Supervision system and method for agricultural non-point source pollution scenarios

Abstract

The present application relates to the field of environmental monitoring and governance, and discloses a supervision system and method for agricultural non-point source pollution scenarios. The system includes a perception module, a data management module, an intelligent analysis module, and an application service module. The perception module collects pollutant concentration, hydrological characteristics, and abnormal behavior data through water quality sensors, high-position video monitoring equipment, and drones; the data management module dynamically integrates, stores, and standardizes the data collected by the perception module; the intelligent analysis module predicts the diffusion path of pollutants in the water body based on the pollution diffusion model, and inverts the spatial location and intensity of the pollution source by combining the video monitoring abnormal behavior records with the diffusion model; the application service module displays the analysis results, triggers warnings, and supports remote governance operations through an intelligent early warning system, a GIS visualization platform, and mobile devices. The present invention can accurately monitor agricultural non-point source pollution, improve the efficiency of pollution source positioning and supervision, and provide technical support for the governance of agricultural non-point source pollution.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental monitoring and governance, and in particular to a supervision system and method for agricultural non-point source pollution scenarios. Background Art

[0003] Traditional methods of monitoring agricultural non-point source pollution usually rely on a combination of manual inspections and fixed monitoring equipment. For example, fixed monitoring points are set up around the river basin or water body to conduct long-term monitoring of water quality parameters such as chemical oxygen demand, ammonia nitrogen, total phosphorus, etc., and regular manual inspections are conducted to confirm pollution behavior.

[0004] However, this approach has the following major problems:

[0005] 1. Limited monitoring scope: Traditional monitoring points are usually deployed at a low density and can only cover some key areas, making it difficult to fully grasp the situation of scattered agricultural pollution point sources, especially in remote areas or hidden pollution discharge behaviors that are difficult to detect.

[0006] 2. Data uniformity: Traditional monitoring is mainly based on water quality monitoring data, and lacks real-time recording and analysis of abnormal behaviors of pollution sources (such as illegal discharge of tail water and sewage discharge through secret pipes), making it difficult to confirm the source of pollutants.

[0007] 3. Response lag: Fixed monitoring equipment usually relies on long-term sampling, and manual investigation and tracing are often required after pollution occurs, resulting in a longer regulatory response time.

[0008] 4. Difficulty in obtaining evidence: Traditional methods lack effective means of obtaining evidence for illegal discharge at night or other concealed discharges. It is difficult to obtain conclusive evidence of illegal behavior by relying solely on manual inspections.

[0009] In recent years, with the increasing pressure on water environment governance, traditional monitoring methods have been unable to meet the needs of precise supervision of complex agricultural non-point source pollution. Some advanced technologies such as drones, video surveillance, and pollution diffusion models have begun to be gradually applied to environmental monitoring, but most of these technologies are applied independently and lack the ability to integrate and comprehensively analyze multi-source data, and cannot form a complete closed loop from data collection, analysis to governance decision-making. Summary of the invention

[0010] In view of the shortcomings of the prior art, the present invention provides a supervision system and method for agricultural non-point source pollution scenarios, which can realize multi-dimensional dynamic monitoring of agricultural non-point source pollution, pollution diffusion path modeling, precise tracing of pollution sources and rapid positioning of pollution discharge behaviors, thereby improving the efficiency and accuracy of pollution source supervision.

[0011] To achieve the above objectives, the present invention is implemented through the following technical solutions: A monitoring system for agricultural non-point source pollution scenes, comprising:

[0012] The perception module is used to collect pollutant concentration, hydrological characteristics and abnormal behavior data through water quality sensors, high-position video monitoring equipment and drones, and provide support for the initial source location of abnormal water behavior;

[0013] The data management module is used to dynamically integrate, store and standardize the data collected by the perception module;

[0014] Intelligent analysis module, which is used to predict the diffusion path of pollutants in the river based on the pollutant diffusion model, and invert the spatial location and intensity of the pollution source by combining the abnormal behavior records of video surveillance with the diffusion model;

[0015] The application service module is used to display analysis results, trigger warnings and support remote governance operations through intelligent early warning systems, data visualization platforms and mobile devices.

[0016] Preferably, the perception module includes:

[0017] Water quality sensors are used to monitor the concentrations of chemical oxygen demand, ammonia nitrogen and total phosphorus, and record the temporal and spatial distribution of water quality anomalies in real time, providing basic data support for pollution diffusion models;

[0018] High-level video surveillance equipment is used to monitor multiple monitoring points in real time. It uses image recognition algorithms to detect abnormal behaviors, including water level changes, fish catching activities, and suspected sewage discharge. It also works in conjunction with diffusion models to locate suspected pollution sources in combination with water quality abnormality data and provide preliminary clues.

[0019] Drones are used for autonomous flight control to locate suspicious ponds or remote areas, and are equipped with thermal imaging equipment and multi-spectral sensors to monitor sewage discharge at night or in hidden areas.

[0020] Preferably, the data management module includes:

[0021] Dynamic integration module, used to connect multi-source data collected by water quality sensors, video surveillance equipment and drones to a shared data center through a unified interface;

[0022] A data storage module, used to store data in blocks according to pollution areas and time indexes;

[0023] The data preprocessing module is used to remove anomalies, denoise and standardize the format of the collected data.

[0024] Preferably, the intelligent analysis module includes:

[0025] The pollution diffusion modeling module is used to dynamically build a pollutant diffusion model based on river flow velocity, pollution diffusion coefficient and water quality sensor data to predict the temporal and spatial distribution of pollutants in the water body;

[0026] The pollution source tracing module preliminarily inverts the spatial location and emission intensity of pollution sources by analyzing the differences between monitoring data and diffusion model simulation results;

[0027] The emission location positioning module is used to determine the specific location of pollution emissions by combining the pollution diffusion modeling results, pollution source tracing results and abnormal behavior records of high-position video surveillance.

[0028] Preferably, the pollution diffusion modeling module uses a convection-diffusion model to dynamically simulate the spatiotemporal distribution of pollutants based on the diffusion characteristics of pollutants in water bodies and flow velocity field information, including:

[0029] The two-dimensional convection-diffusion equation is used to describe the dynamic changes of pollutant concentrations, taking into account the comprehensive effects of flow velocity components, water diffusion coefficients and pollution source intensity on the changes of pollutant concentrations;

[0030] The pollutant concentration C(x,y,t) is taken as the target variable, the velocity components u,v, and the diffusion coefficient D x ,D y And the pollution source intensity S(x,y,t) are input parameters, satisfying the following relationship:

[0031]

[0032] The numerical method is used to discretize and solve the equations, dynamically simulating the diffusion path and concentration distribution of pollutants from the pond to the main road;

[0033] Combine the pollutant concentration data from the water quality sensor and the real-time flow rate data from the flow rate sensor to regularly update the model parameters;

[0034] The diffusion simulation results are combined with the temporal and spatial information of abnormal behaviors recorded by high-position video surveillance to locate the scope of the pollution source and provide data support for subsequent source tracing analysis.

