Water pollution inspection regulation method and system

Abstract

The application discloses a water pollution inspection regulation method and system, relates to the technical field of water pollution inspection regulation, and realizes all-dimensional accurate perception by fusing multi-source data such as hyperspectrum, unmanned ships and Internet of Things and constructing a unified high-temporal and high-spatial resolution water quality distribution map; in combination with real-time water flow data, the diffusion path of pollutants is dynamically simulated, and a pollution source probability distribution is generated, so that pollution tracing is improved from'static speculation' to 'dynamic tracking', the positioning accuracy and efficiency are significantly improved; meanwhile, the prediction and early warning ability for the future change trend of key water quality parameters is enhanced, time is saved for early intervention; based on a water pollution knowledge graph and a multi-objective optimization algorithm, a scientific and self-adaptive change optimization regulation strategy is generated, finally, a closed-loop efficient automatic execution system is formed, and the overall efficiency of water pollution emergency response and normal regulation is greatly improved.

Description

Technical Field

[0001] This application relates to the field of water pollution inspection and control technology, specifically to a water pollution inspection and control method and system. Background Technology

[0002] Significant shortcomings remain in current water environment monitoring and management. First, data sources are scattered; second, traditional pollution source tracing relies heavily on human experience and static models; third, water quality prediction models are often poorly coupled with the dynamic changes of pollution sources and actual hydrological conditions, resulting in limited prediction accuracy; and finally, the formulation and implementation of management strategies are disconnected, relying heavily on manual decision-making, which is inefficient and makes it difficult to achieve synergistic optimization of multiple objectives. Therefore, an intelligent water pollution source tracing and control system capable of achieving a closed-loop system across the entire chain is needed.

[0003] Existing technology, such as the invention application patent with publication number CN118628087A, discloses a data analysis-based method for the operation, maintenance, and inspection of water pollution treatment equipment, relating to the field of equipment maintenance and inspection technology. This data analysis-based method for the operation, maintenance, and inspection of water pollution treatment equipment analyzes current inspection personnel data and wastewater treatment plant inspection platform records to output the number of water pollution treatment equipment that the current inspector can inspect during the current inspection. Then, based on the inspection needs of each piece of equipment and the number of water pollution treatment equipment that the current inspector can inspect, the method marks the water pollution treatment equipment that needs maintenance and inspection during the current inspection. Furthermore, based on data analysis, it generates the current inspection route and adjusts the route based on real-time monitoring of the wastewater treatment effect of each water pollution treatment device in the wastewater treatment plant. This makes the inspection route targeted and efficient, and allows for timely feedback and maintenance for any problems that occur in the wastewater treatment effect of the water pollution treatment equipment.

[0004] Regarding the above-mentioned solutions, the inventors of this application have discovered that the above-mentioned technology has at least the following technical problems:

[0005] 1. Current monitoring methods still have limitations in spatiotemporal resolution, failing to create a comprehensive and high-precision three-dimensional water quality distribution map of the entire basin. This cannot improve the breadth and dimension of the data, nor can it fully reflect the water quality status. The current dynamic pollution source tracing probability model, which does not incorporate real-time flow field data, cannot change the traditional passive mode that can only monitor the current state of pollution. It does not couple water quality distribution with real-time water flow dynamics, lacks the ability to reverse-engineer the diffusion path of pollutants, cannot quickly locate potential pollution source areas, cannot improve the scientificity and accuracy of source tracing, and cannot provide targeted targets for precise prevention and control.

[0006] 2. Current predictive models lack mechanism-driven approaches that integrate the uncertainties of pollution sources. These models cannot predict the changing trends of key water quality parameters over a future period, fail to consider physical diffusion laws or the uncertainties of pollution sources, and thus cannot accurately reflect the complex and ever-changing realities. They cannot be based on different pollution source probabilities or control plans, do not simulate water quality evolution under different scenarios, and cannot provide quantitative evidence for evaluating the effectiveness of different control strategies. Furthermore, current adaptive strategy generation combining knowledge graphs and multi-objective optimization is lacking, hindering intelligent and knowledge-driven decision-making, multi-objective collaborative optimization, and strategies that lack high adaptability and specificity. These strategies cannot address different pollution events and control needs, nor can they achieve a "one-size-fits-all" approach. Summary of the Invention

[0007] To address the aforementioned technical shortcomings, the purpose of this application is to provide a water pollution inspection and control method and system.

[0008] To solve the above-mentioned technical problems, this application adopts the following technical solution: In the first aspect, this application provides a water pollution inspection and control method, which includes the following steps: Step 1, multi-source data fusion: Based on the pre-acquired UAV hyperspectral data, UAV multi-parameter sensor data and fixed monitoring point IoT data, a unified water quality distribution data is generated.

[0009] Step 2, Pollution Source Tracing Analysis: Based on the unified water quality distribution data and the pre-acquired real-time water flow data, analyze the pollutant diffusion path and generate dynamic pollution source probability data.

[0010] Step 3: Water quality prediction: Based on the dynamic pollution source probability data and preset fluid dynamic parameters, predict the key water quality parameters.

