Agricultural non-point source pollution early warning method and system, and electronic device

Abstract

The present invention relates to the field of agricultural environment monitoring, and provides an agricultural non-point source pollution early warning method and system, and an electronic device. The method includes: obtaining first pollution monitoring parameters of an area to be monitored; predicting second pollution monitoring parameters of the area to be monitored within a set time period; when the second pollution monitoring parameters meet the first early warning condition, giving a first-level early warning, and when the second pollution monitoring parameters do not meet the first early warning condition but meet the second early warning condition, giving a second-level early warning; calculating the pollution contribution rate of the area to be monitored to the downstream area through a graph neural network, and if the pollution contribution rate meets the third early warning condition, giving a third-level early warning. It is used to solve the defect that the existing pollution monitoring means are difficult to accurately identify and early warn of pollution and cannot locate the pollution source. Through the hierarchical early warning mechanism, different levels of early warning can be carried out for agricultural non-point source pollution, and real-time monitoring, dynamic early warning and precise treatment of regional pollution can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural environment monitoring, and particularly to a method and system for warning of agricultural non-point source pollution, and an electronic device. Background Art

[0002] Agricultural non-point source pollution refers to the nitrogen, phosphorus, organic matter and other nutrients generated during agricultural production due to the unreasonable use of chemical inputs such as chemical fertilizers, pesticides, and plastic films, as well as the untimely or improper treatment of livestock and poultry aquaculture waste, crop straw, etc. Driven by rainfall and topography, carried by surface and underground runoff and soil erosion, it accumulates excessively in the soil or enters the receiving water body, causing pollution to the ecological environment.

[0003] Paddy fields in the plain river network area are in a long-term flooded state. After fertilizing the paddy fields, fertilizer elements such as nitrogen and phosphorus are easily discharged into the ditches and river networks with drainage water, resulting in agricultural non-point source pollution. Since the water systems in the river network areas where paddy fields are located are often intricate, the pollution monitoring means in the prior art are difficult to accurately identify and warn of pollution, and it is also impossible to locate the pollution sources, resulting in serious environmental pollution caused by agricultural non-point source pollution. Summary of the Invention

[0004] The present invention provides a method and system for warning of agricultural non-point source pollution, and an electronic device, which are used to solve the defects that the pollution monitoring means in the prior art are difficult to accurately identify and warn of pollution, and it is also impossible to locate the pollution sources. The solution of the present application can warn of agricultural non-point source pollution at different levels through a hierarchical warning mechanism, and can realize real-time monitoring, dynamic warning and precise treatment of regional pollution.

[0005] The present invention provides a method for warning of agricultural non-point source pollution, including:

[0006] Obtaining a first pollution monitoring parameter of the area to be monitored;

[0007] Based on the first pollution monitoring parameter, predicting a second pollution monitoring parameter of the area to be monitored within a set time period through a pre-constructed prediction model;

[0008] When the second pollution monitoring parameter meets the first warning condition, a first-level warning is given; when the second pollution monitoring parameter does not meet the first warning condition but meets the second warning condition, a second-level warning is given;

[0009] When the second pollution monitoring parameter does not meet the first warning condition and the second warning condition, calculating the pollution contribution rate of the area to be monitored to the downstream area through a graph neural network. If the pollution contribution rate meets the third warning condition, a third-level warning is given.

[0010] According to the agricultural non-point source pollution warning method provided by the present invention, the prediction model is a coupled model of a Soil and Water Assessment Model (SWAT) and a Long Short-Term Memory (LSTM) network model.

[0011] According to the agricultural non-point source pollution warning method provided by the present invention, the construction process of the prediction model includes:

[0012] Collect training data, where the training data includes historical pollution monitoring parameters of the area to be monitored.

[0013] Based on the training data, train the Soil and Water Assessment Model and the Long Short-Term Memory network model respectively.

[0014] Use the output of the Soil and Water Assessment Model as the input of the Long Short-Term Memory network model, perform iterative optimization on the Long Short-Term Memory network model, and based on the output of the Long Short-Term Memory network model, perform iterative optimization on the Soil and Water Assessment Model.

[0015] According to the agricultural non-point source pollution warning method provided by the present invention, the second pollution monitoring parameter includes soil nitrogen content and the rainfall probability of the area to be monitored within a set time period, and the first warning condition includes:

[0016] The soil nitrogen content exceeds the nitrogen content threshold, and the rainfall probability of the area to be monitored within the set time period exceeds the rainfall probability threshold.

[0017] According to the agricultural non-point source pollution warning method provided by the present invention, the second pollution monitoring parameter includes the phosphorus content at the drainage outlet of the area to be monitored and the downstream flow velocity of the area to be monitored, and the second warning condition includes:

[0018] The phosphorus content at the drainage outlet of the area to be monitored exceeds the phosphorus content threshold, and the downstream flow velocity of the area to be monitored is lower than the flow velocity threshold.

[0019] According to the agricultural non-point source pollution warning method provided by the present invention, the downstream area of the area to be monitored is determined by the following method:

[0020] Construct a virtual river network model of the area to be monitored, where the virtual river network model at least reflects the geographical spatial data and hydrological data of the area to be monitored.

[0021] Based on the virtual river network model, determine the pollutant diffusion path of the area to be monitored.

[0022] Based on the pollutant diffusion path, determine the downstream area of the area to be monitored.

