Method and device for tracing pollution source of sewage outlet of river, electronic equipment and storage medium

Abstract

The application discloses a river pollution outlet pollution source tracing method and device, electronic equipment and a storage medium, comprising: acquiring current water quality parameters, current flow rate data and unmanned aerial vehicle images of a plurality of river pollution outlets in a target river basin; identifying the time sequence change of the current water quality parameters through a long short-term memory network to obtain predicted water quality parameters at a future time, and judging whether a pollution source exists in the target river basin according to the predicted water quality parameters; when the pollution source exists in the target river basin, inputting the current water quality parameters, the current flow rate data and the unmanned aerial vehicle images into a lightweight pollution source identification model deployed on an edge node to obtain a pollution source probability distribution of the target river basin; and determining a target river pollution outlet where the pollution source is located according to the pollution source probability distribution. The application improves the accuracy of pollution source identification and can be widely applied in the field of artificial intelligence technology.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method, device, electronic equipment and storage medium for tracing pollution sources at river sewage outlets. Background Art

[0002] With the acceleration of industrialization and rising urbanization, river outfalls have become a significant source of environmental pollution. Traditional methods for monitoring and tracing pollution sources at river outfalls rely primarily on traditional sensor networks, manual inspections, and sampling. This approach, with its long data collection, transmission, and processing links, is unable to rapidly respond to sudden pollution incidents. Furthermore, source tracing is inefficient, relies on manual decision-making, lacks AI-driven dynamic adjustment and prediction capabilities, and lacks intelligence.

[0003] In recent years, the rapid development of the Internet of Things and AI technologies has provided new solutions for monitoring and tracing pollution sources at river outfalls. Existing solutions rely on fixed models to match pollution sources, failing to dynamically predict pollution source probabilities by integrating historical data. These solutions result in long investigation cycles and low data utilization. High-frequency sensor data creates significant storage pressure, while low-frequency data collection makes it difficult to capture sudden changes in water quality. Furthermore, the lack of integrated analysis of multi-source data (such as water quality parameters, flow rate data, and images) hinders the accuracy of pollution source tracing.

[0004] Explanation of terms:

[0005] LSTM (Long Short-Term Memory): A specialized recurrent neural network (RNN) designed to address the vanishing and exploding gradient problems that plague traditional RNNs when processing long sequences of data. By introducing "memory cells" and a "gating mechanism," LSTM effectively captures long-term dependencies in time series and is widely used in fields such as natural language processing and speech recognition.

[0006] ST-GCN (Spatial-Temporal Graph Convolutional Networks): is a deep learning framework that combines graph convolutional networks with time series analysis, mainly used to process data with spatiotemporal correlation.

[0007] TinyML (Tiny Machine Learning) is a machine learning technology designed for ultra-low-power, resource-constrained embedded devices (such as microcontrollers) to enable real-time intelligent decision-making at the edge. Summary of the Invention

[0008] The purpose of the present invention is to solve one of the technical problems existing in the prior art to at least a certain extent.

[0009] To this end, an object of an embodiment of the present invention is to provide a method for tracing pollution sources at river sewage outlets, which improves the accuracy of pollution source identification.

[0010] Another object of an embodiment of the present invention is to provide a pollution source tracing device for a river sewage outlet.

[0011] In order to achieve the above technical objectives, the technical solutions adopted by the embodiments of the present invention include:

[0012] In one aspect, an embodiment of the present invention provides a method for tracing pollution sources at a river sewage outlet, comprising the following steps:

[0013] Obtain current water quality parameters, current flow rate data, and drone imagery for multiple river outfalls in the target basin;

[0014] Identifying the temporal changes of the current water quality parameters through a long short-term memory network to obtain predicted water quality parameters at future moments, and determining whether there is a pollution source in the target watershed based on the predicted water quality parameters;

[0015] When there is a pollution source in the target watershed, the current water quality parameters, the current flow rate data, and the drone image are input into a lightweight pollution source identification model deployed on the edge node to obtain a probability distribution of the pollution source in the target watershed;

[0016] Determine the target river sewage outlet where the pollution source is located according to the probability distribution of the pollution source;

[0017] The lightweight pollution source identification model is obtained by jointly training and knowledge distilling the pollution diffusion spatiotemporal graph model and the Bayesian network.

[0018] Furthermore, in one embodiment of the present invention, obtaining current water quality parameters, current flow rate data, and drone images of multiple river sewage outlets in the target basin specifically includes:

[0019] Acquiring the current water quality parameters through water quality sensors installed at the sewage outlets of each river;

[0020] Acquiring the current flow rate data by using a flow rate sensor provided at each of the river sewage outlets;

[0021] Regularly taking drone images of the sewage outlets of each river by drone patrol;

[0022] The current water quality parameter, the current flow rate data, and the drone image are transmitted to the edge node.

[0023] Furthermore, in one embodiment of the present invention, identifying the temporal changes of the current water quality parameters through the long short-term memory network to obtain the predicted water quality parameters at a future time, and determining whether the target watershed has a pollution source based on the predicted water quality parameters, specifically includes:

[0024] Generating a time-series water quality parameter sequence of each of the river sewage outlets according to the current water quality parameters;

[0025] Inputting the time series water quality parameter sequence into the pre-trained long short-term memory network to obtain the predicted water quality parameters of each river sewage outlet at a future time;

[0026] Calculating a water quality residual based on the predicted water quality parameter and the current water quality parameter, and calculating a mean and a standard deviation of the water quality residual in a preset residual sliding window;

[0027] determining a dynamic threshold value according to the mean and the standard deviation;

[0028] When the duration during which the water quality residual is greater than the dynamic threshold exceeds a preset duration threshold, it is determined that a pollution source exists in the target watershed.