[0011] Step 4: Generation of Control Strategies: Based on the key water quality parameter data and the preset water pollution knowledge graph, adaptive control strategy data is generated through a multi-objective optimization algorithm.

[0012] Step 5: Execution of coordinated control: Based on the adaptive control strategy data, control the preset water conservancy control facilities to execute water pollution control.

[0013] Preferably, the step of generating unified water quality distribution data based on pre-acquired UAV hyperspectral data, UAV multi-parameter sensor data, and fixed monitoring point IoT data includes: performing spatiotemporal alignment based on the pre-acquired UAV hyperspectral data, UAV multi-parameter sensor data, and fixed monitoring point IoT data to obtain aligned data; performing feature extraction based on the aligned data to generate feature data; and eliminating noise based on the feature data using a weighted fusion calculation formula to generate unified water quality distribution data.

[0014] Preferably, the step of eliminating noise and generating uniform water quality distribution data based on the feature data using a weighted fusion calculation formula includes: specifically using a calculation formula... Obtain uniform water quality distribution data ,in This is represented by the ID corresponding to the feature data. , This is represented by the number of feature data. Represented as the first Each feature data, Represented as the first The weighting factors corresponding to each feature data.

[0015] Preferably, the step of analyzing pollutant diffusion paths and generating dynamic pollution source probability data includes: constructing a dynamic graph structure based on the unified water quality distribution data and pre-acquired real-time water flow data to obtain initial graph topology data; performing pollutant diffusion path analysis using a dynamic adaptive graph neural network model based on the initial graph topology data to generate path analysis data; performing probability calculations based on the path analysis data to generate dynamic pollution source probability data; dynamically adjusting the initial graph topology data based on the real-time water flow data to obtain updated graph topology data; and outputting high-precision source tracing results based on the updated graph topology data and the path analysis data.

[0016] Preferably, the prediction of key water quality parameters based on the dynamic pollution source probability data and preset fluid dynamic parameters includes: performing data fusion and standardization processing based on the dynamic pollution source probability data and preset fluid dynamic parameters to obtain model input data; then applying a physical information neural network model for forward propagation calculation to obtain initial water quality parameter data; performing physical constraint verification on the initial water quality parameter data based on preset fluid dynamic equations to obtain water quality residual data; then optimizing and adjusting the initial water quality parameter data to obtain key water quality parameter data; and finally outputting the key water quality parameter data as a predicted trend for future time periods.

[0017] Preferably, the physical constraint verification of the initial water quality parameter data includes: using the calculation formula of the physical constraint loss function. Calculate the total loss value ,in This is expressed as the mean squared error between the predicted data and the training data. It is expressed as the norm of the residuals of the physical equations. It is represented as a weighting coefficient used to balance data loss and physical loss.

[0018] Preferably, the step of generating adaptive control strategy data based on the key water quality parameter data and a preset water pollution knowledge graph using a multi-objective optimization algorithm includes: performing a knowledge graph query operation based on the key water quality parameter data and the preset water pollution knowledge graph to obtain historical case data and real-time feedback data; performing strategy optimization calculations using a multi-objective optimization algorithm to obtain initial control strategy data; and then performing a balance verification operation on the initial control strategy data to obtain adaptive control strategy data.

[0019] Preferably, the step of performing strategy optimization calculations using a multi-objective optimization algorithm to obtain initial control strategy data includes: using the calculation formula of the multi-objective optimization algorithm. Initial control strategy data were obtained. ,in Represented as decision variables, Represented as an efficiency function, Represented as a cost function, and These are represented as the weighting factors corresponding to the efficiency function and the cost function, respectively; the decision variables that minimize the weighted sum are obtained through the optimization solver to generate initial control strategy data.

[0020] Preferably, the step of controlling a preset water conservancy control facility based on the adaptive control strategy data to perform water pollution control includes: performing instruction parsing operations based on the adaptive control strategy data to obtain facility control parameters; then performing control operations on the preset water conservancy control facility to perform water pollution control, thereby collecting real-time water quality data; and performing parameter dynamic adjustment operations based on the real-time water quality data and the adaptive control strategy data to obtain updated control parameters, thereby performing control operations on the water conservancy control facility to achieve closed-loop execution.

[0021] In a second aspect, this application provides a system for a water pollution inspection and control method, comprising: preferably, a multi-source data fusion module, which generates unified water quality distribution data based on pre-acquired UAV hyperspectral data, unmanned surface vessel multi-parameter sensor data, and fixed monitoring point IoT data.

[0022] The pollution source tracing and analysis module analyzes the pollutant diffusion path based on the unified water quality distribution data and the pre-acquired real-time water flow data, and generates dynamic pollution source probability data.

[0023] The water quality prediction module predicts key water quality parameters based on the dynamic pollution source probability data and preset fluid dynamic parameters.

[0024] The regulation strategy generation module generates adaptive regulation strategy data based on the key water quality parameter data and the preset water pollution knowledge graph through a multi-objective optimization algorithm.

[0025] The linkage control execution module, based on the adaptive control strategy data, controls the preset water conservancy control facilities to perform water pollution control.