[0023] According to the agricultural non-point source pollution early warning method provided by the present invention, calculating the pollution contribution rate of the area to be monitored to the downstream area through a graph neural network includes:

[0024] Calculating the concentration of pollutants in the area to be monitored through the graph neural network, and calculating the concentration of pollutants in the downstream area of the area to be monitored, where the pollutants include nitrogen and phosphorus;

[0025] Based on the pollutant concentration in the area to be monitored and the pollutant concentration in the downstream area of the area to be monitored, determining the pollution contribution rate of the area to be monitored to the downstream area.

[0026] According to the agricultural non-point source pollution early warning method provided by the present invention, the graph neural network is constructed by the following method:

[0027] Taking the area to be monitored as graph nodes and the rivers between the areas to be monitored as edges, constructing a graph structure;

[0028] Performing iterative training on the graph structure to optimize the weights of the edges in the graph structure, and obtaining the graph neural network.

[0029] The present invention also provides an agricultural non-point source pollution early warning system, including:

[0030] A parameter acquisition module for acquiring the first pollution monitoring parameter of the area to be monitored;

[0031] A parameter prediction module for predicting the second pollution monitoring parameter of the area to be monitored within a set time period based on the first pollution monitoring parameter through a pre-constructed prediction model;

[0032] A first early warning module for performing a first-level early warning when the second pollution monitoring parameter meets the first early warning condition, and performing a second-level early warning when the second pollution monitoring parameter does not meet the first early warning condition but meets the second early warning condition;

[0033] A second early warning module for, when the second pollution monitoring parameter does not meet the first and second early warning conditions, calculating the pollution contribution rate of the area to be monitored to the downstream area through a graph neural network, and performing a third-level early warning if the pollution contribution rate meets the third early warning condition.

[0034] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the program, it implements any one of the above agricultural non-point source pollution early warning methods.

[0035] The present invention also provides a non - transitory computer - readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements any one of the above - mentioned agricultural non - point source pollution warning methods.

[0036] The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements any one of the above - mentioned agricultural non - point source pollution warning methods.

[0037] In the agricultural non - point source pollution warning method provided by the present invention, first, the pollution situation in a future time period can be predicted based on a pre - constructed prediction model to achieve early warning of the future pollution situation. And through a multi - level warning method, strict warning conditions are set for each level of warning. When the corresponding warning conditions are met, corresponding warnings are given, which can help the staff accurately judge the agricultural non - point source pollution in the area to be monitored. When neither the first warning condition nor the second warning condition is met, the key monitoring area can also be determined by calculating the pollution contribution rate through a graph neural network for a third - level warning. In this way, the source tracing of the pollution source can also be achieved, which is beneficial to the subsequent rapid treatment of pollution. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the following - described drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts.

[0039] Figure 1 is one of the flow - chart diagrams of the agricultural non - point source pollution warning method provided by the embodiments of the present invention;

[0040] Figure 2 is the second flow - chart diagram of the agricultural non - point source pollution warning method provided by the embodiments of the present invention;

[0041] Figure 3 is the structural diagram of the graph neural network model provided by the embodiments of the present invention;

[0042] Figure 4 is one of the structural diagrams of the agricultural non - point source pollution warning system provided by the embodiments of the present invention;

[0043] Figure 5 is the second structural diagram of the agricultural non - point source pollution warning system provided by the embodiments of the present invention;

[0044] Figure 6 is the physical structural diagram of the electronic device provided by the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0046] Figure 1 It is a schematic flowchart of the agricultural non-point source pollution early warning method provided by an embodiment of the present invention.

[0047] As Figure 1 shown, this embodiment provides an agricultural non-point source pollution early warning method, including:

[0048] Step 101, obtaining a first pollution monitoring parameter of the area to be monitored;

[0049] The area to be monitored can be a plain river network area. There can be multiple areas to be monitored. That is, in practical applications, the plain river network area can be divided into multiple areas to be monitored, and each area to be monitored can be monitored separately. For example, each paddy field can be divided into an area to be monitored and monitored separately.

[0050] The obtained first pollution monitoring parameter can reflect the pollution situation of the area to be monitored. The first pollution monitoring parameter can be obtained from multiple channels. For example, sensors can be deployed in the area to be monitored, and the nitrogen element content and phosphorus element content in the soil can be collected through the sensors. The nitrogen and phosphorus sensors set in the soil can be evenly set in the paddy field, and more sensors can be set at the ridges of the paddy field. For example, nitrogen and phosphorus sensors with double density can be set. The nitrogen and phosphorus sensors can be used to monitor the migration of nitrogen and phosphorus elements under flooded conditions in the paddy field. In practice, multi-parameter water quality sensors can also be set at the drainage outlet of the paddy field to collect the pH value, conductivity, ammonia nitrogen content, total phosphorus data, etc. of the water quality of the paddy field in real time. Buoy-type monitoring stations can also be arranged at the confluence of the plain river network. The monitoring stations are integrated with a flow velocity monitor for monitoring the flow velocity and water quality sensors.

[0051] In addition, data of the area to be monitored can be collected through an aerial remote sensing platform. For example, a drone can carry a hyperspectral and thermal infrared camera. The spectrum of the camera can be from 400 to 1000 nanometers. Within three days after fertilizing the paddy field, it flies along the ditches of the paddy field every day. In practical applications, the drone can scan the paddy field at a low altitude of 30 meters. The drone can retrieve the chlorophyll content through hyperspectral data to mark the areas with uneven fertilization. Satellite data can also be integrated to extract the NDWI index to monitor the suspended sediment content in the river network. For example, the satellite data can be Sentinel-2 satellite data.