[0029] Furthermore, in one embodiment of the present invention, the lightweight pollution source identification model is obtained by the following steps:

[0030] Obtaining historical pollution event data for the target river basin, the historical pollution event data including historical water quality parameters, historical flow rate data, historical drone images, and historical pollution source probabilities for each of the river sewage outlets;

[0031] Constructing a pollution diffusion spatiotemporal graph model with the river sewage outlets as nodes and the water flow directions as edges, inputting the historical pollution event data into the pollution diffusion spatiotemporal graph model to obtain the predicted pollution source probability and the predicted pollutant diffusion path of each river sewage outlet;

[0032] Determining a graph reconstruction loss of the pollution diffusion spatiotemporal graph model according to the predicted pollution source probability and the historical pollution source probability;

[0033] Inputting the historical pollution event data and the predicted pollutant diffusion path into a Bayesian network to obtain the posterior probability of the pollution source of each of the river sewage outlets;

[0034] Determining a posterior probability matching loss of the Bayesian network according to the posterior probability of the pollution source and the historical pollution source probability;

[0035] Determining a joint loss function based on the graph reconstruction loss and the posterior probability matching loss, and updating the parameters of the pollution diffusion spatiotemporal graph model and the Bayesian network through a backpropagation algorithm based on the joint loss function to obtain a trained pollution source identification model;

[0036] The lightweight pollution source identification model is obtained by performing knowledge distillation on the pollution source identification model.

[0037] Furthermore, in one embodiment of the present invention, inputting the historical pollution event data into the pollution diffusion spatiotemporal graph model to obtain the predicted pollution source probability and the predicted pollutant diffusion path of each of the river sewage outlets specifically includes:

[0038] Determining a multimodal embedding vector for each node based on the historical water quality parameters, the historical flow velocity data, the historical drone images, and the historical pollution source probabilities;

[0039] Determine the edge weight of each edge according to the historical flow rate data of adjacent nodes, the historical unmanned image, and the node distance;

[0040] Performing a preset number of information transfers on each node to obtain the updated multimodal embedding vector of each node;

[0041] The predicted pollution source probability and the predicted pollutant diffusion path corresponding to each node are determined according to the updated multimodal embedding vector.

[0042] Furthermore, in one embodiment of the present invention, the inputting of the historical pollution event data and the predicted pollutant diffusion path into a Bayesian network to obtain the posterior probability of the pollution source of each of the river sewage outlets specifically includes:

[0043] Calculating the prior probability of pollution sources of each of the river sewage outlets based on the historical pollution event data through the Bayesian network;

[0044] Determining a likelihood function based on a degree of matching between the predicted pollutant diffusion path and the historical pollution event data;

[0045] The posterior probability of the pollution source of each of the river sewage outlets is calculated based on the prior probability of the pollution source and the likelihood function.

[0046] Furthermore, in one embodiment of the present invention, the method for tracing the pollution source of a river sewage outlet further comprises the following steps:

[0047] Determining the current pollution source probability of each of the river sewage outlets according to the pollution source probability distribution;

[0048] Determining the pollution risk level of each of the river sewage outlets based on the current water quality parameters and the current pollution source probability of each of the river sewage outlets;

[0049] Adjust the water quality parameter sampling frequency, flow rate data sampling frequency and drone image acquisition frequency of each river sewage outlet according to the pollution risk level.

[0050] On the other hand, an embodiment of the present invention provides a pollution source tracing device for a river sewage outlet, comprising:

[0051] A data acquisition module is used to obtain current water quality parameters, current flow rate data, and drone images of multiple river sewage outlets in the target basin;

[0052] A water quality prediction module is used to identify the time series changes of the current water quality parameters through a long short-term memory network, obtain the predicted water quality parameters at a future time, and determine whether there is a pollution source in the target watershed based on the predicted water quality parameters;

[0053] A pollution source identification module is configured to, when a pollution source exists in the target watershed, input the current water quality parameters, the current flow rate data, and the drone image into a lightweight pollution source identification model deployed on an edge node to obtain a probability distribution of the pollution source in the target watershed;

[0054] A pollution source locating module, configured to determine the target river sewage outlet where the pollution source is located based on the probability distribution of the pollution source;

[0055] The lightweight pollution source identification model is obtained by jointly training and knowledge distilling the pollution diffusion spatiotemporal graph model and the Bayesian network.

[0056] On the other hand, an embodiment of the present invention provides an electronic device, which includes a memory, a processor, a program stored on the memory and runnable on the processor, and a data bus for realizing connection and communication between the processor and the memory. When the program is executed by the processor, the method for tracing the pollution source of a river sewage outlet is implemented as described above.

[0057] On the other hand, an embodiment of the present invention also provides a storage medium, which is a computer-readable storage medium used for computer-readable storage. The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the method for tracing the source of pollution from river sewage outlets as described above.

