River water quality parameter monitoring method and system based on deep learning

Abstract

The present invention provides a method and system for monitoring river water quality parameters based on deep learning, which includes self-calibration multi-source data acquisition, diffusive water quality prediction, extreme water quality parameter simulation, reverse diffusion pollution positioning and intelligent monitoring of water quality parameters. The present invention relates to the field of river water quality monitoring technology, specifically to a method and system for monitoring river water quality parameters based on deep learning; by introducing a graph convolutional neural network and a physical diffusion constraint model, the temporal and spatial diffusion trend of pollutants in the river channel is modeled; at the same time, a generative adversarial network and physical verification are combined to construct an extreme pollution event simulation and attribution mechanism; a multimodal Bayesian inversion model and a graph deconvolution structure are further adopted to achieve accurate tracing of pollution sources; the system can dynamically perceive hydrological changes, construct an adaptive threshold judgment mechanism, and realize the prediction, tracking and response to pollution risks, providing efficient and intelligent technical support for river ecological safety management.

Description

Technical Field

[0001] The present invention relates to the technical field of river water quality supervision, and specifically to a river water quality parameter supervision method and system based on deep learning. Background Art

[0002] The deep learning-based river water quality parameter monitoring method and system utilizes deep neural network models to intelligently analyze and predict multi-source river water quality monitoring data. Combining physical diffusion laws, extreme event simulation, and pollution source tracing technology, this comprehensive monitoring approach enables dynamic perception, trend prediction, anomaly identification, and source tracing of river water quality parameters. Its core purpose is to improve the accuracy and timeliness of water quality anomaly warnings, assist environmental management departments in formulating scientific pollution prevention and control strategies, and achieve intelligent and systematic management of river ecosystems.

[0003] However, among the existing intelligent river water quality parameter supervision methods, most of the existing methods remain at the "monitoring-alarm" level, failing to achieve early warning of pollution trends, and the supervision response is often triggered based on a single indicator, ignoring the pollution diffusion chain and source factors, resulting in pollution tracing, diffusion prediction and risk assessment process separation, high model response delay, and inability to achieve linked decision-making under complex hydrological conditions. Technical problems; in the existing water quality prediction methods, there is a problem of insufficient modeling of the river hydrodynamic diffusion process. During heavy rains and floods in rivers, the flow velocity increases sharply and the diffusion distance of pollutants increases, but the model does not take flow velocity changes into account, and the prediction results are still The limitation to the normal diffusion radius leads to a serious underestimation of the warning area, which further leads to technical problems such as the separation of spatial structure and temporal dynamic dependence modeling and the prediction results being inconsistent with physical laws; in the existing water quality parameter simulation methods, there is a lack of coverage of extreme event conditions, the generated results cannot meet the physical consistency constraints, and it is difficult to provide reliable simulation data for abnormal pollution risk drills and regulatory plan deductions; in the existing pollution location methods, there are static tracing or simple reverse simulation, lack of support for multi-source pollution data fusion and uncertainty modeling, and the accuracy of pollution path restoration is limited by data quality and dynamic noise. Summary of the Invention

[0004] In view of the above situation, in order to overcome the defects of the existing technology, the present invention provides a river water quality parameter supervision method and system based on deep learning. In view of the fact that most of the existing intelligent river water quality parameter supervision methods remain at the "monitoring-alarm" level, failing to achieve early warning of pollution trends, and supervision responses are often triggered based on a single indicator, ignoring the pollution diffusion chain and source factors, resulting in pollution tracing, diffusion prediction and risk assessment process separation, high model response delay, and inability to achieve linkage decision-making under complex hydrological conditions. Technical problems, this solution creatively adopts a comprehensive supervision method of integrated prediction, extreme simulation and reverse tracing, realizing technical problems in various hydrological disturbance situations (such as sudden floods, drastic changes in flow rate) It realizes adaptive adjustment of thresholds driven by sensitive factors, so as to quickly identify and generate actionable governance instructions in the early stage of pollution, greatly improving the real-time and scientific nature of basin-level water quality warning and intervention; in view of the problem that the existing water quality prediction methods have insufficient modeling of river hydrodynamic diffusion processes, during the period of heavy rain and flooding in rivers, the flow velocity increases sharply and the diffusion distance of pollutants increases, but the model does not take into account the change in flow velocity, and the prediction results are still limited to the usual diffusion radius, resulting in a serious underestimation of the warning area, which further leads to the technical problems of separate modeling of spatial structure and temporal dynamic dependence and the prediction results are inconsistent with physical laws. This solution creatively adopts the space-time graph convolutional network improved by physical diffusion constraints to perform diffusive water quality prediction. The test not only ensures that the prediction results are physically reasonable, but also improves the generalization ability of the model in the case of data missing or sparse sampling, and significantly enhances the stability of pollution trend evolution modeling; in view of the technical problems that the existing water quality parameter simulation methods lack the ability to cover extreme event conditions, the generated results cannot meet the physical consistency constraints, and it is difficult to provide reliable simulation data for abnormal pollution risk drills and regulatory program deductions, this solution creatively adopts the conditional adversarial generation method combined with physical verification simulation to simulate extreme water quality parameters, realize the causal chain reconstruction of simulated pollution sources and the reasonable inversion of location intensity, thereby effectively supporting advanced regulatory tasks such as simulation inspection, pollution drills and sensitivity analysis; for the existing Some pollution location methods involve static tracing or simple reverse simulation, lack support for multi-source pollution data fusion and uncertainty modeling, and the accuracy of pollution path restoration is limited by technical issues such as data quality and dynamic noise. This solution creatively adopts the standard multimodal Bayesian reverse tracing method to perform reverse diffusion pollution location, achieving high-confidence inversion positioning of pollution sources under the premise of observing pollutant concentrations. At the same time, in conjunction with the constructed graph deconvolution neural network, it enhances the ability to restore the reverse path of pollution in time and space, and ultimately outputs a pollution source probability heat map. By superimposing simulated event probability reference data, it assists regulators in accurately locking in the responsible entities for emissions, thereby improving the pertinence and scientific nature of water environment law enforcement.

[0005] The technical solution adopted by the present invention is as follows: The method for monitoring river water quality parameters based on deep learning provided by the present invention comprises the following steps:

[0006] Step S1: self-calibration multi-source data acquisition;

[0007] Step S2: Diffusive water quality prediction;

[0008] Step S3: extreme water quality parameter simulation;

[0009] Step S4: reverse diffusion pollution positioning;

[0010] Step S5: River water quality parameter supervision.