[0052] In implementation, the nitrogen and phosphorus sensors in the soil can upload the collected parameters once every set time interval. For example, data can be uploaded once every 15 minutes. After removing outliers through the edge node, it can be used as the first pollution monitoring parameter. Outliers include the drift of nitrogen and phosphorus readings caused by water interference.

[0053] Step 102: Based on the first pollution monitoring parameter, through a pre-constructed prediction model, predict the second pollution monitoring parameter of the area to be monitored within a set time period.

[0054] In implementation, the prediction model can be a coupled model of the Soil and Water Assessment Tool (SWAT) and the Long Short-Term Memory (LSTM) network model. Among them, the SWAT can be used to simulate the hydrological cycle of the paddy field, and the LSTM model can be used to predict the second pollution monitoring parameter within a future set time period. In practical applications, it can be used to predict the second pollution monitoring parameter for the next 3 days. Among them, the second pollution monitoring parameter can include the pollution peak.

[0055] In this step, by inputting the first pollution monitoring parameter obtained in step 101 into the prediction model, the nitrogen and phosphorus loss flux can be calculated through the SWAT model, in kg / ha, and the pollutant concentration curve at the drainage outlet of the paddy field in the next three days can be predicted through the LSTM model.

[0056] Step 103: When the second pollution monitoring parameter meets the first warning condition, issue a first-level warning. When the second pollution monitoring parameter does not meet the first warning condition but meets the second warning condition, issue a second-level warning.

[0057] In practical applications, the above first-level warning and second-level warning can be issued at the monitoring station of the staff. For example, warning reminders can be made in the form of indicator lights or buzzers. Exemplarily, the first-level warning can be issued with a yellow indicator light, and the second-level warning can be issued with a red indicator light.

[0058] Step 104, when the second pollution monitoring parameter does not meet the first warning condition and the second warning condition, calculate the pollution contribution rate of the area to be monitored to the downstream area through a graph neural network. If the pollution contribution rate meets the third warning condition, a third-level warning is issued.

[0059] The graph neural network (GNN) in this step can be used to analyze the pollution diffusion path and locate high-risk plots where the pollution contribution rate meets the third warning condition. High-risk plots represent areas that may cause significant pollution hazards to the downstream area. Early location can help with subsequent timely handling.

[0060] In an exemplary embodiment, the prediction model is a coupled model of a soil and water assessment model and a long short-term memory network model.

[0061] In an exemplary embodiment, the construction process of the prediction model includes:

[0062] Collect training data, where the training data includes historical pollution monitoring parameters of the area to be monitored;

[0063] Based on the training data, train the soil and water assessment model and the long short-term memory network model respectively;

[0064] Use the output of the soil and water assessment model as the input of the long short-term memory network model, iteratively optimize the long short-term memory network model, and based on the output of the long short-term memory network model, iteratively optimize the soil and water assessment model.

[0065] The following details the construction process of the prediction model, including the following steps.

[0066] 1. Data collection and preprocessing,

[0067] First, the coupled model needs to collect and organize various types of data, including but not limited to:

[0068] Meteorological data: including precipitation, temperature, humidity, wind speed, etc., usually from weather stations or remote sensing data.

[0069] Soil data: including soil type, soil fertility, soil moisture, etc.

[0070] Agricultural management data: such as data on agricultural activities like fertilization, pesticide use, irrigation, etc.

[0071] Hydrological data: including watershed hydrological characteristics, such as runoff, groundwater level, soil moisture, etc.

[0072] Pollutant data: Concentration data of different types of agricultural pollutants (such as nitrogen, phosphorus, pesticides, etc.) in water bodies or soil.

[0073] Subsequently, a data preprocessing process is carried out, including but not limited to:

[0074] Missing value handling: Use methods such as interpolation, mean filling, or previous value filling to handle missing values in the data.

[0075] Data standardization: Standardize or normalize the input data (such as meteorological data, pollutant concentrations, etc.) so that various types of data have the same scale, enabling the LSTM model to be better trained.

[0076] Time alignment: Align all data according to a unified time step (such as daily, monthly, etc.) to ensure consistency between different data sources.

[0077] 2. Construction of the SWAT model

[0078] The core function of the SWAT model is to simulate the hydrological process and the dynamic behavior of pollutants in soil and water bodies. The construction of the SWAT model usually includes the following steps:

[0079] 2.1 Model initialization

[0080] Input data: Includes meteorological data (precipitation, temperature, etc.), land use data, soil data, watershed boundary data, etc.

[0081] Determine the model area: Divide the watershed, sub-watershed, land use types, etc. according to the research area to establish the basic spatial units of the SWAT model.

[0082] 2.2 Hydrological process simulation

[0083] Precipitation and evaporation: Simulate the infiltration, evaporation, and plant transpiration of precipitation.

[0084] Runoff and soil moisture: Calculate hydrological processes such as surface runoff and groundwater infiltration to obtain data such as soil moisture and runoff.

[0085] Pollutant simulation: Calculate the loss and migration of agricultural non-point source pollutants (such as nitrogen, phosphorus, pesticides, etc.) through the model to obtain the concentration changes of pollutants.