[0058] The advantages and benefits of the present invention will be described in part in the following description and will become apparent from the following description or learned through practice of the present invention:

[0059] The embodiment of the present invention obtains the current water quality parameters, current flow rate data, and drone images of multiple river sewage outlets in the target basin, identifies the time series changes of the current water quality parameters through a long short-term memory network, obtains predicted water quality parameters for the future, and determines whether there is a pollution source in the target basin based on the predicted water quality parameters. When a pollution source exists in the target basin, the current water quality parameters, current flow rate data, and drone images are input into a lightweight pollution source identification model deployed at the edge node to obtain the pollution source probability distribution of the target basin, and the target river sewage outlet where the pollution source is located is determined based on the pollution source probability distribution. The embodiment of the present invention simultaneously integrates multimodal information such as water quality parameters, flow rate data, and drone images, and dynamically predicts the pollution source probability of each river sewage outlet based on a lightweight pollution source identification model obtained by jointly training and knowledge distilling a pollution diffusion spatiotemporal graph model and a Bayesian network, thereby improving the accuracy of pollution source identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following introduction is made to the drawings required for use in the embodiments of the present invention. It should be understood that the drawings introduced below are only for the convenience of clearly describing some embodiments of the technical solutions of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative work.

[0061] Figure 1 A flowchart of the steps of a method for tracing the source of pollution at a river sewage outlet provided by an embodiment of the present invention;

[0062] Figure 2 A schematic diagram of an implementation scenario of the method for tracing the pollution source of a river sewage outlet provided by an embodiment of the present invention;

[0063] Figure 3 A flowchart of step S101 provided in an embodiment of the present invention;

[0064] Figure 4 A flowchart of step S102 provided in an embodiment of the present invention;

[0065] Figure 5 A flowchart of the steps for generating a lightweight pollution source identification model provided by an embodiment of the present invention;

[0066] Figure 6 A schematic diagram of a process for deploying a lightweight pollution source identification model provided by an embodiment of the present invention;

[0067] Figure 7 A flowchart of step S202 provided in an embodiment of the present invention;

[0068] Figure 8A schematic diagram showing the predicted pollution source probability and the predicted pollutant diffusion path output by the pollution diffusion spatiotemporal graph model provided in an embodiment of the present invention;

[0069] Figure 9 A flowchart of step S204 provided in an embodiment of the present invention;

[0070] Figure 10 Another step flow chart of the method for tracing the source of pollution at a river sewage outlet provided by an embodiment of the present invention;

[0071] Figure 11 A schematic diagram of the structure of a pollution source tracing device for a river sewage outlet provided by an embodiment of the present invention;

[0072] Figure 12 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention;

[0073] Figure 13 A schematic diagram of the structure of a storage medium provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0074] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and are not to be construed as limitations on the present application. It should be noted that, although the functional modules are divided in the system schematic and the logical order is shown in the flow chart, in some cases, the steps shown or described may be performed in a different order than the module division in the system schematic or the order in the flow chart. For the step numbers in the following embodiments, they are only provided for the convenience of explanation, and no limitation is placed on the order between the steps. The execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0075] In the description of the present invention, the meaning of "a plurality" is two or more. If there is a description of "first" or "second", it is only used to distinguish technical features and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features. In addition, unless otherwise defined, all technical and scientific terms used in this document have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used in this document are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0076] The river pollution outlet pollution source tracing method provided by the embodiments of the present application can be applied to a terminal, can be applied to a server side, and can also be software running in the terminal or the server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a set-top box, etc.; the server side can be configured as a separate physical server, can be configured as a server cluster or a distributed system formed by multiple physical servers, can also be configured as a cloud server providing basic cloud computing services such as cloud service, cloud database, cloud computing, cloud function, cloud storage, network service, cloud communication, middleware service, domain name service, security service, CDN, and big data and artificial intelligence platform; and the software can be an application that implements the river pollution outlet pollution source tracing method, but is not limited to the above forms.

[0077] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in a distributed computing environment, in which tasks are performed by remote processing devices connected by a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media, including storage devices.

[0078] It should be noted that in each specific embodiment of the present application, when relevant processing needs to be performed on data related to the identity or characteristics of the user, such as user information, user behavior data, user history data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of such data will comply with relevant laws, regulations, and standards in relevant countries and regions. In addition, when the embodiments of the present application need to obtain sensitive personal information of the user, the separate permission or separate consent of the user will be obtained through a pop-up window or a jump to a confirmation page, and after obtaining the separate permission or separate consent of the user, the necessary user-related data for enabling the embodiments of the present application to normally operate will be obtained.

[0079] As Figure 1 Figure 1 shows a step flowchart of a river pollution outlet pollution source tracing method provided by an embodiment of the present application, and Figure 1 The embodiments of the present application provide a river pollution outlet pollution source tracing method, which specifically includes the following steps:

[0080] S101, obtaining current water quality parameters, current flow rate data, and drone images of multiple river sewage outlets in the target basin;

[0081] S102. Identify the temporal changes of current water quality parameters through a long short-term memory network to obtain predicted water quality parameters at future moments, and determine whether there is a pollution source in the target watershed based on the predicted water quality parameters;

[0082] S103. When a pollution source exists in the target watershed, the current water quality parameters, current flow velocity data, and drone images are input into a lightweight pollution source identification model deployed on the edge node to obtain a probability distribution of the pollution source in the target watershed.