[0011] Furthermore, in step S1, the self-calibration multi-source data acquisition is used to deploy sensor nodes and collect multi-source water quality data. Specifically, a multi-source spatiotemporal dynamic calibration fusion perception method is adopted to perform data self-calibration through raw data acquisition to obtain spatiotemporal calibration data, including the following steps:

[0012] Step S11: Deployment of a multi-source sensing network, specifically, selecting key collection nodes along the river channel, deploying a sensor sensing network, collecting data, and obtaining original sensor parameter data of the river;

[0013] The key collection nodes include upstream nodes, midstream nodes, downstream nodes, tributary confluence nodes, and outlet nodes; the sensor perception network includes basic water quality sensors, pollutant sensors, and environmental perception sensors; the original river sensor parameter data includes basic water quality parameters, pollutant parameters, and environmental perception auxiliary parameters;

[0014] Step S12: Dynamic calibration, specifically using a standard reference segment matching method based on a sliding time window to perform data self-calibration to obtain spatiotemporal calibration data.

[0015] Furthermore, in step S2, the diffusive water quality prediction is used to predict the dynamic diffusion trend of pollutants in the river. Specifically, based on the spatiotemporal calibration data, a spatiotemporal graph convolutional network improved by physical diffusion constraints is used to perform diffusive water quality prediction to obtain diffusion prediction data, including the following steps:

[0016] Step S21: constructing a watershed topology map, specifically, formally modeling the river channel into a graph structure based on the spatiotemporal calibration data, including graph node definitions and directed edge definitions, and normalizing the river tributaries by introducing virtual river confluence points to obtain watershed topology map data;

[0017] The graph node definition specifically defines each key acquisition node as a monitoring point, and the monitoring point is represented by a graph node; the directed edge definition specifically defines a directed edge according to the flow direction of the river, and sets the edge weight of the directed edge to the inverse of the hydraulic distance of the river;

[0018] Step S22: embedding the diffusion physical coefficient, specifically by introducing the pollutant concentration, diffusion coefficient and water flow velocity parameters, constructing a continuous physical equation, performing integrated physical constraint modeling, and using the diffusion coefficient and water flow velocity parameters in the continuous physical equation as learnable parameters to perform discrete differentiable parameter training for the physical coefficient embedding, and introducing a physical residual regularization loss function to optimize the model training process for diffusive water quality prediction, thereby obtaining an embedded physical operator; the embedded physical operator specifically includes the continuous physical equation and the physical residual regularization loss function;

[0019] Step S23: spatiotemporal prediction modeling, specifically constructing a graph convolutional neural network with mixed spatiotemporal features as a basic prediction subnet for diffusive water quality prediction, and performing model training based on the embedded physical operator to obtain a spatiotemporal prediction model;

[0020] The graph convolutional neural network with mixed spatiotemporal features includes a spatial feature layer and a temporal feature layer; the spatial feature layer specifically uses a standard graph convolution structure to predict the spatial diffusion between river monitoring points; the temporal feature layer specifically uses a causal expansion convolution structure to predict the temporal evolution characteristics between river detection points;

[0021] Step S24: Diffusive water quality prediction, specifically, by constructing the watershed topology map, embedding the diffusion physical coefficients and the spatiotemporal prediction model, based on the spatiotemporal calibration data, using the spatiotemporal prediction model, performing diffusive water quality prediction to obtain diffusion prediction data.

[0022] Furthermore, in step S3, the extreme water quality parameter simulation is used to enhance the system's ability to identify sudden pollution. Specifically, the extreme water quality parameter simulation is performed using a conditional adversarial generation method combined with physical verification simulation to obtain extreme parameter simulation data, including the following steps:

[0023] Step S31: Generate adversarial network construction, specifically by sequentially constructing a standard generator structure and a discriminator structure, and introducing conditional control labels to generate extreme pollution water quality parameters, thereby obtaining a basic network for generating extreme water quality parameters;

[0024] The condition control labels include extreme event type, spatiotemporal location, and meteorological environment conditions;

[0025] The types of extreme events include industrial emissions, agricultural activities, urban activities, meteorological disasters, and man-made accidents;

[0026] The spatiotemporal location, including the river monitoring point number, longitude, latitude and time of the extreme event;

[0027] The meteorological environmental conditions include rainfall intensity, water body temperature, surrounding temperature, wind speed and humidity parameters;

[0028] Step S32: physical verification simulation, specifically, embedding the physical residual regularization loss function as the physical verification loss into the discriminator structure, performing physical consistency correction on the extreme water quality parameters generated by the extreme water quality parameter generation basic network, and obtaining physical verification optimization simulation parameters;

[0029] Step S33: Parameter anomaly attribution optimization, specifically optimizing simulation parameters based on the physical verification, constructing a simulated extreme water quality parameter space and a reverse attribution parameter optimization network, generating parameter attribution vectors, obtaining extreme water quality parameter attribution reference data, and performing manual inspection and simulation verification of extreme water quality parameters based on the extreme water quality parameter attribution reference data;

[0030] The simulated extreme water quality parameter space includes the location of extreme pollution sources, pollution occurrence time, discharge intensity and flow velocity disturbance parameters;

[0031] The reverse attribution parameter optimization network specifically adopts an encoding-decoding structure, adopts a standard temporal graph attention network as the encoder structure, and adopts a multi-head regression output layer as the decoder structure;

[0032] Reference data for attribution of extreme water quality parameters, including pollution type, location of pollution source, time of pollution occurrence, predicted value of pollution emission intensity, predicted value of pollution flow velocity, and actual pollution comparison value;

[0033] Step S34: extreme water quality parameter simulation, specifically, performing extreme water quality parameter simulation through the generative adversarial network construction, the physical verification simulation, and the parameter anomaly attribution optimization to obtain extreme parameter simulation data;

[0034] The extreme parameter simulation data specifically includes simulated pollution concentration parameter sequences, meteorological parameters, human control parameters and simulated physical verification scores.