[0026] The beneficial effects of this application are as follows: 1. The water pollution inspection and control method and system provided in this application, by integrating multi-source data such as hyperspectral, unmanned vessel and Internet of Things, constructs a unified high spatiotemporal resolution water quality distribution map to achieve full-area three-dimensional accurate perception; combined with real-time water flow data, it dynamically simulates the diffusion path of pollutants and generates the probability distribution of pollution sources, thereby improving pollution source tracing from "static speculation" to "dynamic tracking", significantly improving the accuracy and efficiency of positioning; at the same time, it enhances the ability to predict and warn of the future changing trends of key water quality parameters, buying time for early intervention; based on water pollution knowledge graph and multi-objective optimization algorithm, it generates scientific and adaptive optimization control strategies, and finally forms a closed-loop efficient automated execution system, which greatly improves the overall effectiveness of water pollution emergency response and routine control.

[0027] 2. This application integrates data from UAVs (high-altitude area coverage), unmanned vessels (surface or underwater linear patrols), and fixed points (continuous monitoring of key locations), overcoming the limitations of single monitoring methods in terms of spatiotemporal resolution. It forms a three-dimensional water quality distribution map of the entire basin with no blind spots and high precision, improving the breadth and dimensionality of data, comprehensively reflecting the water quality status, and laying the foundation for high-precision analysis.

[0028] 3. This application combines a dynamic pollution source tracing probability model with real-time flow field data, which changes the traditional passive mode that can only monitor the current state of pollution. By coupling water quality distribution with real-time water flow dynamics, it reverse-engineers the diffusion path of pollutants, quickly identifies potential pollution source areas, improves the scientific nature and accuracy of source tracing, and provides targeted targets for precise prevention and control.

[0029] 4. This application integrates a mechanism-driven prediction model based on the uncertainty of pollution sources, elevating the system from "current status monitoring" to "future prediction." It can anticipate the changing trends of key water quality parameters (such as peak pollutant levels and minimum dissolved oxygen levels) over a future period, providing valuable lead time for intervention measures. It not only considers physical diffusion laws (fluid dynamic parameters) but also incorporates the uncertainty of pollution sources (dynamic probability data), making the prediction results closer to the complex and ever-changing reality. Based on different pollution source probabilities or control plans, it can simulate water quality evolution under different scenarios, providing quantitative evidence for evaluating the effectiveness of different control strategies and assisting in scientific decision-making.

[0030] 5. The adaptive strategy generation combining knowledge graph and multi-objective optimization in this application enables intelligent and knowledge-driven decision-making and achieves multi-objective collaborative optimization. The strategy is dynamically generated based on real-time predicted water quality data and specific optimization objectives, thus possessing high adaptability and pertinence. It can address different pollution events and control needs, achieving "one policy for one event". Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 This is a flowchart illustrating the steps involved in implementing the method described in this application.

[0033] Figure 2 This is a schematic diagram of the system structure connection of this application. Detailed Implementation

[0034] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0035] Please see Figure 1 As shown, this application provides a water pollution inspection and control method in the first aspect, including: Step 1, multi-source data fusion: generating unified water quality distribution data based on pre-acquired UAV hyperspectral data, unmanned vessel multi-parameter sensor data and fixed monitoring point IoT data.

[0036] In a specific example, the process of generating unified water quality distribution data based on pre-acquired UAV hyperspectral data, UAV multi-parameter sensor data, and fixed monitoring point IoT data includes: performing spatiotemporal alignment on the pre-acquired UAV hyperspectral data, UAV multi-parameter sensor data, and fixed monitoring point IoT data to obtain aligned data; performing feature extraction on the aligned data to generate feature data; and eliminating noise using a weighted fusion calculation formula based on the feature data to generate unified water quality distribution data.

[0037] It should be noted that spatiotemporal alignment involves matching and interpolating UAV hyperspectral data, unmanned surface vessel multi-parameter sensor data, and fixed monitoring point IoT data across both temporal and spatial dimensions. This unifies all data points to the same timestamp and geographic coordinate reference system, eliminating data inconsistencies caused by differences in collection time and location. Furthermore, spatiotemporal alignment fills in missing time points using linear interpolation and maps data from different coordinate systems to a unified coordinate system through geographic coordinate transformation, ensuring data consistency across time and space.

[0038] It should be noted that aligned data is a standardized dataset generated after spatiotemporal alignment processing, eliminating multi-source data sets with temporal and spatial biases. Feature extraction identifies and extracts key parameters representing water quality characteristics from the aligned data, such as dissolved oxygen concentration, pH value, turbidity, and chlorophyll content. Principal component analysis is used to reduce data dimensionality and highlight water quality-related features. Furthermore, feature extraction transforms the raw data into more interpretable feature vectors by calculating spectral reflectance ratios or sensor reading transformations to capture parametric data such as the physical and chemical properties of water quality, thereby generating feature data. The feature data is a parametric dataset generated after feature extraction, which is a vector or matrix that concentrates key water quality information and quantifies various pollution indicators of the water body.