[0083] S104. Determine the target river discharge outlet where the pollution source is located based on the probability distribution of the pollution source;

[0084] Among them, the lightweight pollution source identification model is obtained by jointly training and knowledge distilling the pollution diffusion spatiotemporal graph model and Bayesian network.

[0085] like Figure 2 The figure shows a schematic diagram of the implementation scenario of the method for tracing the source of pollution at a river sewage outlet provided by an embodiment of the present invention. Data is obtained through the water quality sensors, flow rate sensors and drones of the perception layer and uploaded to the edge layer. Water quality prediction and pollution source identification and positioning are performed through the edge nodes of the edge layer, and the results are uploaded to the cloud layer through the transmission layer. The server cluster of the cloud layer can visualize the monitoring results of the edge nodes (curve graphs, heat maps).

[0086] The embodiments of the present invention use IoT sensing devices and edge computing nodes to achieve second-level collection and anomaly detection of pollutant concentrations, as well as real-time monitoring and dynamic response of pollution sources. At the same time, they integrate multimodal information such as water quality parameters, flow rate data, and drone images. Based on the lightweight pollution source identification model obtained through joint training and knowledge distillation of the pollution diffusion spatiotemporal graph model and the Bayesian network, the pollution source probability of each river outlet is dynamically predicted, thereby improving the accuracy of pollution source identification. The sensor acquisition frequency is adaptively adjusted according to the pollution risk level, reducing system energy consumption and data transmission pressure, and realizing dynamic resource optimization.

[0087] like Figure 3 FIG. 1 is a flowchart of step S101 provided in an embodiment of the present invention, referring to FIG. Figure 3 As an optional implementation, current water quality parameters, current flow rate data, and drone images of multiple river sewage outlets in the target basin are obtained, which specifically includes:

[0088] S1011. Obtain current water quality parameters through water quality sensors installed at sewage outlets of various rivers;

[0089] S1012, obtaining current flow velocity data through flow velocity sensors installed at sewage outlets of various rivers;

[0090] S1013. Regularly capture drone images of sewage outlets in various rivers through drone patrols;

[0091] S1014: Transmit current water quality parameters, current flow rate data, and drone images to the edge node.

[0092] Specifically, water quality sensors will be deployed at river outfalls and upstream, monitoring parameters ranging from pH (0-14), dissolved oxygen (0-20 mg / L), ammonia nitrogen (0-50 mg / L), and chemical oxygen demand (COD, 0-500 mg / L). Flow rate sensors will also be deployed. The sampling frequency of these sensors will be dynamically adjustable from 1 second (high risk) to 1 hour (low risk). The communication module will utilize LoRa / NB-IoT dual-mode, with a transmission interval of ≤10 seconds (real-time transmission in high-risk situations). Furthermore, drones equipped with hyperspectral cameras will be used for regular patrols to collect image data. The drone parameters are as follows: the hyperspectral camera has a resolution of 400-1000 nm, a spectral resolution of ≤5 nm, and a spatial resolution of 0.5 meters per pixel. The patrol frequency will be once a day in high-risk areas and once a week in low-risk areas.

[0093] After transmitting the current water quality parameters, current flow rate data, and drone images to the edge node, the edge computing node fills in missing values ​​(based on interpolation of the mean values ​​of the previous and next moments) and removes outliers in the data of the water quality sensor and flow rate sensor; then, the improved YOLO model is used to extract the morphological features of the sewage outlet in the drone image and aligns them in time and space with the water quality parameters and flow rate data (synchronize with GPS timestamps).

[0094] like Figure 4 FIG. 1 is a flowchart of step S102 provided in an embodiment of the present invention, referring to FIG. Figure 4 As an optional implementation, a long short-term memory network is used to identify the temporal changes of current water quality parameters, obtain predicted water quality parameters at future moments, and determine whether there is a pollution source in the target watershed based on the predicted water quality parameters. Specifically, the method includes:

[0095] S1021. Generate a time series of water quality parameters for each river sewage outlet based on the current water quality parameters;

[0096] S1022. Input the time series water quality parameter sequence into a pre-trained long short-term memory network to obtain the predicted water quality parameters of each river sewage outlet at a future time;

[0097] S1023. Calculate water quality residuals based on the predicted water quality parameters and the current water quality parameters, and calculate the mean and standard deviation of the water quality residuals in a preset residual sliding window;

[0098] S1024. Determine a dynamic threshold based on the mean and standard deviation;

[0099] S1025. When the duration during which the water quality residual is greater than the dynamic threshold exceeds a preset duration threshold, it is determined that there is a pollution source in the target watershed.

[0100] Specifically, the time series changes of water quality parameters are analyzed in real time based on the pre-trained LSTM model, and abnormal alarms are triggered in combination with dynamic thresholds (comparison of pollutant concentration with historical baselines).

[0101] The input of the LSTM model is a time series water quality parameter sequence , the LSTM model outputs the predicted value , calculate the residual . Set dynamic threshold ,in and is the mean and standard deviation of the residual sliding window (window size = 30 minutes). If it lasts for more than 5 minutes, it is determined that there is a pollution source in the target watershed and an abnormal alarm is triggered.