[0035] Preferably, the pollution source tracing module inverts the spatial distribution and intensity of pollution sources through the difference between pollutant diffusion modeling results and observation data. The pollution source tracing process includes the following steps:

[0036] The pollutant concentration observation data C collected by the perception module obs (x, y, t) and the concentration distribution C obtained by simulation based on pollution diffusion modeling sim (x, y, t; S) to evaluate the deviation between the model prediction and the actual monitoring data;

[0037] A source tracing model with pollution source intensity S(x,y,t) as the optimization variable is constructed to minimize the error between observed data and simulated data, satisfying the following optimization relationship:

[0038] min s(x,y,t) ∫ Ω [C obs (x,y,t)-C sim (x,y,t;S)] 2 dxdy

[0039] The gradient descent method is used to iteratively optimize the pollution source intensity S(x,y,t), and the model error is gradually reduced through the following update formula:

[0040]

[0041] Among them, J(S) is the optimization objective function, α is the learning rate;

[0042] Combined with abnormal behavior data recorded by high-position video surveillance, the traceability results are verified and corrected, and the pollution intensity and time characteristics are matched with the time points of abnormal behavior to further narrow the spatial scope of the pollution source;

[0043] Finally determine the spatial location and emission intensity of the pollution source.

[0044] Preferably, the step of determining the specific location of pollution emission by the emission location positioning module includes:

[0045] Combine the pollutant diffusion path and pollution concentration distribution data generated by the pollution diffusion modeling module, and the candidate locations of pollution sources inverted by the pollution source tracing module;

[0046] Retrieve the time and location of abnormal behavior recorded by high-position video surveillance, and match them with the diffusion time point and path calculated by the diffusion model;

[0047] Prioritize screening the areas in the video where abnormal behavior occurs and is consistent with abnormal water quality data, and ultimately identify the specific location and time of pollution discharge.

[0048] Preferably, the application service module includes:

[0049] Intelligent early warning module, which is used to automatically trigger an alarm for exceeding the standard based on the monitoring data of water quality sensors, the prediction results of the pollution diffusion model and the analysis results of the emission location positioning module, and push detailed information of the abnormal area, including the location and time of possible pollution sources;

[0050] Visualization module, used to dynamically display the pollutant diffusion path, source tracing analysis results and the final emission location of the emission location location module on the GIS map; supports users to view the diffusion prediction and pollution source impact range in different time periods;

[0051] The mobile operation module is used to support remote drone dispatch and real-time viewing of video surveillance, pollution concentration data and positioning results, providing a convenient mobile supervision tool and allowing managers to directly mark and record on-site inspection information.

[0052] Preferably, the visualization module provides dynamic display and prediction functions of pollution diffusion paths on the GIS map based on pollutant concentration data, dynamic pollution source locations and high-level video monitoring records stored in the time series database, and supports concentration change analysis across time dimensions.

[0053] The present invention also provides a method for supervising agricultural non-point source pollution scenes, comprising the following steps:

[0054] Real-time collection of agricultural non-point source pollution data, including pollutant concentrations monitored by water quality sensors, abnormal behavior data recorded by high-position video surveillance, and on-site image information collected by drones;

[0055] Dynamically integrate and preprocess the collected data, including abnormal data removal and format standardization;

[0056] Based on the river flow velocity, diffusion coefficient and pollutant concentration data, a pollution diffusion model is constructed to dynamically simulate the diffusion path and concentration distribution of pollutants in the water body;

[0057] Using the difference between the diffusion model simulation results and the water quality monitoring data, the spatial location and emission intensity of the pollution source are inverted through the source tracing optimization model;

[0058] The specific location and time of pollution emission can be accurately determined by combining diffusion model prediction, source tracing analysis results and abnormal behavior data recorded by high-level video surveillance through time and space matching;

[0059] Intelligent warning is triggered based on concentration thresholds, and abnormal area information is generated in combination with emission location positioning results;

[0060] Dynamically display pollution diffusion paths, traceability results and positioning analysis on GIS maps to support remote scheduling and decision-making on the management side.

[0061] The present invention provides a monitoring system and method for agricultural non-point source pollution scenarios, which has the following beneficial effects:

[0062] 1. The present invention combines pollution diffusion modeling with source tracing optimization, integrates abnormal behavior records of high-level video surveillance and real-time data of water quality sensors, and accurately determines the spatial location and emission intensity of pollution sources. Through cross-validation of multi-dimensional data, the reliability and accuracy of positioning results are greatly improved, avoiding misjudgment caused by a single data source in traditional methods.

[0063] 2. Through multi-point monitoring, efficient data integration and intelligent analysis of the perception module, the present invention realizes a complete closed-loop monitoring process from data collection, dynamic modeling, source tracing analysis to precise positioning of the discharge site. In practical applications, for example, when monitoring sewage discharge in fish ponds, the system first detects water quality abnormalities through sensors, then predicts the pollution diffusion path in combination with the diffusion model, and matches abnormal behaviors through high-position video, and finally accurately locks the sewage discharge fish pond. This process significantly improves the monitoring efficiency and the accuracy of pollution source tracking.

[0064] 3. The present invention relies on intelligent early warning and can dynamically generate early warning information based on the real-time monitoring data of water quality sensors and the prediction results of diffusion models. The system realizes automatic alarm of abnormal behavior and timely notification of excessive pollutant concentrations, providing technical support for rapid response for management personnel and reducing the workload of manual investigation.

[0065] 4. The present invention dynamically displays the pollution diffusion path, source tracing results and final discharge location through GIS maps, so that managers can intuitively understand the diffusion trend and impact range of pollutants in water bodies. Especially in large-scale monitoring scenarios, the spatiotemporal dynamic display function provided by the visualization module provides a clear reference for accurately dispatching drones for evidence collection and formulating governance plans.

[0066] 5. The present invention is applicable to fish ponds, farmland drainage and widely distributed agricultural non-point source pollution scenes. By adjusting the sensor layout density and modeling parameters, it can flexibly adapt to the regulatory needs of different terrains, water system layouts and pollution types. This scene adaptability makes the present invention have a wide range of promotion value in the field of agricultural non-point source pollution control.

[0067] 6. Through the collaborative work of multiple devices, the present invention can not only dynamically monitor the spread of pollutants, but also accurately collect evidence and record on-site through drones. In the monitoring process, based on the traceability and positioning results, the system dispatches drones to fly to suspected pollution discharge points for evidence collection, and further verifies the behavioral characteristics in combination with high-position videos. This organic combination of scientific monitoring and evidence collection not only ensures the closed-loop nature of the monitoring process, but also provides strong evidence support for law enforcement. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is a schematic diagram of the system architecture of the present invention;

[0069] Figure 2 It is a schematic diagram of the method flow of the present invention;

[0070] Figure 3 Schematic diagram of the application of an embodiment of the present invention. DETAILED DESCRIPTION

[0071] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0072] Please see attached Figure 1 The present invention provides a monitoring system for agricultural non-point source pollution scenarios, which uses a variety of perception means and intelligent analysis methods to achieve efficient monitoring, positioning and governance support for agricultural non-point source pollution. The overall system architecture includes a perception module, a data management module, an intelligent analysis module and an application service module, which work together to achieve accurate monitoring and governance of pollution sources. The various modules of the system of the present invention are described in detail below.