[0039] In a specific example, the step of eliminating noise and generating uniform water quality distribution data based on the feature data using a weighted fusion calculation formula includes: specifically using a calculation formula... Obtain uniform water quality distribution data ,in This is represented by the ID corresponding to the feature data. , This is represented by the number of feature data. Represented as the first Each feature data, Represented as the first The weighting factors corresponding to each feature data.

[0040] It should be noted that the weighting factors corresponding to the feature data are calculated based on the noise variance of the feature data, specifically as follows: ,in Represented as the first The noise standard deviation of the feature data is calculated by statistically analyzing the fluctuations of the feature data in historical data. Specifically, it is obtained by calculating the standard deviation of the data sequence, for example, by statistically analyzing the numerical fluctuations of feature data (such as spectral reflectance or sensor readings) within a preset time window.

[0041] It should be noted that the unified water quality distribution data is an integrated dataset generated after weighted fusion. Its spatial distribution map of water quality, which combines the advantages of multi-source data, reflects the comprehensive pollution status of water bodies.

[0042] This application integrates data from UAVs (high-altitude area coverage), unmanned vessels (surface or underwater linear patrols), and fixed points (continuous monitoring of key locations), overcoming the limitations of single monitoring methods in terms of spatiotemporal resolution. It forms a three-dimensional water quality distribution map of the entire basin with no blind spots and high precision, improving the breadth and dimensionality of data, comprehensively reflecting the water quality status, and laying the foundation for high-precision analysis.

[0043] Step 2, Pollution Source Tracing Analysis: Based on the unified water quality distribution data and the pre-acquired real-time water flow data, analyze the pollutant diffusion path and generate dynamic pollution source probability data.

[0044] In a specific example, the analysis of pollutant diffusion paths and the generation of dynamic pollution source probability data includes: constructing a dynamic graph structure based on the unified water quality distribution data and pre-acquired real-time water flow data to obtain initial graph topology data; using a dynamic adaptive graph neural network model to analyze pollutant diffusion paths based on the initial graph topology data to generate path analysis data; performing probability calculations based on the path analysis data to generate dynamic pollution source probability data; dynamically adjusting the initial graph topology data based on the real-time water flow data to obtain updated graph topology data; and outputting high-precision source tracing results based on the updated graph topology data and the path analysis data.

[0045] It should be noted that the construction of the dynamic graph structure is based on unified water quality distribution data and real-time water flow data. Fixed monitoring points are used as nodes in the graph, and water flow directions are used as edges. Node attributes such as pollutant concentration and edge attributes such as flow velocity weights are assigned to form a graph model representing the spatial relationships of water bodies and the paths of pollutant transport. Furthermore, the dynamic graph structure dynamically updates the connections between nodes by integrating real-time water flow change data, ensuring that the graph topology reflects actual hydrodynamic conditions, thereby generating initial graph topology data. This initial graph topology data is the graph model dataset generated after constructing the dynamic graph structure; it contains a graph structure representation of nodes, edges, and their attributes, used to quantify the spatial network of pollutant diffusion in water bodies.

[0046] It should be noted that pollutant diffusion path analysis uses a dynamic adaptive graph neural network model to process the initial graph topology data through graph convolution and attention mechanisms to simulate the propagation process of pollutants in water bodies, generating path data representing the possible trajectories of pollutants. Furthermore, path analysis captures the dependencies between nodes through graph neural network layers and combines historical diffusion patterns to enhance the accuracy of path prediction, thereby generating path analysis data; this path analysis data is the dataset generated after pollutant diffusion path analysis.

[0047] It should be noted that the probability calculation is based on path analysis data. The softmax function is used to calculate the probability value of each potential pollution source, generating probability data representing the likelihood of each pollution source, and then outputting a dynamically updated probability distribution. This probability distribution dataset is denoted as dynamic pollution source probability data.

[0048] It should be noted that dynamic adjustment is based on real-time water flow data. By updating the edge weights and node connections in the initial graph topology, it reflects the impact of changes in water flow velocity and direction on pollutant diffusion, ensuring that the graph structure is consistent with the actual environment. Furthermore, dynamic adjustment uses adaptive algorithms, such as edge reconnection or node attribute updates based on water flow changes, to optimize the performance of the graph neural network and generate updated graph topology data. The updated graph topology data is the revised graph model dataset generated after dynamic adjustment; it is the optimized graph structure and more accurately represents the pollutant transport network under current hydrodynamic conditions.

[0049] It should be noted that the output of high-precision source tracing results is based on updated graph topology data and path analysis data. Through forward propagation and aggregation operations of graph neural networks, the most likely pollution source location is identified, and pollution source location results are generated.

[0050] This application combines a dynamic pollution source tracing probability model with real-time flow field data, which changes the traditional passive mode that can only monitor the current state of pollution. By coupling water quality distribution with real-time water flow dynamics, it reverse-engineers the diffusion path of pollutants, quickly identifies potential pollution source areas, improves the scientific nature and accuracy of source tracing, and provides targeted targets for precise prevention and control.

[0051] Step 3: Water quality prediction: Based on the dynamic pollution source probability data and preset fluid dynamic parameters, predict the key water quality parameters.