[0102] like Figure 5 The figure shows a flow chart of steps for generating a lightweight pollution source identification model provided by an embodiment of the present invention, referring to Figure 5 As an optional implementation, the lightweight pollution source identification model is obtained by the following steps:

[0103] S201. Obtain historical pollution event data for the target river basin, including historical water quality parameters, historical flow rate data, historical drone images, and historical pollution source probabilities for each river outlet;

[0104] S202: Construct a pollution diffusion spatiotemporal graph model using river sewage outlets as nodes and water flow directions as edges, input historical pollution event data into the pollution diffusion spatiotemporal graph model, and obtain predicted pollution source probabilities and predicted pollutant diffusion paths for each river sewage outlet;

[0105] S203, determining the graph reconstruction loss of the pollution diffusion spatiotemporal graph model based on the predicted pollution source probability and the historical pollution source probability;

[0106] S204: Input historical pollution event data and predicted pollutant diffusion paths into the Bayesian network to obtain the posterior probability of pollution sources at each river discharge outlet;

[0107] S205. Determine the posterior probability matching loss of the Bayesian network based on the posterior probability of the pollution source and the historical pollution source probability;

[0108] S206. Determine a joint loss function based on the graph reconstruction loss and the posterior probability matching loss, and update the parameters of the pollution diffusion spatiotemporal graph model and the Bayesian network through a backpropagation algorithm based on the joint loss function to obtain a trained pollution source identification model;

[0109] S207. Perform knowledge distillation on the pollution source identification model to obtain a lightweight pollution source identification model.

[0110] like Figure 6 FIG. 1 is a flow chart of deploying a lightweight pollution source identification model according to an embodiment of the present invention. The specific process is as follows:

[0111] 1) Construct a spatiotemporal graph model of pollution diffusion (ST-GCN), using sewage outlets as nodes and water flow directions as edges, and combine hydrological data (flow velocity, flow) to simulate the migration path of pollutants.

[0112] 2) Based on the Bayesian network (BN) and historical pollution event data, the probability weight of each potential pollution source is calculated to generate a priority screening sequence.

[0113] 3) Spatiotemporal Graph Convolutional Network (ST-GCN) and Bayesian Network (BN) are jointly trained, and the loss function is:

[0114]

[0115] in, is the image reconstruction loss, is the posterior probability matching loss, and is the preset weight parameter.

[0116] 4) Decision fusion: Generate the probability distribution of pollution sources through weighted voting (the confidence level of the spatiotemporal graph convolutional network (ST-GCN) is 60% and the confidence level of the Bayesian network (BN) is 40%).

[0117] 5) Through cloud-edge collaborative architecture and knowledge distillation, a lightweight TinyML model is deployed on edge nodes. The cloud fine-tunes the model on the edge based on transfer learning to achieve dynamic model updates.

[0118] like Figure 7 FIG2 is a flowchart of step S202 provided in an embodiment of the present invention, referring to FIG2 Figure 7 As an optional implementation, historical pollution event data is input into the pollution diffusion spatiotemporal graph model to obtain the predicted pollution source probability and pollutant diffusion path of each river sewage outlet, which specifically includes:

[0119] S2021. Determine a multimodal embedding vector for each node based on historical water quality parameters, historical flow velocity data, historical drone images, and historical pollution source probabilities.

[0120] S2022. Determine the edge weight of each edge based on the historical flow rate data of adjacent nodes, historical unmanned images, and node distances;

[0121] S2023. Perform information transmission on each node a preset number of times to obtain an updated multimodal embedding vector for each node;

[0122] S2024. Determine the predicted pollution source probability and the predicted pollutant diffusion path corresponding to each node based on the updated multimodal embedding vector.

[0123] Specifically, each sewage outlet node is determined based on historical water quality parameters, historical flow rate data, historical drone images, and historical pollution source probabilities. The multimodal embedding vector of is as follows:

[0124]

[0125] Sewage outlet node arrive The edge weights of fused physical diffusion with visual relevance:

[0126]

[0127] in is the structural similarity index, is the image similarity factor of the edge weight in the spatiotemporal graph model;

[0128] The information transfer formula is as follows:

[0129]

[0130] in is the neighbor node, 、 is the node degree, is the trainable weight matrix.

[0131] According to the information transfer formula, information is transferred to each node for a preset number of times to obtain the updated multimodal embedding vector of each node. According to the updated multimodal embedding vector, the predicted pollution source probability and the predicted pollutant diffusion path corresponding to each node can be determined, such as Figure 8 Shown is a schematic diagram of the predicted pollution source probability and the predicted pollutant diffusion path output by the pollution diffusion spatiotemporal graph model provided by an embodiment of the present invention.

[0132] like Figure 9FIG2 is a flowchart of step S204 provided in an embodiment of the present invention, referring to FIG2 Figure 9 As an optional implementation, historical pollution event data and predicted pollutant diffusion paths are input into the Bayesian network to obtain the posterior probability of pollution sources at each river discharge outlet, which specifically includes:

[0133] S2041. Calculate the prior probability of pollution sources at each river discharge outlet based on historical pollution event data using a Bayesian network;

[0134] S2042. Determine a likelihood function based on the matching degree between the predicted pollutant diffusion path and the historical pollution event data;

[0135] S2043. Calculate the posterior probability of the pollution source of each river sewage outlet based on the prior probability of the pollution source and the likelihood function.