[0035] Furthermore, in step S4, the reverse diffusion pollution positioning is used to trace the location of the pollution source. Specifically, based on the extreme parameter simulation data and the diffusion prediction data, a standard multimodal Bayesian reverse tracing method is used to perform reverse diffusion pollution positioning to obtain pollution tracing data, including the following steps:

[0036] Step S41: reverse diffusion modeling, specifically constructing a Bayesian inversion problem mathematical model, using the predicted diffusion concentration in the diffusion prediction data as input data, constructing a reverse diffusion model and performing reverse calculation of pollution source parameters to obtain reverse diffusion parameters of the pollution source;

[0037] The reverse diffusion model includes a forward diffusion sub-model and a reverse inversion model; the forward diffusion sub-model is mathematically modeled using the continuous physical equation; the reverse inversion model is mathematically modeled using a standard Bayesian inversion expression and is sampled and optimized using a standard Markov chain Monte Carlo method;

[0038] Step S42: spatiotemporal deconvolution improvement, specifically, constructing a graph deconvolution neural network with mixed spatiotemporal features based on the spatiotemporal prediction model as the basic prediction subnet for reverse diffusion modeling, thereby obtaining a spatiotemporal deconvolution reverse diffusion prediction model;

[0039] The graph deconvolution neural network with mixed spatiotemporal features includes a spatial reverse path layer and a temporal reverse path layer; the spatial reverse path layer specifically uses a standard graph deconvolution structure to predict the reverse pollution path characteristics between river monitoring points; the temporal reverse path layer specifically uses an inverse dilated convolution structure to predict the reverse order feature dependency of the pollution time series between river detection points;

[0040] Step S43: Visualizing the pollution probability, specifically, using a spatiotemporal deconvolution reverse diffusion prediction model to predict the spatiotemporal data of reverse diffusion pollution sources based on the reverse diffusion parameters of the pollution sources, constructing a spatiotemporal distribution probability heat map of the pollution sources, and obtaining reference data for visualization of the pollution probability;

[0041] Step S44: reverse diffusion pollution positioning, specifically, reverse diffusion pollution positioning is performed based on the pollution probability visualization reference data and the pollution source tracing prediction data output by the spatiotemporal deconvolution inverse diffusion prediction model to obtain pollution source tracing data; the pollution source tracing data specifically includes pollution source location parameters, pollution time parameters, emission intensity parameters, pollution probability visualization reference data and extreme pollution event probability reference data.

[0042] Furthermore, in step S5, the river water quality parameter supervision is used to integrate the prediction, extreme simulation and reverse tracing results and realize monitoring and management. Specifically, a dynamic threshold intelligent supervision decision method is adopted to supervise the river water quality parameters and obtain reference data for river water quality supervision decision-making;

[0043] The dynamic threshold intelligent supervision decision-making method is specifically based on the pollution source tracing data and the diffusion prediction data, by introducing environmental characteristic factors to construct environmental sensitivity factors, performing dynamic threshold calculations to obtain dynamic water quality supervision thresholds, and classifying and supervising water quality abnormality events based on the dynamic water quality supervision thresholds;

[0044] The river water quality regulatory decision-making reference data specifically include pollution source tracing instructions, pollution discharge restriction instructions, verification and monitoring instructions and reference ranking of polluted water quality treatment priorities.

[0045] The river water quality parameter monitoring system based on deep learning provided by the present invention includes a self-calibration multi-source data acquisition module, a diffusive water quality prediction module, an extreme water quality parameter simulation module, a reverse diffusion pollution positioning module and a river water quality parameter monitoring module;

[0046] The self-calibration multi-source data acquisition module is used for self-calibration multi-source data acquisition, obtains spatiotemporal calibration data through self-calibration multi-source data acquisition, and sends the spatiotemporal calibration data to the diffusive water quality prediction module;

[0047] The diffusive water quality prediction module is used for diffusive water quality prediction, obtains diffusion prediction data through diffusive water quality prediction, and sends the diffusion prediction data to the inverse diffusion pollution positioning module and the river water quality parameter monitoring module;

[0048] The extreme water quality parameter simulation module is used for simulating extreme water quality parameters, obtaining extreme parameter simulation data through extreme water quality parameter simulation, and sending the extreme parameter simulation data to the river water quality parameter monitoring module;

[0049] The reverse diffusion pollution positioning module is used for reverse diffusion pollution positioning, obtains pollution source tracing data through reverse diffusion pollution positioning, and sends the pollution source tracing data to the river water quality parameter monitoring module;

[0050] The river water quality parameter supervision module is used for river water quality parameter supervision, and obtains river water quality supervision decision-making reference data through river water quality parameter supervision.

[0051] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0052] (1) In view of the technical problems of the existing intelligent river water quality parameter supervision methods, most of the existing methods remain at the "monitoring-alarm" level and fail to achieve early warning of pollution trends. The supervision response is often triggered by a single indicator, ignoring the pollution diffusion chain and source factors, resulting in the separation of pollution tracing, diffusion prediction and risk assessment processes, high model response delay, and inability to achieve joint decision-making under complex hydrological conditions. This solution creatively adopts a comprehensive supervision method of integrated prediction, extreme simulation and reverse tracing, and realizes the adaptive adjustment of thresholds driven by sensitive factors under various hydrological disturbance situations (such as sudden floods and drastic changes in flow rate), thereby quickly identifying and generating actionable governance instructions in the early stages of pollution, greatly improving the real-time and scientific nature of basin-level water quality warning and intervention;

[0053] (2) In view of the problem that the existing water quality prediction methods do not adequately model the river hydrodynamic diffusion process, during the period of heavy rain and flooding in the river, the flow velocity increases sharply and the diffusion distance of pollutants increases. However, because the model does not take the change of flow velocity into account, the prediction results are still limited to the normal diffusion radius, resulting in a serious underestimation of the warning area. This further leads to technical problems such as the separation of spatial structure and temporal dynamic dependence modeling and the prediction results not being consistent with physical laws. This scheme creatively adopts the space-time graph convolution network improved by physical diffusion constraints to perform diffusive water quality prediction. While ensuring the physical rationality of the prediction results, it can also improve the model generalization ability in the case of data missing or sparse sampling, and significantly enhance the stability of pollution trend evolution modeling;

[0054] (3) In view of the technical problems that the existing water quality parameter simulation methods lack the ability to cover extreme event conditions, the generated results cannot meet the physical consistency constraints, and it is difficult to provide reliable simulation data for abnormal pollution risk drills and regulatory program deductions, this solution creatively adopts a conditional adversarial generation method combined with physical verification simulation to simulate extreme water quality parameters, realize the reconstruction of the causal chain of simulated pollution sources and the reasonable inversion of location intensity, thereby effectively supporting advanced regulatory tasks such as simulation inspection, pollution drills and sensitivity analysis;