[0073] Perception Module

[0074] The perception module is used to collect real-time water quality monitoring data, abnormal behavior data, and on-site pollution characteristic data of the river and surrounding areas. Through the collaborative work of various perception devices in this module, it provides support for subsequent data analysis modules and application service modules. The realization of the perception module combines water quality sensors, high-position video monitoring equipment, and drone monitoring equipment to ensure comprehensive perception and preliminary positioning of pollution sources.

[0075] Generally, the perception module forms a perception network that dynamically covers the entire monitoring area by deploying perception devices at multiple points, and obtains multi-dimensional data related to pollution in real time. As an option, various devices in the perception module can upload data to the data management module using wireless networks or Internet of Things protocols to ensure real-time and reliability.

[0076] In this embodiment, the perception module includes a water quality sensor, a high-position video monitoring device and an unmanned aerial vehicle monitoring device. Each device is connected to the data management module through the Internet of Things technology to collect pollution data of different dimensions.

[0077] Water quality sensor

[0078] In this implementation, water quality sensors are deployed in high-risk agricultural pollution areas such as rivers and fish ponds to collect key water quality parameters in real time, such as chemical oxygen demand (COD), ammonia nitrogen concentration (NH3-N) and total phosphorus concentration (TP). These indicators are important parameters for evaluating the degree of water pollution and the emission intensity of pollution sources.

[0079] As an option, water quality sensors detect pollutant concentrations through electrochemical or optical absorption methods. For example:

[0080] COD can be detected by ultraviolet absorption method, and the pollution concentration is calculated based on the intensity of ultraviolet light absorption by pollutants in water samples.

[0081] NH3-N can be detected by using the ion selective electrode method to measure the concentration of ammonium ions in the solution.

[0082] TP can be detected by measuring the total phosphorus content in water within a specific wavelength range using a spectrophotometer.

[0083] In some embodiments, the water quality sensor is implemented based on optical, electrochemical or ion selective electrode technology. Among them, the monitoring of chemical oxygen demand can be completed by ultraviolet absorption spectroscopy analysis. The sensor calculates the degree of organic pollution by measuring the absorbance at ultraviolet wavelengths.

[0084] Generally, total phosphorus is monitored using chemical colorimetry. The sensor mixes the sample with the reagent, measures the absorbance at a specific wavelength, and ultimately obtains the total phosphorus concentration.

[0085] In a possible implementation, the monitoring data of the water quality sensor is uploaded to the data management module in the form of a time series. The output data of each sensor can be expressed as:

[0086] C i (t) = {COD i (t),NH3-N i (t),TP i (t)}

[0087] Among them, C i (t) is the pollution concentration data collected by sensor i at time t, and the data upload cycle is once per minute.

[0088] In some possible implementations, water quality sensors are distributed at multiple key nodes of the main river, and their spacing is determined according to the flow velocity and diffusion characteristics of the river to ensure the spatial continuity of the monitoring data. The monitoring data of the sensor is uploaded to the data management module in the form of a time series, and the parameters of the pollution diffusion model are dynamically updated in combination with the data of the flow velocity sensor in the sensing area.

[0089] High-position video surveillance equipment

[0090] In this embodiment, high-position video surveillance equipment is installed at commanding heights in the basin or at key locations along the river, and is mainly used to detect abnormal behaviors through image recognition algorithms, such as hidden drainage of sewage pipes, suspicious activities of personnel, and abnormal changes in water levels.

[0091] In some implementations, the video surveillance device uses an embedded artificial intelligence processing unit to perform real-time analysis to identify dynamic features in the image. Specifically, by performing pixel-level change analysis on the boundary of a water body in a video stream, a rapid drop in water level can be detected; at the same time, human activities such as catching fish can be identified through a human behavior feature model.

[0092] As an option, high-position video surveillance equipment combines target detection algorithms to analyze video data in real time. Specifically:

[0093] A convolutional neural network (CNN) model is used to detect abnormal targets in video frames, such as floating pollutants, drainage equipment, and abnormal water level changes.

[0094] The input of the algorithm is the image feature matrix F i (t), the output is the target category and location information in the detection result.

[0095] In one possible implementation, the monitoring device uses the YOLOv5 target detection algorithm, and the model extracts features from the input frame and identifies specific target behaviors. The target detection results are uploaded to the data management module through the IoT interface.

[0096] For example, when pumping equipment is detected near a river, the system will extract the target bounding box information as an alarm event and record the relevant time and location data.

[0097] As an option, high-position video surveillance equipment supports nighttime monitoring and can be equipped with infrared cameras to capture abnormal behavior in low-light environments.

[0098] Generally, high-position video surveillance equipment stores the timestamp and location data of abnormal behavior as log files for subsequent intelligent analysis modules to perform time and space matching. In addition, the equipment supports linkage with the pollution diffusion model, and the location of abnormal behavior can be used as the preliminary candidate point of the pollution source of the model.

[0099] Drone monitoring equipment

[0100] In some embodiments, drone monitoring equipment is used to cover remote areas that cannot be reached by high-position video surveillance, especially at night and in adverse weather conditions, using thermal imaging equipment to achieve dynamic monitoring.

[0101] Specifically, the drone carries the following monitoring equipment:

[0102] Thermal imagers are used to detect temperature anomalies generated by tailwater discharge.

[0103] High-definition cameras to record detailed images of potential contamination areas.

[0104] Generally, drones patrol along a preset route. In one possible implementation, when a high-position video surveillance device or sensor detects an abnormal event, the drone can receive instructions to fly to a specific area to perform fixed-point monitoring.

[0105] The data output of drone monitoring can be expressed as:

[0106] T i (t) = {heat source matrix, image features}

[0107] Among them, the heat source matrix is ​​used to locate the thermal anomaly area of ​​the tailwater discharge point, and the image features are used for subsequent target identification.

[0108] As an option, the drone can automatically return to the cabin to recharge when the battery is low, while uploading all the data collected during the monitoring period.

[0109] Data upload and transmission

[0110] Generally, the perception module uploads the collected data to the data management module in real time through the Internet of Things technology. In this embodiment, the MQTT protocol is used to achieve efficient data transmission, ensuring the real-time and integrity of the data of sensors, high-position video surveillance and drone monitoring equipment.

[0111] Specifically, the uploaded multi-source data includes:

[0112] Sensor Data C i (t), record the time series of pollutant concentration;

[0113] Video data F i (t), contains the target detection results;

[0114] Drone DataT i (t), contains thermal imaging features and high-definition images.

[0115] In one possible implementation, the sensing module can be integrated with rainfall monitoring equipment to assess the impact of precipitation on pollutant dispersion. For example, precipitation data combined with dynamic changes in water quality parameters can help predict the behavior of pollutants spreading with rainwater runoff.

[0116] In this embodiment, the water quality sensors, high-position video surveillance equipment and drones in the perception module are seamlessly integrated through the Internet of Things protocol to form a multimodal perception network covering the entire area.