[0136] Specifically, define the set of potential pollution sources , observational data , then the posterior probability formula is as follows:

[0137]

[0138] Among them, the prior probability Calculation based on the frequency of historical pollution events; likelihood function The matching degree of the pollutant diffusion path is determined by simulating the space-time graph model.

[0139] like Figure 10 FIG2 is another flow chart of the method for tracing the source of pollution at a river sewage outlet provided by an embodiment of the present invention, referring to FIG2. Figure 10 As an optional implementation method, the method for tracing the pollution source of a river sewage outlet further includes the following steps:

[0140] S105. Determine the current pollution source probability of each river sewage outlet based on the pollution source probability distribution;

[0141] S106. Determine the pollution risk level of each river sewage outlet based on the current water quality parameters and current pollution source probability of each river sewage outlet;

[0142] S107. Adjust the water quality parameter sampling frequency, flow velocity data sampling frequency, and drone image acquisition frequency of each river sewage outlet according to the pollution risk level.

[0143] Specifically, the current pollution source probability of each river sewage outlet is determined based on the probability distribution of pollution sources. The river sewage outlets with the highest current pollution source probability are the target river sewage outlets. Subsequently, drone images can be used to identify the characteristics of industrial facilities around the sewage outlets (such as the shape of drainage pipes and enterprise types) to assist manual verification.

[0144] In addition, the embodiment of the present application also dynamically adjusts the water quality parameter sampling frequency, the flow rate data sampling frequency and the unmanned aerial vehicle image acquisition frequency according to the pollution risk level (real-time monitoring in high-risk areas, and reduced frequency sampling in low-risk areas).

[0145] The calculation formula of the pollution risk level is as follows:

[0146]

[0147] The normalized pollutant concentration and the concentration change rate can be calculated according to the current water quality parameters, 、 and The values of 0.5, 0.3 and 0.2 respectively.

[0148] Taking the water quality sensor as an example, the sampling frequency is adjusted as follows:

[0149]

[0150] Taking a tributary of the Yangtze River as an example, 200 sensor terminals are arranged along the river, and an edge computing node is set every 5 kilometers, and an unmanned aerial vehicle patrols once a week; on a certain day, the ammonia nitrogen concentration suddenly increases, the edge node triggers an alarm and uploads data to the cloud, the AI model combines hydrological data to trace the pollution path, and locks a chemical plant 3 kilometers upstream as a high-probability pollution source (probability 82%), and switches the sensors around the chemical plant to real-time monitoring mode, and reduces the frequency to once an hour in other areas.

[0151] The method flow of the embodiment of the present application is described above. It can be understood that the embodiment of the present application realizes second-level collection and abnormal detection of pollutant concentration, and real-time monitoring and dynamic response of the pollution source through the Internet of Things sensing device and the edge computing node; at the same time, multi-modal information such as water quality parameters, flow rate data and unmanned aerial vehicle images are integrated, a light pollution source identification model is obtained based on joint training and knowledge distillation of the pollution diffusion space-time graph model and the Bayesian network, and the pollution source probability of each river sewage outlet is dynamically predicted, which improves the accuracy of pollution source identification; the sensor acquisition frequency is adaptively adjusted according to the pollution risk level, the system energy consumption and data transmission pressure are reduced, and dynamic resource optimization is realized.

[0152] As Figure 11 shown is a structure schematic diagram of a river sewage outlet pollution source tracing device provided by the embodiment of the present application, with reference to Figure 11 , the embodiment of the present application provides a river sewage outlet pollution source tracing device, which comprises:

[0153] A data acquisition module is configured to acquire current water quality parameters, current flow rate data and unmanned aerial vehicle images of a plurality of river sewage outlets in a target river basin.

[0154] The water quality prediction module is used to identify the temporal changes of current water quality parameters through the long short-term memory network, obtain the predicted water quality parameters at future moments, and determine whether there are pollution sources in the target watershed based on the predicted water quality parameters;

[0155] The pollution source identification module is used to input the current water quality parameters, current flow rate data, and drone images into the lightweight pollution source identification model deployed on the edge node when there is a pollution source in the target watershed, and obtain the probability distribution of the pollution source in the target watershed;

[0156] The pollution source location module is used to determine the target river sewage outlet where the pollution source is located based on the probability distribution of the pollution source;

[0157] Among them, the lightweight pollution source identification model is obtained by jointly training and knowledge distilling the pollution diffusion spatiotemporal graph model and Bayesian network.

[0158] The contents of the above method embodiments are all applicable to the present device embodiments. The functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0159] An embodiment of the present invention further provides an electronic device comprising: a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for enabling communication between the processor and the memory. When the program is executed by the processor, the aforementioned method for tracing the source of pollution from a river sewage outlet is implemented. The electronic device can be any intelligent terminal, including a tablet computer and an in-vehicle computer.

[0160] like Figure 12 FIG2 is a schematic diagram of the hardware structure of the electronic device provided by the embodiment of the present invention, referring to FIG2 Figure 12 , an embodiment of the present invention provides an electronic device, including:

[0161] The processor 1201 may be implemented as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is configured to execute relevant programs to implement the technical solutions provided by the embodiments of the present invention.

