid. The standardized components of pollutant accessibility caused by each influencing factor; The number of influencing factors in the accessibility evaluation model is n, which is the number of influencing factors. For example, in the embodiment of this application, n=4.

[0245] This application also provides a terminal device, exemplary of which includes a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to enable the terminal device to perform the above-described methods for non-point source pollution control and stormwater retention.

[0246] The processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, including at least one of a Central Processing Unit (CPU), Graphics Processing Unit (GPU), Network Processor (NP), Digital Signal Processor (DSP), Application-Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor or any conventional processor, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application.

[0247] The memory can be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The memory is used to store computer programs, and the processor can execute the computer programs accordingly after receiving execution instructions.

[0248] This application also provides a computer-readable storage medium for storing the computer program used in the aforementioned terminal device. For example, the computer-readable storage medium may include, but is not limited to, various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0249] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that, in alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0250] In addition, the functional modules or units in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0251] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a smartphone, personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0252] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for non-point source pollution remediation and rainwater flood storage, characterized in that, The method comprises the following steps: Preprocessing the obtained target pixel's pollutant concentration, pollutant accessibility and runoff to obtain target feature data; The pollutant accessibility is used to quantify the possibility of the pollutant reaching the target river from the target pixel; The target feature data is input into a trained accessibility classification model for classification to obtain a pollutant accessibility category; wherein the accessibility classification model comprises a classification tree model; the pollutant accessibility category is the accessibility level output by the accessibility classification model; wherein the construction method of the classification tree model comprises: obtaining training data; the training data comprises index attributes of multiple pixels; the index attributes comprise: pollutant concentration, runoff and pollutant accessibility; each index attribute is divided into levels according to the corresponding rules to obtain multiple sets of index feature data; the classification tree model is generated by using the multiple sets of index feature data and the classification tree algorithm; According to the corresponding land use type of the target pixel, the pollutant accessibility category, and the rainfall and rainwater pollution degree detected by the intelligent interception device, a corresponding rainwater detention and pollutant treatment strategy is determined, specifically comprising: based on a preset division standard, the target area corresponding to the land use type is divided into multiple target catchment areas; according to the pollutant accessibility category corresponding to each target catchment area, and the rainfall and rainwater pollution degree of the target catchment area predicted by the intelligent interception device using the regression tree model, a rainwater detention and pollutant treatment strategy is determined. 2.The method according to claim 1, wherein, The intelligent interception device uses the regression tree model to predict the rainfall and rainwater pollution degree of the target catchment area to determine the rainwater detention and pollutant treatment strategy, comprising: The intelligent interception device uses the generated regression tree model to predict the rainfall and rainwater pollution degree of the rainwater based on the rainwater retention time, rain intensity, rainfall and rainwater pollution degree in the target catchment area, and according to the rainwater pollution degree, the rainwater is diverted. 3.The method according to claim 1, wherein, The target area corresponding to the land use type is divided into multiple target catchment areas based on a preset division standard, comprising: If the land use type is a near-natural area near an upstream drinking water source, the near-natural area is divided into multiple target catchment areas according to a preset unit watershed catchment division; If the land use type is an upstream strong human activity area near an upstream drinking water source, the upstream strong human activity area is divided into multiple target catchment areas according to the preset unit watershed catchment division; wherein the upstream strong human activity area is a strong human activity area upstream; the strong human activity area is a highly urbanized or industrialized area; If the land use type is a downstream strong human activity area, the downstream strong human activity area is divided into multiple target catchment areas according to a drainage unit division. 4.The method according to claim 1, wherein, According to the pollutant accessibility category corresponding to each target catchment area, and the rainfall and rainwater pollution degree of the target catchment area predicted by the intelligent interception device using the regression tree model, a rainwater detention and pollutant treatment strategy is determined, comprising: According to the pollutant accessibility category corresponding to the target catchment area, at least two target nodes of the rainwater flow direction are determined, and a treatment mode at each target node is determined; wherein, if the at least two target nodes include the intelligent interception device; the intelligent interception device determines the rainwater flow direction in the intelligent interception device according to the rainfall and the rainwater pollution degree; If the land use type is a near-natural area near an upstream drinking water source, the target node includes a clean water ditch, an ecological ditch, an intelligent interception device and an ecological reservoir; and the treatment mode includes natural purification; If the land use type is an upstream strong human activity area near an upstream drinking water source, the target node includes a small lake pond reservoir, a sewage treatment plant, a clean water ditch and an intelligent interception device; and the treatment mode includes natural purification, artificial purification, sedimentation and detention; wherein, the upstream strong human activity area is a strong human activity area upstream; and the strong human activity area is a highly urbanized or industrialized area; If the land use type is a downstream strong human activity area, the target node includes a river channel, a sewage treatment plant, an ecological lake / pond, an intelligent interception device and a river channel; and the treatment mode includes natural purification, artificial purification and detention. 5.The method according to claim 4, wherein, The rainwater detention and pollutant treatment strategy further includes: In the case where the land use type is a near-natural area near an upstream drinking water source, if the water quality in the ecological reservoir meets the standard, if the reservoir capacity is less than a preset capacity, the purified water in the ecological reservoir is flowed into the reservoir by controlling the water gate; if the downstream river channel needs ecological water replenishment, the water in the ecological reservoir is replenished to the downstream river channel through the clean water ditch by controlling the water gate; During flood discharge, the upstream water is stored in the ecological reservoir in the near-natural area by controlling the water gate until a preset maximum storage capacity is reached, and then the upstream water is discharged through the clean water ditch; And / or, in the case where the land use type is an upstream strong human activity area near an upstream drinking water source, a small lake pond reservoir is modified between the upstream strong human activity area and the downstream river channel according to the natural conditions; the sewage in the upstream strong human activity area is introduced into the downstream sewage treatment plant for treatment by using the height difference between the upstream strong human activity area and the downstream; And / or, in the case where the land use type is a downstream strong human activity area, when there is a small lake pond reservoir between the upstream strong human activity area and the downstream river channel, the reclaimed water treated by the sewage treatment plant is replenished into the small lake pond reservoir, and then the water is replenished to the downstream river channel after being purified by the small lake pond reservoir; when there is no small lake pond reservoir between the upstream strong human activity area and the downstream river channel, the reclaimed water is sent to the nearest water replenishment point in the downstream strong human activity area through a water replenishment pipe.