[0117] In some implementations, the concentration anomaly data provided by the water quality sensor is linked to the time and location recorded by the high-position video to verify the association between water anomalies and possible pollution discharge behaviors. At the same time, the on-site survey by drones further confirms the pollution characteristics of the abnormal area, providing key support for the final location of the pollution discharge site.

[0118] As an option, the perception module also supports dynamic adjustment of the sampling frequency of sensors and monitoring devices. For example, after detecting excessive concentrations or abnormal behavior, the sampling frequency and video capture rate can be increased to enhance the timeliness and resolution of the data.

[0119] Through the above design, the perception module realizes multi-dimensional and multi-regional pollution data collection, providing comprehensive and accurate basic support for subsequent data management and intelligent analysis.

[0120] Data Management Module

[0121] The data management module is used to receive multi-source pollution data collected by the perception module, and integrate, store, pre-process and standardize it to provide reliable data support for subsequent intelligent analysis and application services. The data management module realizes unified management of multi-dimensional data through dynamic data integration, abnormal data cleaning, standardized processing and efficient storage.

[0122] In this implementation, the data management module mainly includes a dynamic integration submodule, a data storage submodule and a data preprocessing submodule, and each submodule operates in coordination to form a complete processing flow from data collection to unified storage.

[0123] Generally speaking, agricultural non-point source pollution data comes from a wide range of sources and in various formats, and data from different sources are asynchronous in time and space. Therefore, the data management module needs to have dynamic integration and efficient storage capabilities, and be able to clean and standardize the collected data to ensure data quality.

[0124] Dynamically integrate submodules

[0125] In this embodiment, the dynamic integration submodule is used to access multi-source data provided by the perception module, including water quality sensor data, high-position video monitoring data and drone monitoring data.

[0126] As an option, the dynamic integration submodule receives data through a unified IoT protocol (such as MQTT). Specifically, sensor data is uploaded in the form of time series, video surveillance data is uploaded in the form of target feature matrix, and drone surveillance data is uploaded in the form of thermal imaging features and image features.

[0127] In a possible implementation, the dynamic integration submodule indexes the data according to the device identifier (ID) and timestamp to ensure the synchronization of multi-source data in terms of time sequence. The specific processing steps are as follows:

[0128] The water quality parameter C collected by the sensor i (t) Index by timestamp;

[0129] The video surveillance feature matrix F i (t) Classification by target detection event label;

[0130] Set drone monitoring feature T i (t) Classification by regional identification.

[0131] In addition, in some embodiments, the dynamic integration submodule can also support flow data access. For example, during the rainy season, the precipitation data collected by the perception module can be dynamically accessed to the data management module and associated with the water quality parameter data.

[0132] Data storage submodule

[0133] In this embodiment, the data storage submodule uses a time series database (such as InfluxDB) to store dynamically updated pollution monitoring data. The time series database supports high-concurrency writing and historical data query, and can meet the needs of large-scale agricultural non-point source pollution monitoring.

[0134] As an option, the data storage submodule partitions the storage based on the type of polluted data. For example:

[0135] Water quality parameter data is stored by river or fish pond area, and each partition record includes area identification, timestamp and pollution concentration;

[0136] Video surveillance data is stored in partitions according to the location of the surveillance points, and the detection results of each frame of image include the target category, location and detection time;

[0137] Drone data is stored according to patrol missions, and records include flight paths, thermal imaging features, and abnormal area information.

[0138] In a possible implementation, the data storage submodule uses partition indexes to optimize data retrieval efficiency. For example, when querying the time series data of pollutant concentrations in a river, only the corresponding partitions need to be retrieved, thereby reducing unnecessary data scanning.

[0139] Specifically, the data storage submodule can organize the monitoring data into the following structure:

[0140] Data = {(ID, Type, Time, Value)}

[0141] in:

[0142] ID is the data source identifier;

[0143] Type indicates the data type (such as water quality parameters, video features);

[0144] Time is the acquisition time;

[0145] Value is the collected value or characteristic data.

[0146] Data preprocessing submodule

[0147] In this embodiment, the data preprocessing submodule is used to clean abnormal data and standardize the data to improve the quality and consistency of the data.

[0148] In general, abnormal data may be caused by equipment failure, communication interruption or environmental interference. The data preprocessing submodule removes abnormal values ​​through the three sigma method. The formula is as follows:

[0149]

[0150] in:

[0151] C i (t) is the original acquisition value;

[0152] μ(C) is the mean of the contamination parameter;

[0153] σ(C) is the standard deviation.

[0154] As an option, the data preprocessing submodule performs standardization transformation on the collected data to eliminate the dimensional differences between different data sources. The standardization formula is as follows:

[0155]

[0156] in:

[0157] is the standardized data;

[0158] μ(C) and σ(C) are the mean and standard deviation respectively.

[0159] In one possible implementation, the data preprocessing submodule also supports missing data filling operations. For example, when a sensor does not upload data in a specific time period, the module estimates the missing value through interpolation to ensure data integrity.

[0160] In some embodiments, the data management module also supports the extraction and marking of pollution event data. For example, when the intelligent analysis module identifies pollution anomalies, the data management module can extract the relevant monitoring data and mark them as abnormal events for subsequent analysis and governance.

[0161] Through the above design, the data management module of the present invention realizes the efficient integration, storage and processing of multi-source data, and provides high-quality data support for the intelligent analysis module. The data management module can adapt to a variety of agricultural non-point source pollution monitoring scenarios and has strong scalability and robustness.

[0162] Intelligent analysis module

[0163] The intelligent analysis module is used to conduct in-depth analysis of the data collected by the perception module, including predicting the diffusion path of pollutants in water bodies based on pollution diffusion modeling, inverting the location and intensity of pollution sources using the source tracing optimization model, and finally determining the emission location in combination with high-level video surveillance data. The intelligent analysis module provides a solid data foundation for the precise positioning of pollution emissions and the formulation of subsequent governance measures through the collaborative work of multiple models and multiple algorithms.

[0164] In general, the intelligent analysis module combines real-time monitoring data and historical records through dynamic perception and modeling to ensure the accuracy and timeliness of pollution analysis. As an option, the intelligent analysis module can build a linkage model based on multi-dimensional perception data to achieve a closed-loop analysis from pollution spread to pollution source location.

[0165] Specifically, the intelligent analysis module includes a pollution diffusion modeling module, a pollution source tracing module, and an emission site location module. The following is a detailed description of each module.

[0166] Pollution Diffusion Modeling Submodule

[0167] In this embodiment, the pollution diffusion modeling module is used to construct a dynamic pollution diffusion model based on water quality sensor monitoring data, river flow velocity information and pollution diffusion characteristics to simulate the spatiotemporal distribution of pollutants in the water body.

[0168] As an option, the Pollution Diffusion Modeling submodule uses a two-dimensional convection-diffusion equation to describe how pollutant concentrations vary in time and space:

[0169]

[0170] in:

[0171] C(x,y,t) represents the pollutant concentration;

[0172] u, v represent the velocity components of the water body;

[0173] D x ,D y represents the diffusion coefficient;

[0174] S(x,y,t) represents the intensity of the pollution source.