[0052] In a specific example, the prediction of key water quality parameters based on the dynamic pollution source probability data and preset fluid dynamic parameters includes: performing data fusion and standardization processing based on the dynamic pollution source probability data and preset fluid dynamic parameters to obtain model input data; then applying a physical information neural network model for forward propagation calculation to obtain initial water quality parameter data; performing physical constraint verification on the initial water quality parameter data based on preset fluid dynamic equations to obtain water quality residual data; then optimizing and adjusting the initial water quality parameter data to obtain key water quality parameter data; and finally outputting the key water quality parameter data as a predicted trend for future time periods.

[0053] It should be noted that data fusion and standardization, based on dynamic pollution source probability data and fluid dynamics parameters, integrates data from different sources and units into a numerical matrix of a unified format through weighted averaging and normalization methods. This eliminates dimensional differences and enhances data consistency. Fluid dynamics parameters, including flow velocity and diffusion coefficient, describe the characteristics of water movement and provide fundamental parameters for physical constraints. The input data to the physical information neural network model is a standardized numerical matrix that integrates the feature vectors of pollution source probabilities and fluid dynamic conditions.

[0054] In a specific example, the physical constraint verification of the initial water quality parameter data includes: using the calculation formula of the physical constraint loss function. Calculate the total loss value ,in This is expressed as the mean squared error between the predicted data and the training data. It is expressed as the norm of the residuals of the physical equations. It is represented as a weighting coefficient used to balance data loss and physical loss.

[0055] It should be noted that the forward propagation calculation is performed using a physical information neural network model. Specifically, linear transformations and activation function operations are performed through neural network layers to output initial water quality parameter data. The physical information neural network model is a deep learning model that integrates physical equations. Its training process uses a physically constrained loss function to ensure that the output conforms to natural laws. Furthermore, through the convection-diffusion equations... The norm of the residuals of the physical equations is derived. ,in Represented as the residuals of the physical equations, Expressed as pollutant concentration, Indicated as duration, Represented as a flow velocity vector, Represented as the gradient operator, Expressed as diffusion coefficient, Represented as the Laplace operator, This is represented as a pollution source term, which is extracted from dynamic pollution source probability data.

[0056] It should be noted that the effect of the physical equation residuals is to minimize the physical equation residuals to make the predicted trend conform to the actual environmental dynamics. The initial water quality parameter data are predicted values ​​generated after the forward propagation of the physical information neural network model. These predicted values ​​are unverified results, representing water quality parameters at future time points, such as dissolved oxygen concentration and pH value. Physical constraint verification specifically involves calculating the water quality residuals of the initial water quality parameter data based on the fluid dynamics equations. The resulting sequence of water quality residual values ​​is recorded as water quality residual data. Furthermore, the water quality residual data represents the degree of matching between the predicted initial water quality parameter data and the physical equations, quantifying the degree to which the prediction deviates from natural laws.

[0057] It should be noted that the optimization adjustment uses the gradient descent algorithm to iteratively fine-tune the initial water quality parameter data to minimize the water quality residuals, and formats the final predicted values ​​as time series data to generate key water quality parameter data. The key water quality parameter data are adjusted water quality parameter values ​​that conform to physical laws, ensuring that the prediction results are logically sound and reliable from a fluid dynamics perspective.

[0058] This application integrates a mechanism-driven prediction model based on the uncertainty of pollution sources, elevating the system from "current status monitoring" to "future prediction." It can anticipate the changing trends of key water quality parameters (such as peak pollutant levels and minimum dissolved oxygen levels) over a future period, providing valuable lead time for intervention measures. It not only considers physical diffusion laws (fluid dynamic parameters) but also incorporates the uncertainty of pollution sources (dynamic probability data), making the prediction results closer to the complex and ever-changing reality. Based on different pollution source probabilities or control plans, it can simulate water quality evolution under different scenarios, providing quantitative evidence for evaluating the effectiveness of different control strategies and assisting in scientific decision-making.

[0059] Step 4: Generation of Control Strategies: Based on the key water quality parameter data and the preset water pollution knowledge graph, adaptive control strategy data is generated through a multi-objective optimization algorithm.

[0060] In a specific example, the step of generating adaptive control strategy data based on the key water quality parameter data and a preset water pollution knowledge graph using a multi-objective optimization algorithm includes: performing a knowledge graph query operation based on the key water quality parameter data and the preset water pollution knowledge graph to obtain historical case data and real-time feedback data; performing strategy optimization calculations using a multi-objective optimization algorithm to obtain initial control strategy data; and then performing a balance verification operation on the initial control strategy data to obtain adaptive control strategy data.

[0061] It should be noted that the knowledge graph query operation is based on key water quality parameter data and a pre-defined water pollution knowledge graph. It performs node and edge traversal using graph database retrieval technology to match similar historical pollution cases and real-time monitoring feedback information within the knowledge graph. Furthermore, the knowledge graph query operation utilizes the SPARQL query language or a similar graph query protocol, using pollutant concentrations and pH values ​​from the key water quality parameter data as query conditions to retrieve relevant historical treatment plans and real-time environmental change data from the knowledge graph.