[0055] (4) In view of the technical problems that the existing pollution location methods have static tracing or simple reverse simulation, lack support for multi-source pollution data fusion and uncertainty modeling, and the accuracy of pollution path restoration is limited by data quality and dynamic noise, this scheme creatively adopts the standard multimodal Bayesian reverse tracing method to perform reverse diffusion pollution location, and achieves high-confidence inversion location of pollution sources under the premise of observing pollutant concentrations. At the same time, in conjunction with the constructed graph deconvolution neural network, it enhances the ability to restore the reverse path of pollution in time and space, and finally outputs a pollution source probability heat map. By superimposing the simulated event probability reference data, it assists regulators in accurately locking the responsible party for emission, thereby improving the pertinence and scientific nature of water environment law enforcement. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A schematic diagram of the flow chart of the river water quality parameter monitoring method based on deep learning provided by the present invention;

[0057] Figure 2 A schematic diagram of the river water quality parameter monitoring system based on deep learning provided by the present invention;

[0058] Figure 3 Schematic diagram of the process of diffusive water quality prediction in step S2;

[0059] Figure 4 This is a schematic diagram of the process of simulating extreme water quality parameters in step S3;

[0060] Figure 5 This is a schematic diagram of the process of reverse diffusion pollution positioning in step S4.

[0061] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0062] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0063] In the description of the present invention, it should be understood that terms such as "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they should not be understood as limiting the present invention.

[0064] Example 1, see Figure 1 The present invention provides a method for monitoring river water quality parameters based on deep learning, which includes the following steps:

[0065] Step S1: self-calibration multi-source data acquisition;

[0066] Step S2: Diffusive water quality prediction;

[0067] Step S3: extreme water quality parameter simulation;

[0068] Step S4: reverse diffusion pollution positioning;

[0069] Step S5: River water quality parameter supervision.

[0070] By performing the above operations, we can address the technical problems of the existing intelligent river water quality parameter supervision methods, most of which remain at the "monitoring-alarm" level and fail to achieve early warning of pollution trends. The supervision response is often triggered based on a single indicator, ignoring the pollution diffusion chain and source factors, resulting in pollution tracing, diffusion prediction and risk assessment process separation, high model response delay, and inability to achieve linked decision-making under complex hydrological conditions. This solution creatively adopts a comprehensive supervision method of integrated prediction, extreme simulation and reverse tracing, and realizes the adaptive adjustment of thresholds driven by sensitive factors under various hydrological disturbance situations (such as sudden floods and drastic changes in flow rate), thereby quickly identifying and generating actionable governance instructions in the early stages of pollution, greatly improving the real-time and scientific nature of basin-level water quality warnings and interventions.

[0071] Example 2, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S1, the self-calibration multi-source data acquisition is used to deploy sensor nodes and collect multi-source water quality data. Specifically, a multi-source spatiotemporal dynamic calibration fusion perception method is adopted to perform data self-calibration through raw data acquisition to obtain spatiotemporal calibration data, including the following steps:

[0072] Step S11: Deployment of a multi-source sensing network, specifically, selecting key collection nodes along the river channel, deploying a sensor sensing network, collecting data, and obtaining original sensor parameter data of the river;

[0073] The key collection nodes include upstream nodes, midstream nodes, downstream nodes, tributary confluence nodes, and outlet nodes; the sensor perception network includes basic water quality sensors, pollutant sensors, and environmental perception sensors; the original river sensor parameter data includes basic water quality parameters, pollutant parameters, and environmental perception auxiliary parameters;

[0074] The basic water quality parameters include dissolved oxygen, temperature, pH value, turbidity, conductivity and redox potential parameters; the pollutant parameters include ammonia nitrogen, total phosphorus, total nitrogen, COD and heavy metal pollution index parameters; the environmental perception auxiliary parameters include rainfall, wind speed and river flow rate parameters;

[0075] Step S12: Dynamic calibration, specifically using a standard reference segment matching method based on a sliding time window to perform data self-calibration to obtain spatiotemporal calibration data.

[0076] Example 3, see Figure 1 、 Figure 2 and Figure 3 This embodiment is based on the above embodiment. In step S2, the diffusive water quality prediction is used to predict the dynamic diffusion trend of pollutants in the river. Specifically, based on the spatiotemporal calibration data, a spatiotemporal graph convolutional network improved by physical diffusion constraints is used to perform diffusive water quality prediction to obtain diffusion prediction data, including the following steps:

[0077] Step S21: constructing a watershed topology map, specifically, formally modeling the river channel into a graph structure based on the spatiotemporal calibration data, including graph node definitions and directed edge definitions, and normalizing the river tributaries by introducing virtual river confluence points to obtain watershed topology map data;

[0078] The graph node definition specifically defines each key acquisition node as a monitoring point, and the monitoring point is represented by a graph node; the directed edge definition specifically defines a directed edge according to the flow direction of the river, and sets the edge weight of the directed edge to the inverse of the hydraulic distance of the river;

[0079] Step S22: embedding the diffusion physical coefficient, specifically by introducing the pollutant concentration, diffusion coefficient and water flow velocity parameters, constructing a continuous physical equation, performing integrated physical constraint modeling, and using the diffusion coefficient and water flow velocity parameters in the continuous physical equation as learnable parameters to perform discrete differentiable parameter training for the physical coefficient embedding, and introducing a physical residual regularization loss function to optimize the model training process for diffusive water quality prediction, thereby obtaining an embedded physical operator; the embedded physical operator specifically includes the continuous physical equation and the physical residual regularization loss function;

[0080] The calculation formula of the continuous physical equation is:

[0081] ;

[0082] Where, The whole is the concentration time partial derivative, which is used to represent the output of the continuous physical equation. C is the pollutant concentration parameter, t is the time index, D is the diffusivity coefficient, and v is the water velocity parameter. is the water velocity vector, is the concentration gradient parameter, R(·) is the chemical reaction identification term;

[0083] The calculation formula of the physical residual regularization loss function is:

[0084] ;

[0085] Where, is the physical residual regularization loss function, is the regularization coefficient, is the predicted pollutant concentration, The overall is the diffusion law correction term, which is used to indicate the part of the model output that conforms to the diffusion law in space. The overall is the river velocity correction term, which is used to represent the part of the spatial propagation trend of the model output that is affected by river water convection;

[0086] Step S23: spatiotemporal prediction modeling, specifically constructing a graph convolutional neural network with mixed spatiotemporal features as a basic prediction subnet for diffusive water quality prediction, and performing model training based on the embedded physical operator to obtain a spatiotemporal prediction model;