[0162] The memory 1202 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 1202 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1202 and is called by the processor 1201 to execute the pollution source tracing method for river sewage outlets in the embodiments of the present invention.

[0163] Input / output interface 1203, used to implement information input and output;

[0164] Communication interface 1204, used to implement communication interaction between this device and other devices, which can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WiFi, Bluetooth, etc.);

[0165] Bus 1205 , which transmits information between various components of the device (e.g., processor 1201 , memory 1202 , input / output interface 1203 , and communication interface 1204 );

[0166] The processor 1201 , the memory 1202 , the input / output interface 1203 and the communication interface 1204 are connected to each other in communication within the device via the bus 1205 .

[0167] like Figure 13 FIG2 is a schematic diagram of the structure of the storage medium provided by the embodiment of the present invention, referring to FIG2 Figure 13 An embodiment of the present invention further provides a storage medium, which is a computer-readable storage medium used for computer-readable storage. The storage medium stores one or more programs 1301, and the one or more programs 1301 can be executed by one or more processors to implement the above-mentioned method for tracing the source of pollution at river sewage outlets.

[0168] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0169] The embodiment of the present invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs Figure 1 The method shown.

[0170] In some optional embodiments, the function / operation mentioned in the block diagram may not occur in the order mentioned in the operation diagram. For example, depending on the function / operation involved, the two boxes shown in succession can actually be executed substantially simultaneously or the above-mentioned boxes can sometimes be executed in reverse order. In addition, the embodiment presented and described in the flow chart of the present invention is provided in an exemplary manner for the purpose of providing a more comprehensive understanding of the technology. The disclosed method is not limited to the operation and logic flow presented herein. Optional embodiments are contemplated in which the order of the various operations is changed and the sub-operations described as a part of a larger operation are performed independently.

[0171] In addition, although the present invention is described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the above-mentioned functions and / or features can be integrated into a single physical device and / or software module, or one or more functions and / or features can be implemented in separate physical devices or software modules. It is also understood that a detailed discussion of the actual implementation of each module is not necessary for understanding the present invention. More specifically, given the properties, functions, and internal relationships of the various functional modules in the devices disclosed herein, the actual implementation of the module will be understood within the routine skills of an engineer. Therefore, a person skilled in the art can implement the present invention set forth in the claims using ordinary skills without undue experimentation. It is also understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0172] If the above functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the above methods in various embodiments of the present invention. The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0173] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0174] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable media on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other media, and then editing, interpreting, or processing it in another suitable manner as necessary, and then storing it in a computer memory.

[0175] It should be understood that various components of the present invention may be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods may be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof may be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field-programmable gate array (FPGA), etc.

[0176] In the above description of this specification, reference to the terms "one embodiment / example," "another embodiment / example," or "certain embodiments / examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0177] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

[0178] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A method for tracing the pollution source of a river sewage outlet, characterized in that: The following steps are involved: Obtain current water quality parameters, current flow rate data, and drone imagery for multiple river outfalls in the target basin; Identifying the temporal changes of the current water quality parameters through a long short-term memory network to obtain predicted water quality parameters at future moments, and determining whether there is a pollution source in the target watershed based on the predicted water quality parameters; When there is a pollution source in the target watershed, the current water quality parameters, the current flow rate data, and the drone image are input into a lightweight pollution source identification model deployed on the edge node to obtain a probability distribution of the pollution source in the target watershed; Determine the target river sewage outlet where the pollution source is located according to the probability distribution of the pollution source; The lightweight pollution source identification model is obtained by the following steps: Obtaining historical pollution event data for the target river basin, the historical pollution event data including historical water quality parameters, historical flow rate data, historical drone images, and historical pollution source probabilities for each of the river sewage outlets; Constructing a pollution diffusion spatiotemporal graph model with the river sewage outlets as nodes and the water flow directions as edges, inputting the historical pollution event data into the pollution diffusion spatiotemporal graph model to obtain the predicted pollution source probability and the predicted pollutant diffusion path of each river sewage outlet; Determining a graph reconstruction loss of the pollution diffusion spatiotemporal graph model according to the predicted pollution source probability and the historical pollution source probability; Inputting the historical pollution event data and the predicted pollutant diffusion path into a Bayesian network to obtain the posterior probability of the pollution source of each river sewage outlet; Determining the posterior probability matching loss of the Bayesian network according to the posterior probability of the pollution source and the historical pollution source probability; Determining a joint loss function based on the graph reconstruction loss and the posterior probability matching loss, and updating the parameters of the pollution diffusion spatiotemporal graph model and the Bayesian network through a backpropagation algorithm based on the joint loss function to obtain a trained pollution source identification model; The lightweight pollution source identification model is obtained by performing knowledge distillation on the pollution source identification model.

2. A method for tracing the pollution source of a river sewage outlet according to claim 1, characterized in that: The method of obtaining current water quality parameters, current flow rate data, and drone images of multiple river sewage outlets in the target basin specifically includes: Acquiring the current water quality parameters through water quality sensors installed at the sewage outlets of each river; Acquiring the current flow rate data by using a flow rate sensor provided at each of the river sewage outlets; Regularly taking drone images of the sewage outlets of each river by drone patrol; The current water quality parameter, the current flow rate data, and the drone image are transmitted to the edge node.