[0062] It should be noted that historical case data refers to records of historical water pollution events stored in the knowledge graph, including successful control strategies, their implementation effects, and cost data. Its purpose is to provide a benchmark reference for multi-objective optimization and reduce the randomness of strategy generation. Real-time feedback data is dynamic data simultaneously acquired during knowledge graph query operations. It represents instantaneous sampling of the environmental state and includes real-time water quality change information and control execution feedback from monitoring equipment, ensuring that the strategy adapts to dynamic conditions.

[0063] It should be noted that the balancing verification operation is based on the initial control strategy data. Verification is performed through threshold decisions and consistency checks to ensure a reasonable balance between efficiency and cost. Furthermore, the balancing verification operation uses preset efficiency thresholds and cost upper limit thresholds to compare the efficiency and cost values ​​in the initial control strategy data. When the efficiency value is less than the efficiency threshold or the cost value is greater than the cost upper limit threshold, the strategy is fine-tuned. For example, the weighting factors corresponding to the efficiency function and the cost function are adjusted using an iterative reweighting method, and the strategy is recalculated to generate adaptive control strategy data. Further, the adaptive control strategy data consists of optimized and verified water pollution control schemes, a set of feasible strategies that balance efficiency and cost. Its purpose is to provide instruction data for the execution of coordinated control measures, achieving efficient and economical water pollution management.

[0064] In a specific example, the step of performing policy optimization calculations using a multi-objective optimization algorithm to obtain initial control policy data includes: calculation formulas using a multi-objective optimization algorithm. Initial control strategy data were obtained. ,in Represented as decision variables, Represented as an efficiency function, Represented as a cost function, and These are represented as the weighting factors corresponding to the efficiency function and the cost function, respectively; the decision variables that minimize the weighted sum are obtained through the optimization solver to generate initial control strategy data.

[0065] It should be noted that decision variables are adjustable parameters in multi-objective optimization algorithms (such as gate opening and pump station flow rate). They represent the operational settings of water conservancy control facilities, and physically, are numerical representations of control actions. Their function is to transform abstract strategies into concrete execution instructions. The efficiency function is one of the objective functions in multi-objective optimization algorithms, representing the pollution removal rate or the degree of water quality improvement. Physically, it is a mathematical model of pollution removal efficiency, and its function is to provide efficiency indicators for optimization, ensuring the effectiveness of the strategy. The cost function is another objective function in multi-objective optimization algorithms, representing energy consumption or equipment operating costs. Physically, it is a functional representation of economic cost or energy usage, and its function is to provide cost constraints for optimization, avoiding excessive waste.

[0066] It should be noted that the weighting factors corresponding to the efficiency function and the cost function are balancing parameters in multi-objective optimization algorithms. They represent the trade-off between efficiency and cost, and physically represent the numerical representation of the importance of the objective. Their function is to prioritize efficiency or cost by adjusting the values ​​of these weighting factors, and they are calculated based on the effect-cost ratio from historical case data and resource availability from real-time feedback data. .

[0067] It should be noted that the initial control strategy data is a control scheme generated after strategy optimization calculations, which has not been finally verified.

[0068] The adaptive strategy generation combining knowledge graphs and multi-objective optimization in this application enables intelligent and knowledge-driven decision-making and achieves multi-objective collaborative optimization. The strategy is dynamically generated based on real-time predicted water quality data and specific optimization objectives, thus possessing high adaptability and specificity. It can cope with different pollution events and control needs, achieving "one policy for one event".

[0069] Step 5: Execution of coordinated control: Based on the adaptive control strategy data, control the preset water conservancy control facilities to execute water pollution control.

[0070] In a specific example, controlling a preset water conservancy control facility based on the adaptive control strategy data to perform water pollution control includes: performing instruction parsing operations based on the adaptive control strategy data to obtain facility control parameters; then performing control operations on the preset water conservancy control facility to perform water pollution control, thereby collecting real-time water quality data; and performing parameter dynamic adjustment operations based on the real-time water quality data and the adaptive control strategy data to obtain updated control parameters, thereby performing control operations on the water conservancy control facility to achieve closed-loop execution.

[0071] It should be noted that the command parsing operation is based on adaptive control strategy data. It decodes and extracts specific control commands from the strategy, such as gate opening degree, pump station flow rate, or chemical dosage, to transform the abstract strategy into executable commands. Furthermore, the command parsing operation uses a predefined data format protocol to identify key fields in the strategy data and map them to corresponding facility operation parameters, ensuring the accuracy and operability of the commands. The facility control parameters are the set of parameters generated after the command parsing operation. They represent the specific operational settings of the water conservancy control facilities (numerical representations of control actions), such as opening percentage or flow rate value. Their function is to provide input for the control operation, transforming the strategy into actual control actions.

[0072] It should be noted that control operations, based on facility control parameters, drive water conservancy and regulation facilities through control signals, such as adjusting gate positions, regulating pump station operation status, or controlling chemical dosing devices, to change water flow characteristics or pollutant concentrations. Furthermore, control operations use industrial control protocols to convert digital parameters into physical actions, realizing the physical intervention process for water pollution control. Further, water pollution control is the actual process executed after control operations; it means reducing or eliminating water pollutants through facility operations (human-mediated water purification), directly improving water quality indicators to meet environmental standards.