[0087] The graph convolutional neural network with mixed spatiotemporal features includes a spatial feature layer and a temporal feature layer; the spatial feature layer specifically uses a standard graph convolution structure to predict the spatial diffusion between river monitoring points; the temporal feature layer specifically uses a causal expansion convolution structure to predict the temporal evolution characteristics between river detection points;

[0088] The modeling calculation formula of the spatial feature layer is:

[0089] ;

[0090] Where H (l+1) It is the output feature matrix of the graph convolution of the l+1 layer, which is used to represent the spatial diffusion prediction characteristics between river monitoring points. is a nonlinear activation function, is the degree matrix of the adjacency matrix, is the self-connected adjacency matrix, H (l) is the output feature matrix of the graph convolution at layer l, W (l) is the spatial feature layer weight of the lth layer;

[0091] The modeling calculation formula of the time feature layer is:

[0092] ;

[0093] Where y t is the output feature vector at the current time t, which is used to represent the predicted value of the time evolution feature between the river detection points. K is the total scale of the causal expansion convolution kernel, which is used to represent the total number of time periods. k is the convolution kernel index, d is the expansion factor parameter, and w k is the convolution kernel weight corresponding to the kth convolution kernel, is an input sequence feature vector, specifically used to represent the watershed topology data;

[0094] Preferably, Table 1 is an example table of parameters of the graph convolutional neural network with mixed spatiotemporal features. As shown in the table, the model type of the spatial feature layer is a standard graph convolution structure, and the parameter types include the number of graph convolution layers, activation function, feature dimension, and random inactivation ratio; the model type of the temporal feature layer is a causal expansion convolution structure, and the parameter types include the number of causal expansion convolution layers, convolution kernel size, expansion factor, activation function, and feature dimension; the basic training parameters for model training are based on the embedded physical operator, and the parameter types include optimizer, learning rate, training batch, and maximum training rounds;

[0095] Table 1 Example of parameters of graph convolutional neural network with mixed spatiotemporal features

[0096]

[0097] Step S24: Diffusive water quality prediction, specifically, by constructing the watershed topology map, embedding the diffusion physical coefficients and the spatiotemporal prediction model, based on the spatiotemporal calibration data, using the spatiotemporal prediction model, performing diffusive water quality prediction to obtain diffusion prediction data.

[0098] By performing the above operations, the problem of insufficient modeling of the river hydrodynamic diffusion process in existing water quality prediction methods is solved. During periods of heavy rain and flooding in rivers, the flow velocity increases sharply and the diffusion distance of pollutants increases. However, because the model does not take into account the change in flow velocity, the prediction results are still limited to the normal diffusion radius, resulting in a serious underestimation of the warning area. This further leads to technical problems such as the separation of spatial structure and temporal dynamic dependence modeling and the inconsistency of prediction results with physical laws. This solution creatively uses a spatiotemporal graph convolutional network improved by physical diffusion constraints to perform diffusive water quality prediction. While ensuring the physical rationality of the prediction results, it can also improve the model generalization ability in the case of data missing or sparse sampling, and significantly enhance the stability of pollution trend evolution modeling.

[0099] Example 4, see Figure 1 、 Figure 2 and Figure 4 This embodiment is based on the above embodiment. In step S3, the extreme water quality parameter simulation is used to enhance the system's ability to identify sudden pollution. Specifically, the extreme water quality parameter simulation is performed using a conditional adversarial generation method combined with physical verification simulation to obtain extreme parameter simulation data, including the following steps:

[0100] Step S31: Generate adversarial network construction, specifically by sequentially constructing a standard generator structure and a discriminator structure, and introducing conditional control labels to generate extreme pollution water quality parameters, thereby obtaining a basic network for generating extreme water quality parameters;

[0101] The standard generator structure adopts a 5-layer transposed convolutional network and generates a sequence of simulated pollution parameters;

[0102] The discriminator structure adopts a standard three-dimensional convolutional network and outputs a probability value of data authenticity;

[0103] The condition control labels include extreme event type, spatiotemporal location, and meteorological environment conditions;

[0104] The types of extreme events include industrial emissions, agricultural activities, urban activities, meteorological disasters, and man-made accidents;

[0105] The spatiotemporal location, including the river monitoring point number, longitude, latitude and time of the extreme event;

[0106] The meteorological environmental conditions include rainfall intensity, water body temperature, surrounding temperature, wind speed and humidity parameters;

[0107] Preferably, the industrial emissions category specifically includes industrial leakage, illegal discharge and sudden increase in sewage load; the agricultural activities category includes fertilizer leakage, pesticide seepage and livestock and poultry breeding wastewater leakage; the urban activities category includes urban drainage network pollution, rainwater and sewage mixing pollution and urban sewage pipe leakage; the meteorological disaster category includes heavy rain, drought and flood backflow; the man-made sudden accidents category includes vehicle accidents, illegal mud discharge into rivers and poisoning pollution;

[0108] Step S32: physical verification simulation, specifically, embedding the physical residual regularization loss function as the physical verification loss into the discriminator structure, performing physical consistency correction on the extreme water quality parameters generated by the extreme water quality parameter generation basic network, and obtaining physical verification optimization simulation parameters;

[0109] Step S33: Parameter anomaly attribution optimization, specifically optimizing simulation parameters based on the physical verification, constructing a simulated extreme water quality parameter space and a reverse attribution parameter optimization network, generating parameter attribution vectors, obtaining extreme water quality parameter attribution reference data, and performing manual inspection and simulation verification of extreme water quality parameters based on the extreme water quality parameter attribution reference data;

[0110] The simulated extreme water quality parameter space includes the location of extreme pollution sources, pollution occurrence time, discharge intensity and flow velocity disturbance parameters;

[0111] The reverse attribution parameter optimization network specifically adopts an encoding and decoding structure, and adopts a standard temporal graph attention network as the encoder structure and a multi-head regression output layer as the decoder structure;

[0112] Reference data for attribution of extreme water quality parameters, including pollution type, location of pollution source, time of pollution occurrence, predicted value of pollution emission intensity, predicted value of pollution flow velocity, and actual pollution comparison value;

[0113] Step S34: extreme water quality parameter simulation, specifically, performing extreme water quality parameter simulation through the generative adversarial network construction, the physical verification simulation, and the parameter anomaly attribution optimization to obtain extreme parameter simulation data;

[0114] The extreme parameter simulation data specifically includes simulated pollution concentration parameter sequences, meteorological parameters, human control parameters and simulated physical verification scores.