[0175] In one possible implementation, the diffusion model is discretized and solved by the finite difference method. With the time step Δt and the space step Δx, Δy as units, the model discretization formula is:

[0176]

[0177] Specifically, the water velocity u, v and the diffusion coefficient D x ,D y The output of the model simulation is the concentration distribution of pollutants in the river or fish pond, which is used to identify high-risk areas of pollutants.

[0178] In one possible implementation, model parameters (such as flow rate and diffusion coefficient) are dynamically updated through real-time flow rate sensors and water quality monitoring data to ensure the accuracy of simulation results.

[0179] Generally speaking, the concentration distribution data output by the pollution diffusion modeling module will be directly provided to the pollution source tracing module to provide basic data support for tracing optimization.

[0180] Pollution source tracing submodule

[0181] In this embodiment, the pollution source tracing module inverts the spatial distribution and emission intensity of the pollution sources by performing a difference analysis between the pollutant observation data collected by the perception module and the diffusion modeling results.

[0182] As an option, the pollution source tracing submodule establishes a tracing optimization model to determine the pollution source parameters by minimizing the difference between the model simulation value and the actual observation value. The optimization objective function is:

[0183] min s(x,y,t) ∫ Ω [C obs (x,t,t)-C sim (x,y,t;S)] 2 dxdy

[0184] in:

[0185] C obs (x, y, t) are actual observed values;

[0186] C sim (x, y, t; S) are the values ​​simulated by the diffusion model.

[0187] In one possible implementation, the provenance optimization model is solved by the gradient descent method. The gradient update formula is:

[0188]

[0189] in:

[0190] S k represents the pollution source intensity of the kth iteration;

[0191] J(S) is the objective function;

[0192] α is the learning rate.

[0193] Specifically, the initial parameters of the source tracing model are provided by empirical rules or historical data, and the spatial distribution and intensity information of pollution sources are output after iterative solution.

[0194] In one possible implementation, the pollution source tracing module combines the temporal and spatial data of the perception module to further improve the accuracy of the tracing results by analyzing the changing trends and diffusion paths of the monitored concentrations.

[0195] Generally, the candidate locations of pollution sources and emission intensity information generated by the pollution source tracing module will be passed as input to the emission location location module.

[0196] Emission location location module

[0197] In this embodiment, the emission location positioning module combines the pollution diffusion modeling results, the pollution source tracing results and the abnormal behavior data recorded by the high-position video surveillance to finally determine the specific pollution emission location.

[0198] In some implementations, the emission location location module first matches the abnormal behavior time and space data recorded by the high-level video surveillance with the pollution path simulated by the diffusion model based on the candidate pollution source locations provided by the source tracing module. For example:

[0199] If the abnormal behavior recorded in the video occurs at time t v The pollutant diffusion time t predicted by the diffusion model c If the matching conditions are met within a certain range, the location of the abnormal behavior in the video will be taken as the priority confirmed emission location.

[0200] As an option, the discharge location location module further classifies the categories of abnormal behaviors, such as identifying the difference between fish catching behavior and sewage discharge behavior, to improve the certainty of the discharge location.

[0201] In one possible implementation, the module also combines images and thermal imaging data collected by the drone to verify the traceability and positioning results to ensure the accuracy of the final results.

[0202] Generally, the emission location positioning module will output the final confirmed pollution emission location and its corresponding time to the application service module for subsequent early warning and response.

[0203] Through the above implementation, those skilled in the art can realize the complete technical solution of the intelligent analysis module based on the disclosed content. The intelligent analysis module works together through pollution diffusion modeling, pollution source tracing and emission location positioning, providing a scientific and reliable technical basis for the monitoring and positioning of pollutants, ensuring the efficient operation of the agricultural non-point source pollution supervision system.

[0204] Application service module

[0205] The application service module is used to convert the analysis results generated by the intelligent analysis module into intuitive visual information, and realize intelligent early warning and remote control functions. Through intelligent early warning, GIS visualization display and mobile terminal operation support, this module realizes seamless connection from data results to decision-making, greatly improving the efficiency and accuracy of pollution supervision.

[0206] In general, the application service module needs to integrate the results of the perception module and the intelligent analysis module in real time, including pollutant concentration data, diffusion path, source analysis and emission location information. As an option, the module can display the pollution distribution and diffusion trend through dynamic maps, combined with threshold-triggered alarms and mobile notifications, to achieve real-time pollution monitoring and response.

[0207] Specifically, the application service module includes an intelligent warning module, a visualization module and a mobile operation module. The following is a detailed description of each sub-module.

[0208] Intelligent early warning submodule

[0209] In this embodiment, the intelligent early warning module is used to automatically trigger an alarm and push detailed information of abnormal areas based on the pollutant concentration threshold set by the system, combined with diffusion modeling and emission site positioning results.

[0210] As an option, the warning condition is defined by a preset pollutant concentration threshold. For example, when the pollutant concentration C(x,y,t) in a certain area exceeds the preset threshold C threshold When , an alarm event is triggered, and the alarm logic is expressed as:

[0211]

[0212] Specifically, warning events include the following information:

[0213] The spatial location of the abnormal area;

[0214] Types and concentrations of pollutants exceeding the standard;

[0215] The time when the alert is triggered.

[0216] In one possible implementation, the warning information also includes abnormal behavior records and emission location results, such as the time and location of abnormal behavior recorded by high-position video surveillance, and the final confirmed pollution emission point.

[0217] Generally, the alarm information of the intelligent early warning module will be pushed to the visualization module and the mobile operation module at the same time to provide support for subsequent decision-making and execution.

[0218] In a possible implementation, the warning information is pushed to the administrator's terminal device in the form of a text message or application notification, and the notification content includes the alarm details and recommended emergency treatment measures.

[0219] Visualization submodule

[0220] In this embodiment, the visualization module is used to dynamically display the results of the intelligent analysis module on the GIS map, including information such as pollution diffusion path, pollution source distribution and emission location.

[0221] As an option, the base map of the GIS map can be satellite remote sensing images or vector topographic maps, and the following information layers can be superimposed:

[0222] Pollutant concentration distribution layer: Based on the output of the pollution diffusion model, a color gradient is used to represent the pollution concentration in different areas;

[0223] Pollution source location layer: displays the pollution source location and emission intensity determined by the source tracing model;

[0224] Monitoring equipment distribution layer: mark the specific locations of sensors, high-position video surveillance and drones in the perception module.

[0225] Specifically, the rendering of pollutant concentration distribution uses an interpolation algorithm to convert the data of discrete monitoring points into a continuous distribution map. For example, the concentration distribution can be calculated based on the Kriging interpolation method:

[0226]

[0227] in:

[0228] C(x,y) is the concentration value of the target point;

[0229] C i is the concentration at the monitoring point;

[0230] w i is the interpolation weight.

[0231] As an option, the visualization module supports dynamic display of the time dimension. For example, users can slide the timeline to view the pollution diffusion path and traceability results at different time points, and intuitively understand the diffusion trend of pollutants.