6. The method according to any one of claims 1-5, wherein, The method for generating the regression tree model includes: Each target variable in a training data set is sorted according to a preset rule to obtain a continuous variable training set; the training data set includes rainwater detention time, rain intensity, rainfall and rainwater pollution degree; and the target variable includes rainfall and rainwater pollution degree; For each target variable, a corresponding candidate split point is calculated according to adjacent target variables in the continuous variable training set; According to the continuous variable training set, a regression tree node is constructed, and the square difference loss corresponding to each candidate split point is calculated by traversing all candidate split points, and the candidate split point corresponding to the minimum square difference loss is selected as the optimal split point; The optimal split point is recursively divided until the termination condition is met, and the final regression tree model is output.

7. A non-point source pollution remediation and stormwater detention control system, characterized in that, The system implements the method for surface source pollution remediation and rainwater storage according to any one of claims 1-6 by controlling the electrically controlled gates.

8. The area source pollution remediation and stormwater detention control system of claim 7, wherein, The system comprises an intelligent interception device, wherein the intelligent interception device comprises a water inlet, a high-pollution rainwater outlet, a low-pollution rainwater outlet, a first well chamber, a second well chamber, a debris screen, an anti-backflow plate, and a plurality of electrically controlled gates. The high-pollution rainwater outlet and the low-pollution rainwater outlet are arranged below the first well chamber, and each outlet is provided with an electrically controlled gate. The water inlet is in communication with the second well chamber, and the second well chamber is provided with the debris screen between the first well chamber. The anti-backflow plate is arranged below the debris screen. The intelligent interception device can divert high-pollution rainwater through the high-pollution rainwater outlet and low-pollution rainwater through the low-pollution rainwater outlet by controlling the corresponding electrically controlled gates according to the rainfall and the pollution degree of the rainwater.

Images