[0115] By performing the above operations, in order to address the technical problems in existing water quality parameter simulation methods, such as the lack of coverage of extreme event conditions, the inability of generated results to meet physical consistency constraints, and the difficulty in providing reliable simulation data for abnormal pollution risk drills and regulatory program deductions, this solution creatively adopts a conditional adversarial generation method combined with physical verification simulation to simulate extreme water quality parameters, realize the reconstruction of the causal chain of simulated pollution sources and the reasonable inversion of location intensity, thereby effectively supporting advanced regulatory tasks such as simulation inspection, pollution drills and sensitivity analysis.

[0116] Example 5, see Figure 1 、 Figure 2 and Figure 5 This embodiment is based on the above embodiment. In step S4, the reverse diffusion pollution positioning is used to trace the location of the pollution source. Specifically, based on the extreme parameter simulation data and the diffusion prediction data, a standard multimodal Bayesian reverse tracing method is used to perform reverse diffusion pollution positioning to obtain pollution tracing data, including the following steps:

[0117] Step S41: reverse diffusion modeling, specifically constructing a Bayesian inversion problem mathematical model, using the predicted diffusion concentration in the diffusion prediction data as input data, constructing a reverse diffusion model and performing reverse calculation of pollution source parameters to obtain reverse diffusion parameters of the pollution source;

[0118] The reverse diffusion model includes a forward diffusion sub-model and a reverse inversion model; the forward diffusion sub-model is mathematically modeled using the continuous physical equation; the reverse inversion model is mathematically modeled using a standard Bayesian inversion expression and is sampled and optimized using a standard Markov chain Monte Carlo method;

[0119] The calculation formula of the standard Bayesian inversion expression is:

[0120] ;

[0121] Where, The overall probability distribution term is used to represent the predicted diffusion concentration C under known conditions. obs Under the premise of The probability distribution of , where is the pollution source parameter, C obs is the predicted diffusion concentration in the diffusion prediction data, is the proportional sign, The overall likelihood function term is used to represent the pollution source parameter Get the predicted diffusion concentration C obs The probability of It is a priori probability distribution term, which is used to express the priori probability value of the pollution source parameter directly without introducing the predicted diffusion concentration;

[0122] Step S42: spatiotemporal deconvolution improvement, specifically, constructing a graph deconvolution neural network with mixed spatiotemporal features based on the spatiotemporal prediction model as the basic prediction subnet for reverse diffusion modeling, thereby obtaining a spatiotemporal deconvolution reverse diffusion prediction model;

[0123] The graph deconvolution neural network with mixed spatiotemporal features includes a spatial reverse path layer and a temporal reverse path layer; the spatial reverse path layer specifically uses a standard graph deconvolution structure to predict the reverse pollution path characteristics between river monitoring points; the temporal reverse path layer specifically uses an inverse dilated convolution structure to predict the reverse order feature dependency of the pollution time series between river detection points;

[0124] Step S43: Visualizing the pollution probability, specifically, using a spatiotemporal deconvolution reverse diffusion prediction model to predict the spatiotemporal data of reverse diffusion pollution sources based on the reverse diffusion parameters of the pollution sources, constructing a spatiotemporal distribution probability heat map of the pollution sources, and obtaining reference data for visualization of the pollution probability;

[0125] Step S44: reverse diffusion pollution positioning, specifically, reverse diffusion pollution positioning is performed based on the pollution probability visualization reference data and the pollution source tracing prediction data output by the spatiotemporal deconvolution inverse diffusion prediction model to obtain pollution source tracing data; the pollution source tracing data specifically includes pollution source location parameters, pollution time parameters, emission intensity parameters, pollution probability visualization reference data and extreme pollution event probability reference data.

[0126] By performing the above operations, in order to address the technical problems in existing pollution location methods, such as static tracing or simple reverse simulation, lack of support for multi-source pollution data fusion and uncertainty modeling, and the accuracy of pollution path restoration is limited by data quality and dynamic noise, this solution creatively adopts the standard multimodal Bayesian reverse tracing method to perform reverse diffusion pollution location, achieving high-confidence inversion positioning of pollution sources under the premise of observing pollutant concentrations. At the same time, in conjunction with the constructed graph deconvolution neural network, the reverse path restoration capability of pollution in time and space is enhanced, and finally a pollution source probability heat map is output. By superimposing simulated event probability reference data, it assists regulators in accurately locking in the responsible entities for emissions, thereby improving the targeted nature and scientific nature of water environment law enforcement.

[0127] Example 6, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S5, the river water quality parameter supervision is used to integrate prediction, extreme simulation and reverse tracing results and realize monitoring and management. Specifically, a dynamic threshold intelligent supervision decision-making method is used to supervise river water quality parameters and obtain reference data for river water quality supervision decision-making.

[0128] The dynamic threshold intelligent supervision decision-making method is specifically based on the pollution source tracing data and the diffusion prediction data, by introducing environmental characteristic factors to construct environmental sensitivity factors, performing dynamic threshold calculations to obtain dynamic water quality supervision thresholds, and classifying and supervising water quality abnormality events based on the dynamic water quality supervision thresholds;

[0129] The calculation formula for the dynamic threshold calculation is:

[0130] ;

[0131] Where, is the dynamic water quality regulation threshold, is the sliding mean of the i-th water quality parameter in the historical data of the current basin, is the standard deviation of the i-th water quality parameter in the historical data of the current basin, where i is the water quality parameter index, which is used to represent the water quality parameters in the pollution source tracing data and the diffusion prediction data. It is the environmental sensitivity factor, which is calculated by introducing temperature, flow rate and rainfall parameters and constructing a lightweight multi-layer perceptron to adjust the training conditions. The specific calculation formula of the environmental sensitivity factor is: , where MLP(·) is a lightweight multilayer perceptron representation function, using a three-layer fully connected network, the input is set as the environmental parameter vector, the output is set as the scalar adjustment factor, Rain(t) is the rainfall parameter, Flow(t) is the flow rate parameter, and Temperature(t) is the temperature parameter;

[0132] The river water quality regulatory decision-making reference data specifically include pollution source tracing instructions, pollution discharge restriction instructions, verification and monitoring instructions and reference ranking of polluted water quality treatment priorities.