[0086] The outputs of the SWAT model include hydrological data and pollutant data. Among them, hydrological data includes runoff, evaporation, soil moisture, etc., and pollutant data includes the loss amounts of nitrogen and phosphorus, pesticide concentrations, etc. These output data will be used as the input for the subsequent LSTM model.

[0087] 3. Construction and training of the LSTM model

[0088] The LSTM (Long Short-Term Memory) model is used to process time series data and capture long-term dependencies therein. In the coupled SWAT-LSTM model, the LSTM model is used to further process the output of the SWAT model and other relevant time series data (such as meteorological data, agricultural management data, etc.), thereby improving the accuracy of pollutant concentration prediction.

[0089] The inputs to the LSTM model include: pollutant concentrations output by the SWAT model, time series data, and seasonal characteristics, etc. Among them, the pollutant concentrations output by the SWAT model include, for example, nitrogen and phosphorus concentrations, runoff, soil moisture, etc., the time series data includes, for example, meteorological data, fertilization amount, irrigation amount, etc., and the seasonal characteristics include months, quarters, etc. These data help the LSTM capture seasonal fluctuations.

[0090] 3.1 LSTM Model Architecture Design

[0091] Input layer: The input layer accepts multiple input features, mainly including historical pollutant concentrations, hydrological data, meteorological data, and agricultural management data output by the SWAT model.

[0092] LSTM hidden layer: The LSTM layer is used to learn long-term dependencies in the time series data. Multiple LSTM layers can be set, and the number of neurons in each layer is determined according to the complexity of the problem.

[0093] Fully connected layer: The output of the LSTM layer will be passed to the fully connected layer (Dense Layer) to generate the final prediction result through the fully connected layer.

[0094] Output layer: The output layer is used to predict the pollutant concentration at a future time step.

[0095] 3.2 LSTM Model Training

[0096] Loss function: Usually, the mean squared error (MSE) is used as the loss function, and the optimization goal is to minimize the difference between the predicted value and the actual value.

[0097] Optimization algorithm: Algorithms such as the Adam optimizer or RMSprop are used to optimize the training process of the LSTM model.

[0098] Hyperparameter tuning: The hyperparameters of the LSTM (such as learning rate, batch size, number of neurons in the hidden layer, etc.) are adjusted through grid search or random search to improve the model performance.

[0099] 3.3 LSTM Model Prediction

[0100] After the model training is completed, the LSTM model can be used to predict the future pollutant concentrations. Based on historical data, the LSTM model can predict the change trend of pollutant concentrations at a future moment.

[0101] 4. Integration of SWAT-LSTM Coupled Model

[0102] The coupling of the SWAT-LSTM model effectively combines the outputs of the SWAT model and the LSTM model to improve the prediction ability of pollutant concentrations.

[0103] 4.1 Data Flow and Coupling Method

[0104] The first stage (SWAT simulation): First, use the SWAT model to simulate the hydrological process and output the preliminary concentration data of pollutants (such as runoff, nitrogen and phosphorus loss, etc.).

[0105] The second stage (LSTM prediction): Input the data output by the SWAT model and other time series data into the LSTM model for time series prediction. Through learning historical data, the LSTM model can predict the change trend of pollutant concentrations in the future for a period of time.

[0106] 4.2 Coupling Process

[0107] Operation of the SWAT model: First, conduct SWAT model simulation to obtain data such as pollutant concentrations, surface runoff, and soil moisture.

[0108] Training of the LSTM model: Use historical data (including pollutant concentrations and hydrological data, meteorological data, etc. output by SWAT) to train the LSTM model to learn the patterns and long-term dependencies in the time series.

[0109] Joint optimization: Through joint training and feedback mechanisms, the output of the SWAT model can provide more accurate inputs for the LSTM model, and the output of the LSTM model can also provide references for subsequent simulations of the SWAT model.

[0110] After the model training is completed, the coupled model can also be used to predict agricultural non-point source pollution and monitor the changes in pollutant concentrations in real time.

[0111] Combining the advantages of the SWAT and LSTM models, the change trend of pollutant concentrations can be predicted at multiple time steps, so as to take corresponding environmental protection measures in advance.

[0112] The mean square error (MSE), root mean square error (RMSE), and R² value can also be used as evaluation indicators. By comparing with actual monitoring data, the prediction accuracy of the model can be evaluated, and the model can be further optimized according to the evaluation results.

[0113] In the exemplary embodiment, the second pollution monitoring parameter includes the soil nitrogen content and the rainfall probability in the area to be monitored within a set time period, and the first warning condition includes:

[0114] The soil nitrogen content exceeds a nitrogen content threshold, and the rainfall probability in the monitored area within a set time period exceeds a rainfall probability threshold.

[0115] In implementation, the nitrogen content threshold can be 150 mg / kg and the rainfall probability threshold can be 60%.

[0116] In an exemplary embodiment, the second pollution monitoring parameter includes the phosphorus content at the drain outlet of the monitored area and the downstream flow rate of the monitored area, and the second early warning condition includes:

[0117] The phosphorus content at the drainage outlet of the area to be monitored exceeds the phosphorus content threshold, and the downstream flow velocity of the area to be monitored is lower than the flow velocity threshold.

[0118] In implementation, the phosphorus content threshold may be 0.4 mg / L, and the flow velocity threshold of the downstream flow velocity may be 0.1 m / s.