[0232] In one possible implementation, the visualization module also integrates high-position video surveillance and real-time images taken by drones. Specifically, the system marks the camera location and shooting range of the high-position surveillance on the GIS map, and allows users to click to view video playback or real-time images.

[0233] Generally speaking, the functions of the visualization module also include the display of emission location results, such as displaying the location and time information of the pollution emission point in the form of annotations on the map, and providing background information for source tracing analysis.

[0234] Mobile terminal operation submodule

[0235] In this embodiment, the mobile terminal operation module is used to enable management personnel to remotely view monitoring data and analysis results, and supports drone scheduling and task execution.

[0236] As an option, mobile capabilities include the following:

[0237] Real-time data viewing: Users can view pollution diffusion paths, monitoring data and early warning events through mobile applications.

[0238] Drone dispatch: After discovering an abnormally polluted area, users can send instructions through the mobile terminal to control the drone to fly to the designated location and collect on-site images.

[0239] Remote governance operations: Combined with the governance recommendations of the optimal control module, users can directly issue instructions through the mobile terminal to allocate governance resources or adjust emission reduction strategies.

[0240] As an option, the mobile terminal operation module allows users to directly issue drone tasks. For example, when the system pushes alarm information for a certain area, managers can select the area through the mobile terminal and send automatic inspection instructions to the drone.

[0241] In one possible implementation, the mobile operation module supports multi-user collaboration. For example, managers at different levels can share pollution information and task progress in real time and jointly formulate governance measures.

[0242] Generally speaking, the interface design of the mobile operation module is based on the principle of simplicity and intuitiveness, ensuring that managers can quickly obtain key information and respond in complex scenarios.

[0243] Through the above implementation, those skilled in the art can realize the complete technical solution of the application service module based on the public content. The application service module provides comprehensive support for the supervision of agricultural non-point source pollution through the combination of intelligent early warning, visualization and mobile terminal functions, ensuring efficient execution and feedback loop from data to action.

[0244] In general, the present invention collects data from water quality sensors, high-position video surveillance and drone equipment in real time through the perception module, combines the dynamic integration and standardized processing of multi-source data with the data management module, uses the intelligent analysis module to achieve pollutant diffusion modeling, source tracing analysis and precise positioning of emission locations, and provides intelligent early warning, GIS visualization and mobile terminal operation support through the application service module, forming a closed-loop process from data collection, analysis to governance decision-making. The present invention can efficiently and accurately monitor agricultural non-point source pollution, significantly improve supervision efficiency, and provide strong technical support for water environment protection and governance.

[0245] Please see attached Figure 2 The present invention also provides a supervision method for agricultural non-point source pollution scenarios, aiming to use multi-source perception data and intelligent analysis technology to achieve dynamic monitoring, accurate source tracing and efficient early warning of agricultural non-point source pollution. This method forms a closed-loop governance system from perception to decision-making through the process of data collection, dynamic integration, diffusion modeling, source tracing analysis and result visualization, which can effectively improve the efficiency and accuracy of water pollution supervision.

[0246] This method relies on real-time data and efficient modeling algorithms, combined with GIS visualization and intelligent early warning technology, to provide scientific support for the tracing and control of agricultural non-point source pollution. The following is a detailed description of the implementation of each step of the method.

[0247] S1. Real-time collection of agricultural non-point source pollution data

[0248] In this embodiment, step S1 involves real-time collection of agricultural non-point source pollution data, including multi-source data collected by water quality sensors, high-position video monitoring equipment and drones.

[0249] Specifically:

[0250] Water quality sensors are deployed on main roads and important pond areas to monitor the concentrations of key pollutants such as chemical oxygen demand, ammonia nitrogen and total phosphorus. The frequency of data collection can be dynamically adjusted according to regional pollution risks.

[0251] High-level video surveillance equipment records abnormal behavior around ponds and sewage pipes, including human activities, fish catching behavior, and water level changes. Video surveillance data is marked with timestamps and locations for subsequent matching and analysis.

[0252] Drones are responsible for collecting high-resolution images and thermal imaging data in remote areas, and collecting evidence of abnormal behavior in key areas, supplementing the deficiencies of high-level monitoring.

[0253] S2. Dynamically integrate and preprocess the collected data

[0254] In this implementation, step S2 dynamically integrates and preprocesses the collected data.

[0255] Specifically:

[0256] The dynamic integration module accesses the data streams of water quality sensors, video surveillance equipment and drones through a unified interface, and archives the data according to spatial and temporal indexes.

[0257] The preprocessing module removes anomalies from the monitoring data and filters out erroneous data caused by equipment failure or environmental interference. At the same time, it uses a standardized format to uniformly convert data from different devices to ensure compatibility for subsequent analysis.

[0258] In some embodiments, the monitoring data of the water quality sensor is subjected to noise filtering to improve the accuracy of the pollutant concentration data.

[0259] S3. Based on the river flow velocity, diffusion coefficient and pollutant concentration data, a pollution diffusion model is constructed to dynamically simulate the diffusion path and concentration distribution of pollutants in the water body.

[0260] In this embodiment, step S3 constructs a pollution diffusion model based on the river flow velocity, diffusion coefficient and pollutant concentration data to dynamically simulate the diffusion path and concentration distribution of pollutants in the water body.

[0261] Specifically:

[0262] The diffusion model uses flow velocity and diffusion coefficient as core parameters, and simulates the diffusion trajectory of pollutants from the source to the downstream by updating the flow velocity and pollutant concentration in real time.

[0263] The simulation results of the diffusion path are discretized and stored based on time and space, providing data support for subsequent traceability analysis and anomaly matching.

[0264] S4. Using the difference between the diffusion model simulation results and the water quality monitoring data, the spatial location and emission intensity of the pollution source are inverted through the source tracing optimization model.

[0265] In this embodiment, step S4 utilizes the difference between the diffusion model simulation results and the water quality monitoring data to invert the spatial location and emission intensity of the pollution source through the source tracing optimization model.

[0266] Specifically:

[0267] The source tracing module compares the monitored pollutant concentration with the concentration predicted by the diffusion model, and inverts the specific location of the pollution source by minimizing the error between the two.

[0268] The source tracing analysis results mark the candidate points of the pollution sources and their emission intensities, providing candidate areas for the subsequent positioning module.

[0269] S5. Comprehensive diffusion model prediction, source tracing analysis results and abnormal behavior data recorded by high-position video surveillance, accurately determine the specific location and time of pollution emissions through time and space matching

[0270] In this embodiment, step S5 integrates the diffusion model prediction, source tracing analysis results and abnormal behavior data recorded by high-position video surveillance to accurately determine the specific location and time of pollution emission.

[0271] Specifically:

[0272] The system matches the time of abnormal behavior recorded by high-position video surveillance with the predicted path of the diffusion model, giving priority to screening candidate pollution sources that are consistent in time and space.

[0273] Combined with the categories of abnormal behavior records (such as sewage pipe discharge or fish catching behavior), the specific location and time of pollution discharge can be ultimately determined.

