Monitoring and early warning method and device for water environment pollution and storage medium

Abstract

The invention discloses a monitoring and early warning method and device for water environment pollution and a storage medium, and the method comprises the steps: determining a plurality of target monitoring points preset in a target region, and generating a first graph structure used for representing the migration motion of pollutants and a second graph structure used for representing the diffusion and degradation of the pollutants based on the plurality of target monitoring points; inputting the first graph structure into a preset first graph neural network, and outputting a third graph structure; inputting the second graph structure into a preset second graph neural network, and outputting a fourth graph structure; performing feature extraction and fusion on the third graph structure and the fourth graph structure, and generating a first pollutant concentration feature vector corresponding to each target monitoring point; and inputting the pollutant concentration feature vector into a risk prediction model, determining a pollution degree corresponding to each target monitoring point, and determining whether to carry out early warning based on the pollution degree. The technical effect of monitoring and early warning water environment pollution is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of water environment monitoring, and in particular to a monitoring and early warning method, device and storage medium for water environment pollution. Background Art

[0002] Water is a precious resource that humans rely on for survival. Its ecological environment and pollution level are of vital importance to humans. However, the current water pollution monitoring and early warning are still in a relatively primitive state, with a low level of intelligence. For example, by investigating and evaluating the pollution sources, the main pollutants in the water environment of the target area can be determined. A third party agency can be commissioned to conduct sampling and analysis to determine the pollution level in the water environment of the target area. The transparency of the water environment in the target area can be observed with the naked eye to determine the pollution level.

[0003] However, the above-mentioned method of monitoring the pollution level of the water environment in the target area has some defects. For example, although the main pollutants in the water environment of the target area can be determined by investigating and evaluating the pollution sources, the pollution sources cannot be located, and the water quality of the water environment in the target area cannot be accurately evaluated and predicted. For another example, although the pollution level in the water environment of the target area can be determined by entrusting a third-party agency to conduct sampling and analysis, the water quality monitoring equipment can only collect water quality data from discrete sites and cannot obtain more comprehensive water quality data. For another example, since the water environment is open and complex, pollutants will diffuse and migrate, and the interactions between the various pollutants are complex and there are many coupling factors. Therefore, how to accurately predict the changes in the pollution level of the water environment and monitor and warn based on the prediction results is an urgent problem to be solved.

[0004] The publication number is CN119005539A, and the name is a water quality monitoring and early warning method and system. It includes the following steps: collecting multidimensional water quality parameter data through multiple sets of sensors to obtain time series data, preprocessing, and obtaining preprocessed time series data; constructing a multi-factor water quality prediction model, which includes a single factor prediction module, a feature fusion module, a Transformer encoder module, and a prediction output module. The single factor prediction module is a single factor time series monitoring model, and the preprocessed time series data multi-factor water quality prediction model obtains the multidimensional water quality parameter prediction value of the future time step; constructing an early warning model based on the multi-factor time series prediction results, and the multi-dimensional water quality parameter prediction value is processed by the early warning model based on the multi-factor time series prediction results to generate an early warning signal.

[0005] The publication number is CN117491585A, and the name is a water ecological pollution monitoring method, device and system based on a time series network. It includes: obtaining topographic survey data of the target river section, building a two-dimensional river section model; determining the calculation area of ​​the target river section; gridding the calculation area of ​​the target river section; setting relevant parameters for the two-dimensional river section model, conducting hydrodynamic analysis of the target river section, and outputting the hydrodynamic analysis result data; constructing a water quality prediction model based on a time series network; obtaining a time series data set of historical water quality indicators of each outlet of the target river section based on the hydrodynamic analysis result data and the pollution source data generated by the historical monitoring data, and inputting it into the water quality prediction model based on the time series network for training and testing, and then predicting the water quality monitoring data of the target river section; generating monitoring report data based on the predicted water quality monitoring data.

[0006] With respect to the technical problem existing in the above-mentioned prior art that it is currently impossible to accurately predict the changes in the degree of water pollution and to carry out monitoring and early warning based on the prediction results, no effective solution has been proposed yet. Summary of the invention

[0007] The embodiments of the present disclosure provide a monitoring and early warning method, device and storage medium for water environment pollution, so as to at least solve the technical problem existing in the prior art that it is currently impossible to accurately predict the changes in the degree of water environment pollution and to perform monitoring and early warning based on the prediction results.

[0008] According to one aspect of an embodiment of the present disclosure, a monitoring and early warning method for water environment pollution is provided, including: determining a plurality of target monitoring points pre-set in a target area, and based on the plurality of target monitoring points, respectively generating a first graph structure for representing the migration and movement of pollutants and a second graph structure for representing the diffusion and degradation of pollutants; constructing a first graph neural network corresponding to the first graph structure, and inputting the first graph structure into the first graph neural network, thereby outputting a third graph structure; constructing a second graph neural network corresponding to the second graph structure, and inputting the second graph structure into the second graph neural network, thereby outputting a fourth graph structure; performing feature extraction and fusion on the third graph structure and the fourth graph structure, and generating a first pollutant concentration feature vector corresponding to each target monitoring point; and inputting the pollutant concentration feature vector into a pre-trained risk prediction model to determine the degree of pollution corresponding to each target monitoring point, and based on the degree of pollution corresponding to each target monitoring point, determining whether to issue an early warning.

[0009] According to another aspect of an embodiment of the present disclosure, a storage medium is further provided, the storage medium including a stored program, wherein when the program is running, a processor executes any one of the methods described above.