[0073] It should be noted that real-time data acquisition is based on pre-set monitoring equipment, such as water quality sensors or IoT nodes, to continuously collect water parameters, including dissolved oxygen concentration, pH value, turbidity, or pollutant concentration, to obtain the immediate environmental state after regulation. Furthermore, real-time data acquisition uses data transmission protocols to aggregate sensor readings and align timestamps, generating continuous time-series data. Further, real-time water quality data is the dataset obtained after real-time data acquisition; it represents a set of water quality indicators after regulation is implemented (instantaneous sampled values ​​of the water environment state), serving to provide feedback input for dynamic adjustments and to evaluate the regulation effect and detection deviations.

[0074] It should be noted that the dynamic parameter adjustment operation is based on real-time water quality data and adaptive control strategy data. By comparing the actual water quality indicators with the expected targets of the strategy, the parameter deviation is calculated, and the facility parameters are recalculated using optimization calculations to minimize the deviation and adapt to environmental changes. Furthermore, according to the feedback loop principle, the dynamic parameter adjustment operation applies control theory such as PID regulation or model prediction to iteratively adjust the control parameters to ensure continuous and effective control.

[0075] It should be noted that the updated control parameters are the revised parameter set generated after the dynamic adjustment of parameters. They represent the optimized facility operation settings (the latest instruction values ​​for adaptive control), and their function is to improve control accuracy through closed-loop feedback. Furthermore, the control operation, based on the updated control parameters, drives the water conservancy control facilities again to execute the adjusted control actions, forming a feedback loop that enables the system to respond to environmental changes in real time, maintaining the continuity and adaptability of the control process.

[0076] It should be noted that closed-loop execution is the system behavior achieved after control operation. It means that the regulation process includes a cyclical mechanism of monitoring, adjustment and execution. It is also the realization of a negative feedback control system. Its function is to ensure the continuous optimization, stability and reliability of water pollution regulation and avoid the limitations of single regulation.

[0077] Please see Figure 2 As shown, this application provides a system for a water pollution inspection and control method in its second aspect.

[0078] The system 100 of the water pollution inspection and control method described in this invention can be installed in an electronic device. Depending on the functions implemented, the system 100 may include a multi-source data fusion module 101, a pollution source tracing and analysis module 102, a water quality prediction module 103, a control strategy generation module 104, and a linkage control execution module 105. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0079] In this embodiment, the functions of each module / unit are as follows: The multi-source data fusion module generates unified water quality distribution data based on pre-acquired UAV hyperspectral data, unmanned vessel multi-parameter sensor data, and fixed monitoring point IoT data.

[0080] The pollution source tracing and analysis module analyzes the pollutant diffusion path based on the unified water quality distribution data and the pre-acquired real-time water flow data, and generates dynamic pollution source probability data.

[0081] The water quality prediction module predicts key water quality parameters based on the dynamic pollution source probability data and preset fluid dynamic parameters.

[0082] The regulation strategy generation module generates adaptive regulation strategy data based on the key water quality parameter data and the preset water pollution knowledge graph through a multi-objective optimization algorithm.

[0083] The linkage control execution module, based on the adaptive control strategy data, controls the preset water conservancy control facilities to perform water pollution control.

[0084] This application provides a water pollution inspection and control method and system that integrates multi-source data such as hyperspectral data, unmanned surface vessels, and IoT data to construct a unified high spatiotemporal resolution water quality distribution map, achieving comprehensive three-dimensional and precise perception. Combined with real-time water flow data, it dynamically simulates pollutant diffusion paths and generates pollution source probability distributions, thereby upgrading pollution source tracing from "static speculation" to "dynamic tracking," significantly improving positioning accuracy and efficiency. Simultaneously, it enhances the ability to predict and warn of future changes in key water quality parameters, buying time for early intervention. Based on a water pollution knowledge graph and multi-objective optimization algorithms, it generates scientific and adaptive optimization control strategies, ultimately forming a closed-loop, highly efficient, automated execution system, significantly improving the overall effectiveness of water pollution emergency response and routine control.