[0274] In some embodiments, thermal imaging data from drones is used to verify the accuracy of traceability results and further enhance positioning accuracy.

[0275] S6. Trigger intelligent warning based on concentration threshold and generate abnormal area information based on emission location positioning results

[0276] In this embodiment, step S6 triggers an intelligent warning based on a concentration threshold, and generates abnormal area information in combination with the emission location positioning result.

[0277] Specifically:

[0278] The system makes threshold judgments on the monitoring data of the water quality sensor. If the pollutant concentration at a certain monitoring point exceeds the preset value, an early warning is immediately triggered.

[0279] The warning information includes detailed data of the area where the concentration exceeds the standard, the time of abnormal behavior and the final located emission point, which is pushed to the management end for decision-making.

[0280] S7. Dynamically display pollution diffusion paths, source tracing results and location analysis on GIS maps

[0281] In this embodiment, step S7 dynamically displays the pollution diffusion path, source tracing results and positioning analysis on the GIS map.

[0282] Specifically:

[0283] GIS maps visualize diffusion paths in the form of dynamic trajectories, showing the distribution of pollutants at different points in time.

[0284] The system marks candidate pollution source points and the final emission locations on the map, and allows users to click to view detailed analysis results.

[0285] The management end can directly dispatch drones through GIS maps to achieve accurate inspections of key areas.

[0286] In some implementations, the GIS map also displays the overlay results of historical data, providing managers with a basis for analyzing the long-term trends of pollution sources.

[0287] Through the above-mentioned specific implementation methods, the method of the present invention realizes the complete process from multi-source data collection, dynamic modeling to precise positioning, which can significantly improve the supervision efficiency of agricultural non-point source pollution and provide reliable technical support for the protection and management of water environment.

[0288] Example:

[0289] Please see attached Figure 3 This embodiment is based on the actual application of agricultural non-point source pollution scenarios. Taking the fish pond area of ​​a certain river basin as a typical case, combined with the grid fish pond layout and sensor layout points shown in the figure, the specific application of the present invention is described. During the implementation process, through the deployment of sensors, monitoring data collection and analysis, the location of pollution sources and accurate monitoring of abnormal behaviors are achieved.

[0290] In this embodiment, the monitoring area includes multiple fish ponds in a grid layout, which are connected by drainage channels and eventually merge into the main stream represented by the blue line. As a potential source of agricultural non-point source pollution, the tailwater and sewage discharge of fish ponds may pollute downstream water bodies. In order to achieve accurate supervision of pollution spread and sewage discharge, water quality sensors are arranged at the locations marked in yellow and green, and multi-dimensional perception is carried out in combination with high-position video surveillance and drone inspections.

[0291] Sensor placement

[0292] In this embodiment, chemical oxygen demand, ammonia nitrogen, and total phosphorus water quality sensors are installed at the yellow and green marked sensor layout points respectively to monitor the pollutant concentration in real time. The selection of these layout points is based on the following:

[0293] Yellow marking points: located at the main intersections of fish ponds and main streams, to monitor the pollutant concentration of fish pond tailwater before it is discharged into the main road.

[0294] Green marking points: key nodes located on the main roads, used to monitor the diffusion path and concentration changes of pollutants in the main stream.

[0295] Generally, the density of sensor deployment is related to the number of fish ponds and the flow rate of the river basin to ensure the spatial continuity of the monitoring data coverage area. Sensors upload monitoring data through the Internet of Things technology and index it in combination with time and space information.

[0296] Data collection and preprocessing

[0297] In this embodiment, water quality sensors collect pollutant concentration data in real time, and high-position video monitoring equipment and drones work together to record abnormal behavior in the fish pond area. The monitoring data is transmitted to the data management module through the dynamic integration module and the following pre-processing operations are performed:

[0298] Eliminate abnormalities in monitoring data and filter out erroneous values ​​caused by equipment failure;

[0299] Standardize sensor data, video surveillance data, and drone data into a unified temporal and spatial format;

[0300] The filtering algorithm is applied to denoise the pollutant concentration data collected by the sensor.

[0301] Diffusion modeling and source analysis

[0302] In this embodiment, the diffusion model is used to predict the diffusion path and concentration distribution of pollutants after they are discharged from the fish pond into the main road. Specifically, it includes:

[0303] Combined with the real-time data from the sensor deployment points, the river flow velocity and diffusion coefficient are dynamically calculated;

[0304] A two-dimensional convection-diffusion model was constructed to simulate the diffusion process of pollutants between fish ponds and mainstream streams.

[0305] Subsequently, the source tracing optimization model was used to analyze the differences between the monitoring data and the simulation data, and to invert the location of the pollution source and the emission intensity. The source tracing module combines time and space data to match the possible emission behavior of the fish pond with the pollution diffusion path.

[0306] Discharge location location

[0307] Combined with the diffusion modeling and traceability analysis results, the abnormal behavior recorded by the high-level video surveillance equipment is used to further verify the sewage discharge behavior of the fish pond. In this embodiment, the system comprehensively matches the following data:

[0308] The time of abnormal behavior in the fish pond recorded by high-position video surveillance (such as sudden drop in water level, pipe discharge behavior);

[0309] The time it takes for pollutants to diffuse to the yellow marked point in the diffusion model;

[0310] Pollutant concentration trends at green marked points.

[0311] By matching time and space, the system eventually locates a specific fish pond as the source of pollution emissions and marks the specific time of emission.

[0312] Intelligent early warning and visual display

[0313] Once the fish pond discharge behavior is determined, this embodiment triggers an alarm through the intelligent early warning module and pushes abnormal area information to the management end. The early warning information includes:

[0314] The specific location of the fish pond and the time of discharge;

[0315] Pollutant concentration data for yellow and green marked points;

[0316] Auxiliary forensic information collected by video surveillance and drones.

[0317] The pollution diffusion path and the sewage discharge location of fish ponds are dynamically displayed on the GIS map. Managers can view early warning information through mobile terminals and dispatch drones to patrol key areas and collect further evidence.

[0318] This embodiment realizes dynamic supervision of the gridded fish pond area and accurate positioning of pollution emission behaviors through sensor layout, diffusion modeling and source tracing analysis, emission location positioning and intelligent early warning display. This method provides complete technical support for the efficient supervision of agricultural non-point source pollution, and significantly improves the scientificity and practicality of supervision work.

[0319] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The supervision system for agricultural non-point source pollution scenarios is characterized by: include: The perception module is used to collect pollutant concentration, hydrological characteristics and abnormal behavior data through water quality sensors, high-position video monitoring equipment and drones, and provide support for the initial source location of abnormal water behavior; The data management module is used to dynamically integrate, store and standardize the data collected by the perception module; Intelligent analysis module, which is used to predict the diffusion path of pollutants in the river based on the pollutant diffusion model, and invert the spatial location and intensity of the pollution source by combining the abnormal behavior records of video surveillance with the diffusion model; The application service module is used to display analysis results, trigger warnings and support remote governance operations through intelligent early warning systems, data visualization platforms and mobile devices.