[0119] In an exemplary embodiment, after a first-level warning is issued, water pollution control can also be carried out on rice fields, including pushing a "suspend drainage" command and starting a drone re-inspection. After a second-level warning is issued, the water conservancy facilities can be linked to close the gates, and a variable fertilization map can be generated. The fertilizer machine can be controlled through the agricultural machinery Internet of Things to reduce the application amount around the ridge by 20%, that is, to reduce fertilization.

[0120] In an exemplary embodiment, the downstream area of ​​the area to be monitored is determined by the following method:

[0121] Constructing a virtual river network model of the area to be monitored, wherein the virtual river network model at least reflects the geographic spatial data and hydrological data of the area to be monitored;

[0122] Based on the virtual river network model, determining the pollutant diffusion path of the area to be monitored;

[0123] Based on the pollutant diffusion path, a downstream area of ​​the area to be monitored is determined.

[0124] The virtual river network model reflects the path of pollutant diffusion by simulating river systems, watersheds and their hydrological processes. The following types of data are usually required to generate a virtual river network model:

[0125] a. Geospatial Data

[0126] Digital Elevation Model (DEM): Provides height information of the earth's surface and is used to generate topographic features of watersheds and river networks. DEM data helps determine the direction of water flow and the connectivity of rivers.

[0127] Geographical location of rivers and water bodies: including information such as coordinates, lengths, and watershed boundaries of rivers, lakes, and wetlands. River grid data (such as hydrological zoning data) in GIS (Geographic Information System) datasets can be used.

[0128] Land use / land cover data: These data reflect the land uses (such as agriculture, forest, urban, wetland, etc.) in different regions of the watershed, and have a direct impact on the distribution of pollution sources and the initial concentrations of pollutants.

[0129] b. Hydrological data

[0130] Precipitation data: Historical and real-time precipitation amounts, especially precipitation data in heavy rain scenarios, directly affect the water flow in the watershed and the diffusion of pollutants.

[0131] Flow data: The water flow rate and discharge in rivers, which are used to simulate the transport process of pollutants. The water discharge in different river sections of the watershed can be estimated based on historical hydrological data or real-time monitoring data.

[0132] c. Pollutant data

[0133] Pollutant concentration data: Including concentration data of pollutants such as nitrogen, phosphorus, and heavy metals in water bodies. These data can be collected in real time through devices such as soil sensors and water quality monitoring points.

[0134] Pollution source distribution data: Determine the specific locations of pollution sources (such as farmlands, industrial discharge outlets, etc.) and understand the intensities of pollution sources.

[0135] d. Hydrological models and meteorological data

[0136] SWAT model or similar hydrological models: These models can help calculate the water flow distribution in the watershed, the loss amounts of pollutants, and the hydrological responses of the watershed.

[0137] Meteorological data: Factors such as temperature, wind speed, and humidity have an impact on hydrological processes such as precipitation and evaporation, and can affect the water flow dynamics after heavy rain.

[0138] e. Channel attributes

[0139] Channel geometric parameters: Parameters such as channel width, depth, and slope can be used to estimate the water flow rate and water carrying capacity of rivers.

[0140] Hydraulic data: Including flow velocity, discharge, etc., which are used to calculate the diffusion of pollutants in water bodies.

[0141] The process of simulating the pollutant diffusion path in a heavy rain scenario using a virtual river network model involves the following steps:

[0142] a. Construction of the virtual river network

[0143] Terrain Modeling: Using DEM data, through the analysis of water flow direction, determine the location of each node (monitoring point or geographical unit) in the watershed and its adjacent nodes. Based on the precipitation and river flow characteristics of the watershed, establish a virtual river network.

[0144] Generation of Network Structure: According to the actual terrain and hydrological data of the watershed (such as watershed boundary, river path, etc.), construct a river network diagram. Nodes represent river reaches or water body units, and edges represent the direction and intensity of water flow.

[0145] b. Simulation of Precipitation and Water Flow

[0146] Precipitation Input: In the case of heavy rain, the precipitation increases sharply. Therefore, it is necessary to dynamically input the precipitation according to the heavy rain forecast data (such as precipitation intensity and precipitation distribution).

[0147] Water Flow Simulation: Through hydrological models (such as SWAT, HEC-HMS, etc.), simulate the surface water flow caused by precipitation, and determine the propagation path of pollutants in the water flow. According to parameters such as flow velocity, river channel slope, and water depth, calculate how pollutants diffuse from upstream to downstream.

[0148] c. Simulation of Pollutant Diffusion

[0149] Initial Concentration of Pollutants: According to the concentration data of pollution sources (such as agricultural fertilization, industrial emissions, etc.), set the initial pollutant concentration. Pollutants usually diffuse from the pollution source nodes to the surrounding water bodies or soils.

[0150] Pollutant Transport Process: In the virtual river network model, pollutants will spread with the movement of water flow. The diffusion of pollutants is related to factors such as the velocity of water flow, the geometric shape of the river channel, and the water body exchange rate.

[0151] River Channel Propagation: The diffusion of pollutants in water bodies follows the physical processes of dispersion, diffusion, and convection. In areas with higher water flow rates, pollutants diffuse faster, and vice versa.

[0152] Calculation of Pollutant Concentration: During the simulation process, the pollutant concentration at each node (such as river reach or water quality monitoring point) changes with time. The mass conservation equation and flow velocity equation can be used to calculate the concentration change of pollutants at each time step.

[0153] d. Specific Simulation Steps in the Case of Heavy Rain

[0154] Precipitation Input: During heavy rain, the precipitation increases significantly, resulting in an increase in surface runoff. Therefore, input the precipitation and precipitation intensity of heavy rain into the model.