[0133] Example 7, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. The river water quality parameter monitoring system based on deep learning provided by the present invention includes a self-calibration multi-source data acquisition module, a diffusive water quality prediction module, an extreme water quality parameter simulation module, a reverse diffusion pollution positioning module and a river water quality parameter monitoring module;

[0134] The self-calibration multi-source data acquisition module is used for self-calibration multi-source data acquisition, obtains spatiotemporal calibration data through self-calibration multi-source data acquisition, and sends the spatiotemporal calibration data to the diffusive water quality prediction module;

[0135] The diffusive water quality prediction module is used for diffusive water quality prediction, obtains diffusion prediction data through diffusive water quality prediction, and sends the diffusion prediction data to the inverse diffusion pollution positioning module and the river water quality parameter monitoring module;

[0136] The extreme water quality parameter simulation module is used for simulating extreme water quality parameters, obtaining extreme parameter simulation data through extreme water quality parameter simulation, and sending the extreme parameter simulation data to the river water quality parameter monitoring module;

[0137] The reverse diffusion pollution positioning module is used for reverse diffusion pollution positioning, obtains pollution source tracing data through reverse diffusion pollution positioning, and sends the pollution source tracing data to the river water quality parameter monitoring module;

[0138] The river water quality parameter supervision module is used for river water quality parameter supervision, and obtains river water quality supervision decision-making reference data through river water quality parameter supervision.

[0139] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0140] While the embodiments of the present invention have been shown and described, it will be apparent to those skilled in the art that various changes, modifications, substitutions, and alterations can be made to the embodiments without departing from the principles and spirit of the invention.

[0141] The present invention and its embodiments are described above. This description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs structures and embodiments similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. A river water quality parameter monitoring method based on deep learning, characterized by: The method comprises the following steps: Step S1: self-calibration multi-source data acquisition to obtain spatiotemporal calibration data; Step S2: Diffusive water quality prediction, using a spatiotemporal graph convolutional network improved by physical diffusion constraints to perform diffusive water quality prediction and obtain diffusion prediction data, including the following steps: Step S21: Construction of watershed topology map; Step S22: Embedding of diffusion physical coefficients, by introducing pollutant concentration, diffusion coefficient and water flow velocity parameters, constructing continuous physical equations, and performing integrated physical constraint modeling; Step S23: Spatiotemporal prediction modeling; Step S24: Diffusive water quality prediction; Step S3: extreme water quality parameter simulation, using a conditional adversarial generation method combined with physical verification simulation to simulate extreme water quality parameters and obtain extreme parameter simulation data; Step S4: reverse diffusion pollution positioning, using the standard multimodal Bayesian reverse tracing method to perform reverse diffusion pollution positioning and obtain pollution tracing data, including the following steps: step S41: reverse diffusion modeling; step S42: spatiotemporal deconvolution improvement, specifically, based on the spatiotemporal prediction model, constructing a spatiotemporal feature mixed graph deconvolution neural network as the basic prediction subnet for reverse diffusion modeling, and obtaining a spatiotemporal deconvolution reverse diffusion prediction model; the spatiotemporal feature mixed graph deconvolution neural network includes a spatial reverse path layer and a temporal reverse path layer; the spatial reverse path layer specifically adopts a standard graph deconvolution structure to predict the reverse pollution path characteristics between river monitoring points; the temporal reverse path layer specifically adopts a reverse dilation convolution structure to predict the reverse feature dependency of the pollution time series between river detection points; step S43: pollution probability visualization; step S44: reverse diffusion pollution positioning; Step S5: River water quality parameter supervision, using a dynamic threshold intelligent supervision decision-making method to supervise river water quality parameters and obtain reference data for river water quality supervision decision-making.

2. The method for monitoring river water quality parameters based on deep learning according to claim 1, characterized in that: In step S1, the self-calibration multi-source data acquisition is used to deploy sensor nodes and collect multi-source water quality data. Specifically, a multi-source spatiotemporal dynamic calibration fusion perception method is used to perform data self-calibration through raw data acquisition to obtain spatiotemporal calibration data, including the following steps: Step S11: Deployment of a multi-source sensing network, specifically, selecting key collection nodes along the river channel, deploying a sensor sensing network, collecting data, and obtaining original sensor parameter data of the river; The key collection nodes include upstream nodes, midstream nodes, downstream nodes, tributary confluence nodes and outlet nodes; the sensor perception network includes basic water quality sensors, pollutant sensors and environmental perception sensors; the original river sensor parameter data includes basic water quality parameters, pollutant parameters and environmental perception auxiliary parameters; Step S12: Dynamic calibration, specifically using a standard reference segment matching method based on a sliding time window to perform data self-calibration to obtain spatiotemporal calibration data.

3. The method for monitoring river water quality parameters based on deep learning according to claim 2 is characterized in that: In step S2, the diffusive water quality prediction is used to predict the dynamic diffusion trend of pollutants in the river. Specifically, based on the spatiotemporal calibration data, a spatiotemporal graph convolutional network improved by physical diffusion constraints is used to perform diffusive water quality prediction to obtain diffusion prediction data, including the following steps: Step S21: constructing a watershed topology map, specifically, formally modeling the river channel into a graph structure based on the spatiotemporal calibration data, including graph node definitions and directed edge definitions, and normalizing the river tributaries by introducing virtual river confluence points to obtain watershed topology map data; Step S22: embedding the diffusion physical coefficient, specifically by introducing the pollutant concentration, diffusion coefficient and water flow velocity parameters, constructing a continuous physical equation, performing integrated physical constraint modeling, and using the diffusion coefficient and water flow velocity parameters in the continuous physical equation as learnable parameters to perform discrete differentiable parameter training for the physical coefficient embedding, and introducing a physical residual regularization loss function to optimize the model training process for diffusive water quality prediction, thereby obtaining an embedded physical operator; the embedded physical operator specifically includes the continuous physical equation and the physical residual regularization loss function; Step S23: spatiotemporal prediction modeling, specifically constructing a graph convolutional neural network with mixed spatiotemporal features as a basic prediction subnet for diffusive water quality prediction, and performing model training based on the embedded physical operator to obtain a spatiotemporal prediction model; The graph convolutional neural network with mixed spatiotemporal features includes a spatial feature layer and a temporal feature layer; the spatial feature layer specifically uses a standard graph convolution structure to predict the spatial diffusion between river monitoring points; the temporal feature layer specifically uses a causal expansion convolution structure to predict the temporal evolution characteristics between river detection points; Step S24: Diffusive water quality prediction, specifically, by constructing the watershed topology map, embedding the diffusion physical coefficients and the spatiotemporal prediction model, based on the spatiotemporal calibration data, using the spatiotemporal prediction model, performing diffusive water quality prediction to obtain diffusion prediction data.