2. The monitoring system for agricultural non-point source pollution scenes according to claim 1 is characterized in that: The perception module comprises: Water quality sensors are used to monitor the concentrations of chemical oxygen demand, ammonia nitrogen and total phosphorus, and record the temporal and spatial distribution of water quality anomalies in real time, providing basic data support for pollution diffusion models; High-level video surveillance equipment is used to monitor multiple monitoring points in real time. It uses image recognition algorithms to detect abnormal behaviors, including water level changes, fish catching activities, and suspected sewage discharge. It also works in conjunction with diffusion models to locate suspected pollution sources in combination with water quality abnormality data and provide preliminary clues. Drones are used for autonomous flight control to locate suspicious ponds or remote areas, and are equipped with thermal imaging equipment and multi-spectral sensors to monitor sewage discharge at night or in hidden areas.

3. The monitoring system for agricultural non-point source pollution scenes according to claim 1 is characterized in that: The data management module comprises: Dynamic integration module, used to connect multi-source data collected by water quality sensors, video surveillance equipment and drones to a shared data center through a unified interface; A data storage module, used to store data in blocks according to pollution areas and time indexes; The data preprocessing module is used to remove anomalies, denoise and standardize the format of the collected data.

4. The monitoring system for agricultural non-point source pollution scenes according to claim 1 is characterized in that: The intelligent analysis module includes: The pollution diffusion modeling module is used to dynamically build a pollutant diffusion model based on river flow velocity, pollution diffusion coefficient and water quality sensor data to predict the temporal and spatial distribution of pollutants in the water body; The pollution source tracing module preliminarily inverts the spatial location and emission intensity of pollution sources by analyzing the differences between monitoring data and diffusion model simulation results; The emission location positioning module is used to determine the specific location of pollution emissions by combining the pollution diffusion modeling results, pollution source tracing results and abnormal behavior records of high-position video surveillance.

5. The monitoring system for agricultural non-point source pollution scenes according to claim 4 is characterized in that: The pollution diffusion modeling module uses a convection-diffusion model to dynamically simulate the temporal and spatial distribution of pollutants based on the diffusion characteristics of pollutants in water bodies and flow velocity field information, including: The two-dimensional convection-diffusion equation is used to describe the dynamic changes of pollutant concentrations, taking into account the comprehensive effects of flow velocity components, water diffusion coefficients and pollution source intensity on the changes of pollutant concentrations; The pollutant concentration C(x,y,t) is taken as the target variable, the velocity components u,v, and the diffusion coefficient D x ,D y And the pollution source intensity S(x,y,t) are input parameters, satisfying the following relationship: The numerical method is used to discretize and solve the equations, dynamically simulating the diffusion path and concentration distribution of pollutants from the pond to the main road; Combine the pollutant concentration data from the water quality sensor and the real-time flow rate data from the flow rate sensor to regularly update the model parameters; The diffusion simulation results are combined with the temporal and spatial information of abnormal behaviors recorded by high-position video surveillance to locate the scope of the pollution source and provide data support for subsequent source tracing analysis.

6. The monitoring system for agricultural non-point source pollution scenes according to claim 5 is characterized in that: The pollution source tracing module inverts the spatial distribution and intensity of pollution sources through the difference between the pollutant diffusion modeling results and the observation data. The pollution source tracing process includes the following steps: The pollutant concentration observation data C collected by the perception module obs (x, y, t) and the concentration distribution C obtained by simulation based on pollution diffusion modeling sim (x, y, t; S) to evaluate the deviation between the model prediction and the actual monitoring data; A source tracing model with pollution source intensity S(x,y,t) as the optimization variable is constructed to minimize the error between observed data and simulated data, satisfying the following optimization relationship: min S(x,y,t) ∫ Ω [C obs (x,t,t)-C sim (x,y,t;S)] 2 dxdy The gradient descent method is used to iteratively optimize the pollution source intensity S(x,y,t), and the model error is gradually reduced through the following update formula: Among them, J(S) is the optimization objective function, α is the learning rate; Combined with abnormal behavior data recorded by high-position video surveillance, the traceability results are verified and corrected, and the pollution intensity and time characteristics are matched with the time points of abnormal behavior to further narrow the spatial scope of the pollution source; Finally determine the spatial location and emission intensity of the pollution source.

7. The monitoring system for agricultural non-point source pollution scenes according to claim 6 is characterized in that: The step of determining the specific location of pollution emission by the emission location positioning module includes: Combine the pollutant diffusion path and pollution concentration distribution data generated by the pollution diffusion modeling module, and the candidate locations of pollution sources inverted by the pollution source tracing module; Retrieve the time and location of abnormal behavior recorded by high-position video surveillance, and match them with the diffusion time point and path calculated by the diffusion model; Prioritize screening the areas in the video where abnormal behavior occurs and is consistent with abnormal water quality data, and ultimately identify the specific location and time of pollution discharge.

8. The monitoring system for agricultural non-point source pollution scenes according to claim 1 is characterized in that: The application service module includes: Intelligent early warning module, which is used to automatically trigger an alarm for exceeding the standard based on the monitoring data of water quality sensors, the prediction results of the pollution diffusion model and the analysis results of the emission location positioning module, and push detailed information of the abnormal area, including the location and time of possible pollution sources; Visualization module, used to dynamically display the pollutant diffusion path, source tracing analysis results and the final emission location of the emission location location module on the GIS map; supports users to view the diffusion prediction and pollution source impact range in different time periods; The mobile operation module is used to support remote drone dispatch and real-time viewing of video surveillance, pollution concentration data and positioning results, providing a convenient mobile supervision tool and allowing managers to directly mark and record on-site inspection information.

9. The monitoring system for agricultural non-point source pollution scenes according to claim 8 is characterized in that: The visualization module provides dynamic display and prediction functions of pollution diffusion paths on GIS maps based on pollutant concentration data, dynamic pollution source locations and high-level video monitoring records stored in the time series database, and supports concentration change analysis across time dimensions.

10. A method for monitoring agricultural non-point source pollution scenes, based on the system according to any one of claims 1 to 9, characterized in that: The following steps are involved: Real-time collection of agricultural non-point source pollution data, including pollutant concentrations monitored by water quality sensors, abnormal behavior data recorded by high-position video surveillance, and on-site image information collected by drones; Dynamically integrate and preprocess the collected data, including abnormal data removal and format standardization; Based on the river flow velocity, diffusion coefficient and pollutant concentration data, a pollution diffusion model is constructed to dynamically simulate the diffusion path and concentration distribution of pollutants in the water body; Using the difference between the diffusion model simulation results and the water quality monitoring data, the spatial location and emission intensity of the pollution source are inverted through the source tracing optimization model; The specific location and time of pollution emission can be accurately determined by combining diffusion model prediction, source tracing analysis results and abnormal behavior data recorded by high-level video surveillance through time and space matching; Intelligent warning is triggered based on concentration thresholds, and abnormal area information is generated in combination with emission location positioning results; Dynamically display pollution diffusion paths, traceability results and positioning analysis on GIS maps to support remote scheduling and decision-making on the management side.

Images