[0155] Dynamic Changes in Water Flow: As precipitation increases, the water flow rate also increases, thus affecting the propagation path and rate of pollutants.

[0156] Pollutant diffusion path tracking: Based on the dynamic changes of water flow, simulate how pollutants diffuse with the water flow to different river sections and water bodies. In some cases, pollutants may enter tributaries or downstream lakes and other water bodies.

[0157] Pollutant concentration prediction: Use a time-step-based model to predict the changes in pollutant concentration at each node to identify hotspots of pollutant diffusion.

[0158] e. Early warning and response

[0159] Based on the propagation path and concentration of pollutants, an early warning system can be set up at key nodes (such as water quality monitoring points, drainage outlets, etc.) to monitor the pollutant diffusion situation in real time and generate emergency response measures (such as closing drainage outlets, starting ecological restoration, etc.) according to the pollution diffusion path and concentration.

[0160] In this embodiment, determining the pollutant diffusion path has the following beneficial effects:

[0161] 1. Identify pollution sources and pollution diffusion pathways

[0162] Pollution source location: By simulating the diffusion path of pollutants, the source and diffusion direction of pollutants can be accurately identified, helping to determine the location of pollution sources (such as farmland, industrial discharge outlets, etc.).

[0163] Analysis of pollution propagation pathways: By simulating the diffusion paths of pollutants in different scenarios, it is possible to reveal how pollutants spread from the source to other areas (such as rivers, lakes, groundwater, etc.) and predict the flow and diffusion trends of pollutants.

[0164] 2. Predict the spatial and temporal distribution of pollutants

[0165] Spatial distribution: Simulating the pollutant diffusion path can help predict the concentration changes of pollutants at different geographical locations (such as river sections, water bodies, soil, etc.), so as to understand the spatial distribution of pollutants in the basin.

[0166] Temporal variation: By simulating the changes of pollutants over time, it is possible to predict the dynamic changes of pollutant concentration in the future for a period of time, providing a time reference for pollution prevention and control.

[0167] 3. Evaluate the impact of pollutants on the environment and ecosystem

[0168] Impact assessment: Simulating the pollutant diffusion path helps to evaluate the potential impact of pollutants on the environment, ecosystem and water quality. For example, by predicting whether pollutants will enter drinking water sources, wetlands or ecologically sensitive areas, the threats to biodiversity, water quality and ecosystem functions can be evaluated.

[0169] Risk assessment: By evaluating the diffusion path of pollutants, it is possible to quantify the pollution risk of pollutants to downstream areas, water bodies, and ecosystems, providing a scientific basis for management departments to take corresponding prevention and control measures.

[0170] 4. Optimize water resource management and pollution prevention

[0171] Early warning and emergency response: By simulating the diffusion path of pollutants, early warning information can be provided for pollution incidents. Especially under extreme weather conditions such as heavy rain, the prediction of the pollutant diffusion path can help take emergency response measures in advance, such as closing drainage outlets, strengthening water quality monitoring, or starting water treatment systems.

[0172] Formulation of treatment strategies: Simulating the diffusion path of pollutants helps to design and optimize pollution prevention and control measures. For example, understanding the propagation path of pollutants from the source to the target area can help formulate reasonable water quality purification and pollutant treatment strategies (such as ecological restoration, sewage treatment, source control, etc.).

[0173] 5. Support decision-making and policy implementation

[0174] Decision support: Simulating the pollutant diffusion path can provide decision support for the government, environmental management agencies, scientific research personnel, etc. By understanding the path and trend of pollutant diffusion, relevant departments can make reasonable environmental protection policies and measures to promote pollution control.

[0175] Monitoring and supervision: Through the simulation of the pollutant diffusion path, the government can better monitor the emissions of pollution sources to ensure the effective implementation of relevant regulations and policies.

[0176] 6. Improve the ability to trace pollutants

[0177] Pollution source tracing: In some pollution accidents or environmental pollution incidents, by simulating the pollutant diffusion path, the source and propagation path of pollutants can be traced. This is very important for accountability, environmental restoration, and the formulation of remedial measures.

[0178] In an exemplary embodiment, calculating the pollution contribution rate of the area to be monitored to the downstream area by the graph neural network includes:

[0179] Calculating the concentration of pollutants in the area to be monitored by the graph neural network, and calculating the concentration of pollutants in the downstream area of the area to be monitored, where the pollutants include nitrogen and phosphorus;

[0180] Determining the pollution contribution rate of the area to be monitored to the downstream area based on the pollutant concentration in the area to be monitored and the pollutant concentration in the downstream area of the area to be monitored.

[0181] In an exemplary embodiment, the graph neural network is constructed by the following method:

[0182] Taking the area to be monitored as graph nodes and the rivers between the areas to be monitored as edges, a graph structure is constructed;

[0183] The graph structure is iteratively trained to optimize the weights of the edges in the graph structure, and the graph neural network is obtained.

[0184] Figure 3 It is a schematic structural diagram of the graph neural network model provided by the embodiments of the present invention.

[0185] As Figure 3 shown, specifically, when establishing the graph structure, each monitoring point can be defined as a node , and each hydrological connectivity is an edge . The weight of the edge represents the intensity of hydrological flow or the ability of pollution propagation during the process of pollutants flowing from node to node .