3. The method for tracing the pollution source of a river sewage outlet according to claim 1, characterized in that: The method of identifying the temporal changes of the current water quality parameters through the long short-term memory network to obtain the predicted water quality parameters at a future time, and determining whether there is a pollution source in the target watershed based on the predicted water quality parameters, specifically includes: Generating a time-series water quality parameter sequence of each of the river sewage outlets according to the current water quality parameters; Inputting the time series water quality parameter sequence into the pre-trained long short-term memory network to obtain the predicted water quality parameters of each river sewage outlet at a future time; Calculating a water quality residual based on the predicted water quality parameter and the current water quality parameter, and calculating a mean and a standard deviation of the water quality residual in a preset residual sliding window; determining a dynamic threshold value according to the mean and the standard deviation; When the duration during which the water quality residual is greater than the dynamic threshold exceeds a preset duration threshold, it is determined that a pollution source exists in the target watershed.

4. The method for tracing the pollution source of a river sewage outlet according to claim 1, characterized in that: Inputting the historical pollution event data into the pollution diffusion spatiotemporal graph model to obtain the predicted pollution source probability and the predicted pollutant diffusion path of each river sewage outlet specifically includes: Determining a multimodal embedding vector for each node based on the historical water quality parameters, the historical flow velocity data, the historical drone images, and the historical pollution source probabilities; Determining an edge weight for each edge based on the historical flow rate data of adjacent nodes, the historical drone image, and the node distance; Performing a preset number of information transfers on each node to obtain the updated multimodal embedding vector of each node; The predicted pollution source probability and the predicted pollutant diffusion path corresponding to each node are determined according to the updated multimodal embedding vector.

5. The method for tracing the pollution source of a river sewage outlet according to claim 1, characterized in that: Inputting the historical pollution event data and the predicted pollutant diffusion path into the Bayesian network to obtain the posterior probability of the pollution source of each river sewage outlet specifically includes: Calculating the prior probability of pollution sources of each of the river sewage outlets based on the historical pollution event data through the Bayesian network; Determining a likelihood function based on a degree of matching between the predicted pollutant diffusion path and the historical pollution event data; The posterior probability of the pollution source of each of the river sewage outlets is calculated based on the prior probability of the pollution source and the likelihood function.

6. A method for tracing pollution sources at river sewage outlets according to any one of claims 1 to 5, characterized in that: The method for tracing the pollution source of a river sewage outlet further comprises the following steps: Determining the current pollution source probability of each of the river sewage outlets according to the pollution source probability distribution; Determining the pollution risk level of each of the river sewage outlets based on the current water quality parameters and the current pollution source probability of each of the river sewage outlets; Adjust the water quality parameter sampling frequency, flow rate data sampling frequency and drone image acquisition frequency of each river sewage outlet according to the pollution risk level.

7. A pollution source tracing device for a river sewage outlet, characterized in that: include: A data acquisition module is used to obtain current water quality parameters, current flow rate data, and drone images of multiple river sewage outlets in the target basin; A water quality prediction module is used to identify the time series changes of the current water quality parameters through a long short-term memory network, obtain the predicted water quality parameters at a future time, and determine whether there is a pollution source in the target watershed based on the predicted water quality parameters; A pollution source identification module is configured to, when a pollution source exists in the target watershed, input the current water quality parameters, the current flow rate data, and the drone image into a lightweight pollution source identification model deployed on an edge node to obtain a probability distribution of the pollution source in the target watershed; A pollution source locating module, configured to determine the target river sewage outlet where the pollution source is located based on the probability distribution of the pollution source; The lightweight pollution source identification model is obtained by the following steps: Obtaining historical pollution event data for the target river basin, the historical pollution event data including historical water quality parameters, historical flow rate data, historical drone images, and historical pollution source probabilities for each of the river sewage outlets; Constructing a pollution diffusion spatiotemporal graph model with the river sewage outlets as nodes and the water flow directions as edges, inputting the historical pollution event data into the pollution diffusion spatiotemporal graph model to obtain the predicted pollution source probability and the predicted pollutant diffusion path of each river sewage outlet; Determining a graph reconstruction loss of the pollution diffusion spatiotemporal graph model according to the predicted pollution source probability and the historical pollution source probability; Inputting the historical pollution event data and the predicted pollutant diffusion path into a Bayesian network to obtain the posterior probability of the pollution source of each river sewage outlet; Determining the posterior probability matching loss of the Bayesian network according to the posterior probability of the pollution source and the historical pollution source probability; Determining a joint loss function based on the graph reconstruction loss and the posterior probability matching loss, and updating the parameters of the pollution diffusion spatiotemporal graph model and the Bayesian network through a backpropagation algorithm based on the joint loss function to obtain a trained pollution source identification model; The lightweight pollution source identification model is obtained by performing knowledge distillation on the pollution source identification model.

8. An electronic device, characterized in that: The electronic device includes a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory. When the program is executed by the processor, the steps of the method for tracing the pollution source of a river sewage outlet as described in any one of claims 1 to 6 are realized.

9. A storage medium, which is a computer-readable storage medium and is used for computer-readable storage, characterized in that: The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the method for tracing the pollution source of a river sewage outlet as described in any one of claims 1 to 6.

Images