[0085] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0086] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0087] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0088] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0089] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0090] 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 it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for water pollution inspection and regulation, characterized in that, The method comprises the following steps: Step one, multi-source data fusion: based on the pre-acquired unmanned aerial vehicle hyperspectral data, unmanned ship multi-parameter sensor data and fixed monitoring point Internet of Things data, unified water quality distribution data is generated; Step two, pollution source analysis: based on the unified water quality distribution data and pre-acquired real-time water flow data, the diffusion path of the pollutant is analyzed, and dynamic pollution source probability data is generated; Step three, water quality prediction: based on the dynamic pollution source probability data and pre-set fluid mechanics parameters, key water quality parameter data is predicted; The prediction of the key water quality parameter data based on the dynamic pollution source probability data and the pre-set fluid mechanics parameters comprises: Based on the dynamic pollution source probability data and the pre-set fluid mechanics parameters, data fusion and standardization processing are performed to obtain model input data, and then a physical information neural network model is applied for forward propagation calculation to obtain initial water quality parameter data; based on a pre-set fluid mechanics equation, the initial water quality parameter data is subjected to physical constraint verification to obtain water quality residual data; then the initial water quality parameter data is adjusted and optimized to obtain key water quality parameter data; and the key water quality parameter data is output as the prediction trend of the future time period; Step four, control strategy generation: based on the key water quality parameter data and a pre-set water pollution knowledge graph, adaptive control strategy data is generated through a multi-objective optimization algorithm; The generation of the adaptive control strategy data based on the key water quality parameter data and the pre-set water pollution knowledge graph through the multi-objective optimization algorithm comprises: Based on the key water quality parameter data and the pre-set water pollution knowledge graph, a knowledge graph query operation is performed to obtain historical case data and real-time feedback data, and a multi-objective optimization algorithm is used for strategy optimization calculation to obtain initial control strategy data; then, the initial control strategy data is subjected to balance verification operation to obtain adaptive control strategy data; The strategy optimization calculation through the multi-objective optimization algorithm to obtain the initial control strategy data comprises: By the calculation formula of the multi-objective optimization algorithm Obtain initial control strategy data Wherein Is expressed as a decision variable, Is expressed as an efficiency function, Is expressed as a cost function, And Corresponding to the weight factor of the efficiency function and the weight factor corresponding to the cost function respectively; obtain the decision variable that makes the weighted sum minimum through the optimization solver to generate the initial control strategy data; Step five, linkage control execution: based on the adaptive control strategy data, a pre-set water conservancy control facility is controlled to execute water pollution control.

2. The method of claim 1, wherein, The generation of the unified water quality distribution data based on the pre-acquired unmanned aerial vehicle hyperspectral data, unmanned ship multi-parameter sensor data and fixed monitoring point Internet of Things data comprises: Based on the pre-acquired unmanned aerial vehicle hyperspectral data, unmanned ship multi-parameter sensor data and fixed monitoring point Internet of Things data, spatio-temporal alignment is performed to obtain alignment data; based on the alignment data, feature extraction is performed to generate feature data; based on the feature data, a weighted fusion calculation formula is used to eliminate noise, and unified water quality distribution data is generated.

3. The method of claim 2, wherein the water pollution inspection and control method is characterized by, The generation of the unified water quality distribution data based on the feature data through the weighted fusion calculation formula to eliminate noise comprises: Specifically, by a calculation formula deriving the unified water quality distribution data wherein is represented as a number corresponding to the characteristic data, , is represented as a number of the characteristic data, is represented as the characteristic data, is represented as the weight factor corresponding to the characteristic data.

4. The method of claim 1, wherein the method further comprises: The analysis of the diffusion path of the pollutant and the generation of the dynamic pollution source probability data comprise: Based on the unified water quality distribution data and the pre-acquired real-time water flow data, a dynamic graph structure is constructed to obtain initial graph topology data; based on the initial graph topology data, a dynamic adaptive graph neural network model is used for pollution diffusion path analysis to generate path analysis data; based on the path analysis data, probability calculation is performed to generate dynamic pollution source probability data; based on the real-time water flow data, the initial graph topology data is dynamically adjusted to obtain updated graph topology data; based on the updated graph topology data and the path analysis data, a high-precision tracing result is output.

5. The method of claim 1, wherein the method further comprises: The physical constraint verification of the initial water quality parameter data includes: The formula for calculating the physical constraint loss function The total loss value is derived where is the mean squared error between the predicted data and the training data, is the norm of the residual of the physical equation, is the weight coefficient used to balance the data loss and the physical loss.

6. The method of claim 1, wherein the method further comprises: The adaptive control strategy data is used to control the preset water conservancy control facility to perform water pollution control, including: Based on the adaptive control strategy data, an instruction analysis operation is performed to obtain facility control parameters; then the preset water conservancy control facility is controlled to perform water pollution control, so as to collect real-time water quality data; based on the real-time water quality data and the adaptive control strategy data, a parameter dynamic adjustment operation is performed to obtain updated control parameters, and then the water conservancy control facility is controlled to realize closed-loop execution.

7. A system for performing the water pollution inspection and regulation method according to any one of claims 1 to 6, characterized in that, It includes: A multi-source data fusion module generates unified water quality distribution data based on pre-acquired unmanned aerial vehicle hyperspectral data, unmanned ship multi-parameter sensor data, and fixed monitoring point Internet of Things data; A pollution tracing analysis module analyzes the diffusion path of pollutants based on the unified water quality distribution data and the pre-acquired real-time water flow data, and generates dynamic pollution source probability data; A water quality prediction module predicts key water quality parameter data based on the dynamic pollution source probability data and pre-set fluid mechanics parameters; A control strategy generation module generates adaptive control strategy data based on the key water quality parameter data and a pre-set water pollution knowledge graph through a multi-objective optimization algorithm; A linkage control execution module controls the preset water conservancy control facility based on the adaptive control strategy data to perform water pollution control.

Images