[0186] After the graph structure is constructed, the training mechanism of the graph neural network (such as gradient descent, etc.) can be used to optimize the edge weights so that the model can learn appropriate pollution propagation patterns and accurately calculate the pollution contribution rate.

[0187] In the GNN model, information is passed through the neighbor nodes of the graph. Each node updates its own state through the information of its neighbor nodes, and the update rule of the information is usually expressed as:

[0188]

[0189] is the hidden state of node after the j-th iteration (which can be understood as the feature representation of the node), is the weight of edge (i.e., hydrological connectivity), is the set of neighbor nodes of node , is the activation function (such as ReLU), and b is the bias term.

[0190] The initial pollution amount can be initialized based on actual pollution concentration data.

[0191] At each moment t, the transfer of the pollution amount is based on the pollution amounts of the neighbor nodes and the edge weights. The specific pollution propagation formula is:

[0192]

[0193] Among them, is the external pollution source of the node .

[0194] In practical applications, in order to calculate the pollution contribution rate, the contribution of each monitoring point can be reflected based on the final concentration of pollutants. Assume that the target pollution point is the node . The pollution contribution of each node to the node can be calculated through the following steps:

[0195] Pollution contribution rate can be defined as the contribution ratio of the node to the pollution concentration of the node . The specific calculation method is as follows:

[0196] Among them: is the pollution concentration of the node at the final moment T, is the pollution concentration of the target pollution point at the final moment T.

[0197] Next, the agricultural non-point source pollution early warning system provided by the present invention will be described. The agricultural non-point source pollution early warning system described below can be mutually referred to the agricultural non-point source pollution early warning method described above.

[0198] Figure 4 is one of the structural schematic diagrams of the agricultural non-point source pollution early warning system provided by the embodiments of the present invention.

[0199] Figure 5 is another structural schematic diagram of the agricultural non-point source pollution early warning system provided by the embodiments of the present invention.

[0200] As Figure 4 and Figure 5 shown, the agricultural non-point source pollution early warning system provided in this embodiment includes:

[0201] A parameter acquisition module 401, configured to acquire first pollution monitoring parameters of the area to be monitored;

[0202] A parameter prediction module 402, configured to predict second pollution monitoring parameters of the area to be monitored within a set time period based on the first pollution monitoring parameters through a pre-constructed prediction model;

[0203] A first early warning module 403, configured to perform a first-level early warning when the second pollution monitoring parameters meet the first early warning condition, and perform a second-level early warning when the second pollution monitoring parameters do not meet the first early warning condition but meet the second early warning condition;

[0204] The second warning module 404 is used to calculate the pollution contribution rate of the area to be monitored to the downstream area through a graph neural network when the second pollution monitoring parameter does not meet the first warning condition and the second warning condition. If the pollution contribution rate meets the third warning condition, a third-level warning is issued.

[0205] As Figure 5 shown, the agricultural non-point source pollution warning system provided in this embodiment can collect the required first pollution monitoring parameters through the data collection layer. Specifically, the parameters can be collected through an aerial remote sensing platform and a ground sensing network and sent to the data collection layer. The data of the data collection layer is then sent to the cloud platform layer, where multi-source data fusion is performed on all the data. After that, data analysis and warning are carried out in the data analysis module. The data analysis module can include the above-mentioned parameter acquisition module 401, parameter prediction module 402, first warning module 403, and second warning module 404. After data processing by the above data analysis module, a warning result can be obtained. The various warning results obtained can be sent to the decision support interface of the warning system. The decision support interface can be displayed using a display screen. On the decision support interface, on the one hand, the warning signals of the first-level warning, second-level warning, and third-level warning can be displayed. In addition, a pollution heat map can be generated according to the above-mentioned collected first pollution monitoring parameters. The pollution heat map can more intuitively reflect the pollution severity of each area. Moreover, after the warning is issued, certain treatment countermeasures can be provided according to the pollution severity. For example, the amount of fertilization can be adjusted, that is, a variable fertilization prescription map can be provided.

[0206] Figure 6 An example of the physical structure diagram of an electronic device is shown, as Figure 6 shown, the electronic device may include: a processor 610, a communication interface 620, a memory 630, and a communication bus 640. Among them, the processor 610, the communication interface 620, and the memory 630 complete mutual communication through the communication bus 640. The processor 610 can call the logical instructions in the memory 630 to execute the agricultural non-point source pollution warning method, which includes:

[0207] Obtain the first pollution monitoring parameter of the area to be monitored;

[0208] Based on the first pollution monitoring parameter, predict the second pollution monitoring parameter of the area to be monitored within a set time period through a pre-constructed prediction model;

[0209] When the second pollution monitoring parameter meets the first warning condition, a first-level warning is issued. When the second pollution monitoring parameter does not meet the first warning condition but meets the second warning condition, a second-level warning is issued;

[0210] When the second pollution monitoring parameter does not meet the first warning condition and the second warning condition, calculate the pollution contribution rate of the area to be monitored to the downstream area through a graph neural network. If the pollution contribution rate meets the third warning condition, issue a third-level warning.