4. The method for monitoring river water quality parameters based on deep learning according to claim 3 is characterized in that: In step S3, the extreme water quality parameter simulation is used to enhance the system's ability to identify sudden pollution. Specifically, the extreme water quality parameter simulation is performed using a conditional adversarial generation method combined with physical verification simulation to obtain extreme parameter simulation data, including the following steps: Step S31: Generate adversarial network construction, specifically by sequentially constructing a standard generator structure and a discriminator structure, and introducing conditional control labels to generate extreme pollution water quality parameters, thereby obtaining a basic network for generating extreme water quality parameters; The condition control labels include extreme event type, spatiotemporal location, and meteorological environment conditions; The types of extreme events include industrial emissions, agricultural activities, urban activities, meteorological disasters, and man-made accidents; Step S32: physical verification simulation, specifically, embedding the physical residual regularization loss function as the physical verification loss into the discriminator structure, performing physical consistency correction on the extreme water quality parameters generated by the extreme water quality parameter generation basic network, and obtaining physical verification optimization simulation parameters; Step S33: Parameter anomaly attribution optimization, specifically optimizing simulation parameters based on the physical verification, constructing a simulated extreme water quality parameter space and a reverse attribution parameter optimization network, generating parameter attribution vectors, obtaining extreme water quality parameter attribution reference data, and performing manual inspection and simulation verification of extreme water quality parameters based on the extreme water quality parameter attribution reference data; The simulated extreme water quality parameter space includes the location of extreme pollution sources, pollution occurrence time, discharge intensity and flow velocity disturbance parameters; The reverse attribution parameter optimization network specifically adopts an encoding-decoding structure, adopts a standard temporal graph attention network as the encoder structure, and adopts a multi-head regression output layer as the decoder structure; Reference data for attribution of extreme water quality parameters, including pollution type, pollution source location, pollution occurrence time, pollution emission intensity prediction value, pollution flow velocity prediction value, and actual pollution comparison value; Step S34: extreme water quality parameter simulation, specifically, performing extreme water quality parameter simulation through the generative adversarial network construction, the physical verification simulation and the parameter anomaly attribution optimization to obtain extreme parameter simulation data.

5. The method for monitoring river water quality parameters based on deep learning according to claim 4 is characterized in that: In step S34, the extreme parameter simulation data specifically includes simulated pollution concentration parameter sequence, meteorological parameters, human control parameters and simulated physical verification scores.

6. The method for monitoring river water quality parameters based on deep learning according to claim 5, characterized in that: In step S4, the reverse diffusion pollution positioning is used to trace the location of the pollution source. Specifically, based on the extreme parameter simulation data and the diffusion prediction data, a standard multimodal Bayesian reverse tracing method is used to perform reverse diffusion pollution positioning to obtain pollution tracing data, including the following steps: Step S41: reverse diffusion modeling, specifically constructing a Bayesian inversion problem mathematical model, using the predicted diffusion concentration in the diffusion prediction data as input data, constructing a reverse diffusion model and performing reverse calculation of pollution source parameters to obtain reverse diffusion parameters of the pollution source; The reverse diffusion model includes a forward diffusion sub-model and a reverse inversion model; the forward diffusion sub-model is mathematically modeled using the continuous physical equation; the reverse inversion model is mathematically modeled using a standard Bayesian inversion expression and is sampled and optimized using a standard Markov chain Monte Carlo method; Step S42: spatiotemporal deconvolution improvement; Step S43: Visualizing the pollution probability, specifically, using a spatiotemporal deconvolution reverse diffusion prediction model to predict the spatiotemporal data of reverse diffusion pollution sources based on the reverse diffusion parameters of the pollution sources, constructing a spatiotemporal distribution probability heat map of the pollution sources, and obtaining reference data for visualization of the pollution probability; Step S44: reverse diffusion pollution positioning, specifically, reverse diffusion pollution positioning is performed based on the pollution probability visualization reference data and the pollution source tracing prediction data output by the spatiotemporal deconvolution inverse diffusion prediction model to obtain pollution source tracing data; the pollution source tracing data specifically includes pollution source location parameters, pollution time parameters, emission intensity parameters, pollution probability visualization reference data and extreme pollution event probability reference data.

7. The method for monitoring river water quality parameters based on deep learning according to claim 6, characterized in that: In step S5, the river water quality parameter supervision is used to integrate prediction, extreme simulation and reverse tracing results and realize monitoring and management. Specifically, a dynamic threshold intelligent supervision decision-making method is used to supervise the river water quality parameters and obtain reference data for river water quality supervision decision-making; The dynamic threshold intelligent supervision decision-making method is specifically based on the pollution tracing data and the diffusion prediction data. By introducing environmental characteristic factors to construct environmental sensitivity factors, dynamic threshold calculation is performed to obtain dynamic water quality supervision thresholds, and water quality abnormality events are classified and supervised based on the dynamic water quality supervision thresholds; the river water quality supervision decision-making reference data specifically includes pollution tracing instructions, pollution discharge restriction instructions, verification and monitoring instructions and reference ranking of polluted water quality treatment priorities.

8. A river water quality parameter monitoring system based on deep learning, for implementing the river water quality parameter monitoring method based on deep learning as described in any one of claims 1 to 7, characterized in that: It includes a self-calibration multi-source data acquisition module, a diffusive water quality prediction module, an extreme water quality parameter simulation module, a reverse diffusion pollution positioning module and a river water quality parameter supervision module.

9. The river water quality parameter monitoring system based on deep learning according to claim 8 is characterized by: The self-calibration multi-source data acquisition module is used for self-calibration multi-source data acquisition, obtains spatiotemporal calibration data through self-calibration multi-source data acquisition, and sends the spatiotemporal calibration data to the diffusive water quality prediction module; The diffusive water quality prediction module is used for diffusive water quality prediction, obtains diffusion prediction data through diffusive water quality prediction, and sends the diffusion prediction data to the inverse diffusion pollution positioning module and the river water quality parameter monitoring module; The extreme water quality parameter simulation module is used for simulating extreme water quality parameters, obtaining extreme parameter simulation data through extreme water quality parameter simulation, and sending the extreme parameter simulation data to the river water quality parameter monitoring module; The reverse diffusion pollution positioning module is used for reverse diffusion pollution positioning, obtains pollution source tracing data through reverse diffusion pollution positioning, and sends the pollution source tracing data to the river water quality parameter monitoring module; The river water quality parameter supervision module is used for river water quality parameter supervision, and obtains river water quality supervision decision-making reference data through river water quality parameter supervision.

Images