[0211] In addition, when the logical instructions in the above-mentioned memory 630 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this 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 for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0212] On the other hand, the present invention also provides a computer program product. The computer program product includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the agricultural non-point source pollution warning method provided by the above-mentioned various methods. The method includes:

[0213] Obtain the first pollution monitoring parameter of the area to be monitored;

[0214] Based on the first pollution monitoring parameter, predict the second pollution monitoring parameter of the area to be monitored within a set time period through a pre-constructed prediction model;

[0215] When the second pollution monitoring parameter meets the first warning condition, issue a first-level warning. When the second pollution monitoring parameter does not meet the first warning condition but meets the second warning condition, issue a second-level warning;

[0216] When the second pollution monitoring parameter does not meet the first warning condition and the second warning condition, calculate the pollution contribution rate of the area to be monitored to the downstream area through a graph neural network. If the pollution contribution rate meets the third warning condition, issue a third-level warning.

[0217] On the other hand, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is used to execute the agricultural non-point source pollution warning method provided by the above-mentioned various methods. The method includes:

[0218] Obtain the first pollution monitoring parameter of the area to be monitored;

[0219] Based on the first pollution monitoring parameter, predict the second pollution monitoring parameter of the area to be monitored within a set time period through a pre-constructed prediction model;

[0220] When the second pollution monitoring parameter meets the first warning condition, issue a first-level warning. When the second pollution monitoring parameter does not meet the first warning condition but meets the second warning condition, issue a second-level warning;

[0221] When the second pollution monitoring parameter does not meet the first warning condition and the second warning condition, calculate the pollution contribution rate of the area to be monitored to the downstream area through a graph neural network. If the pollution contribution rate meets the third warning condition, issue a third-level warning.

[0222] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative effort.

[0223] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions in essence or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of each embodiment or some parts of the embodiments.

[0224] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present invention.

Claims

1. Agricultural non-point source pollution early warning method, characterized in that, Including: Obtain the first pollution monitoring parameter of the area to be monitored; Based on the first pollution monitoring parameter, predict the second pollution monitoring parameter of the area to be monitored within a set time period through a pre-constructed prediction model; When the second pollution monitoring parameter meets the first warning condition, issue a first-level warning. When the second pollution monitoring parameter does not meet the first warning condition but meets the second warning condition, issue a second-level warning; When the second pollution monitoring parameter does not meet the first warning condition and the second warning condition, calculate the pollution contribution rate of the area to be monitored to the downstream area through a graph neural network. If the pollution contribution rate meets the third warning condition, issue a third-level warning; The second pollution monitoring parameter includes the soil nitrogen content and the rainfall probability of the area to be monitored within a set time period. The first warning condition includes: The soil nitrogen content exceeds the nitrogen content threshold, and the rainfall probability of the area to be monitored within a set time period exceeds the rainfall probability threshold; The second pollution monitoring parameter includes the phosphorus content of the drainage outlet of the area to be monitored and the downstream flow velocity of the area to be monitored. The second warning condition includes: The phosphorus content of the drainage outlet of the area to be monitored exceeds the phosphorus content threshold, and the downstream flow velocity of the area to be monitored is lower than the flow velocity threshold; The downstream area of the area to be monitored is determined by the following method: Construct a virtual river network model of the area to be monitored, and the virtual river network model at least reflects the geographical spatial data and hydrological data of the area to be monitored; Based on the virtual river network model, determine the pollutant diffusion path of the area to be monitored, including the simulation of precipitation and water flow, the simulation of pollutant diffusion, and the simulation under rainstorm scenarios; Based on the pollutant diffusion path, determine the downstream area of the area to be monitored; The calculation of the pollution contribution rate of the area to be monitored to the downstream area through the graph neural network includes: Calculate the concentration of pollutants in the area to be monitored through the graph neural network, and calculate the concentration of pollutants in the downstream area of the area to be monitored. The pollutants include nitrogen and phosphorus; Based on the pollutant concentration in the area to be monitored and the pollutant concentration in the downstream area of the area to be monitored, determine the pollution contribution rate of the area to be monitored to the downstream area; The graph neural network is constructed by the following method: Taking the area to be monitored as graph nodes and the rivers between the areas to be monitored as edges, construct a graph structure; Iteratively train the graph structure to optimize the weights of the edges in the graph structure to obtain the graph neural network.

2. The agricultural non-point source pollution early warning method according to claim 1, wherein The prediction model is a coupled model of a soil and water assessment model and a long short-term memory network model.

3. The agricultural non-point source pollution early warning method according to claim 2, characterized in that The construction process of the prediction model includes: Collect training data, and the training data includes the historical pollution monitoring parameters of the area to be monitored; Based on the training data, train the soil and water assessment model and the long short-term memory network model respectively; Use the output of the soil and water assessment model as the input of the long short-term memory network model, iteratively optimize the long short-term memory network model, and iteratively optimize the soil and water assessment model based on the output of the long short-term memory network model.

4. An agricultural non-point source pollution early warning system, which is applied to the agricultural non-point source pollution early warning method described in any one of claims 1-3, and is characterized in that, It includes: A parameter acquisition module for acquiring the first pollution monitoring parameter of the area to be monitored; A parameter prediction module for predicting the second pollution monitoring parameter of the area to be monitored within a set time period through a pre-constructed prediction model based on the first pollution monitoring parameter; A first warning module for giving a first-level warning when the second pollution monitoring parameter meets the first warning condition, and giving a second-level warning when the second pollution monitoring parameter does not meet the first warning condition but meets the second warning condition; A second warning module for calculating the pollution contribution rate of the area to be monitored to the downstream area through a graph neural network when the second pollution monitoring parameter does not meet the first warning condition and the second warning condition, and giving a third-level warning if the pollution contribution rate meets the third warning condition.

5. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the agricultural non-point source pollution warning method described in any one of claims 1-3.

Images