[0010] According to another aspect of an embodiment of the present disclosure, a monitoring and early warning device for water environment pollution is also provided, including: a graph structure generation module, used to determine a plurality of target monitoring points pre-set in a target area, and based on the plurality of target monitoring points, respectively generate a first graph structure for representing the migration movement of pollutants and a second graph structure for representing the diffusion and degradation of pollutants; a first graph structure output module, used to construct a first graph neural network corresponding to the first graph structure, and input the first graph structure into the first graph neural network, thereby outputting a third graph structure; a second graph structure output module, used to construct a second graph neural network corresponding to the second graph structure, and input the second graph structure into the second graph neural network, thereby outputting a fourth graph structure; a fusion module, used to extract and fuse features of the third graph structure and the fourth graph structure, and generate a first pollutant concentration feature vector corresponding to each target monitoring point; and an early warning module, used to input the first pollutant concentration feature vector into a pre-trained risk prediction model, determine the degree of pollution corresponding to each target monitoring point, and determine whether to issue an early warning based on the degree of pollution corresponding to each target monitoring point.

[0011] According to another aspect of an embodiment of the present disclosure, a monitoring and early warning device for water environment pollution is also provided, including: a processor; and a memory, connected to the processor, for providing the processor with instructions for processing the following processing steps: determining a plurality of target monitoring points pre-set in a target area, and based on the plurality of target monitoring points, respectively generating a first graph structure for representing the migration movement of pollutants and a second graph structure for representing the diffusion and degradation of pollutants; constructing a first graph neural network corresponding to the first graph structure, and inputting the first graph structure into the first graph neural network, thereby outputting a third graph structure; constructing a second graph neural network corresponding to the second graph structure, and inputting the second graph structure into the second graph neural network, thereby outputting a fourth graph structure; performing feature extraction and fusion on the third graph structure and the fourth graph structure, and generating a first pollutant concentration feature vector corresponding to each target monitoring point; and inputting the first pollutant concentration feature vector into a pre-trained risk prediction model to determine the degree of pollution corresponding to each target monitoring point, and based on the degree of pollution corresponding to each target monitoring point, determining whether to issue an early warning.

[0012] The present application provides a monitoring and early warning method for water environment pollution. First, the processor determines a plurality of target monitoring points preset in the target area, and based on the plurality of target monitoring points, generates a first graph structure for representing the migration movement of pollutants and a second graph structure for representing the diffusion and degradation of pollutants. Then, the processor constructs a first graph neural network corresponding to the first graph structure, and inputs the first graph structure into the first graph neural network, thereby outputting a third graph structure. At the same time, the processor constructs a second graph neural network corresponding to the second graph structure, and inputs the second graph structure into the second graph neural network, thereby outputting a fourth graph structure. After that, the third graph structure and the fourth graph structure are feature extracted and fused, and a first pollutant concentration feature vector corresponding to each target monitoring point is generated. Finally, the processor inputs the first pollutant concentration feature vector into a pre-trained risk prediction model, determines the degree of pollution corresponding to each target monitoring point, and determines whether to issue an early warning based on the degree of pollution corresponding to each target monitoring point.

[0013] Referring to the above content, it can be known that the present application generates a first graph structure for representing the migration of pollutants and a second graph structure for representing the diffusion and degradation of pollutants. And because the migration of pollutants between two adjacent target monitoring points will affect each other, when the first graph neural network is used to process the data of the first graph structure, the first association relationship between the two adjacent nodes (corresponding to the target monitoring points) (i.e., the edge between the two adjacent nodes) can be retained, so that the third graph structure finally outputted not only retains the data information of each node, but also retains the structural information between the two adjacent nodes.

[0014] Similarly, since the diffusion of pollutants between two adjacent target monitoring points will affect each other, when the second graph neural network is used to process the data of the second graph structure, the second association relationship between the two adjacent nodes (corresponding to the target monitoring points) (i.e., the edge between the two adjacent nodes) can be retained, so that the fourth graph structure finally output not only retains the data information of each node, but also retains the structural information between the two adjacent nodes.

[0015] Therefore, when the third graph structure and the fourth graph structure are subjected to feature extraction and fusion, the first pollutant concentration feature vector corresponding to each target monitoring point generated in the end contains relatively comprehensive feature information. And when the first pollutant concentration feature vector is input into the pre-trained risk prediction model, the risk prediction model can accurately predict the pollution degree corresponding to each target monitoring point. Finally, based on the pollution degree corresponding to each target monitoring point, it is selected whether to issue an early warning.

[0016] This solves the technical problem in the prior art that it is currently impossible to accurately predict changes in the degree of water pollution and to perform monitoring and early warning based on the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present disclosure and constitute a part of the present application. The illustrative embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation on the present disclosure. In the drawings: Figure 1 is a hardware structure block diagram of a computing device for implementing the method according to Embodiment 1 of the present application; Figure 2 is a schematic diagram of a monitoring and early warning system for water environment pollution according to Example 1 of the present application; Figure 3 It is a modular schematic diagram of the monitoring and early warning platform according to Example 1 of the present application; Figure 4 It is a flow chart of the monitoring and early warning method for water environment pollution according to Example 1 of the present application; Figure 5 is a schematic diagram of the first graph structure according to Embodiment 1 of the present application; Figure 6 is a schematic diagram of the second graph structure according to Embodiment 1 of the present application; Figure 7 is a schematic diagram of the risk prediction model described in Example 1 of the present application; Figure 8 is a schematic diagram of a monitoring and early warning device for water environment pollution according to Example 2 of the present application; and Fig. 9 It is a schematic diagram of the monitoring and early warning device for water environment pollution described in Example 3 of the present application. DETAILED DESCRIPTION

[0018] In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only embodiments of a part of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present disclosure.

[0019] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or devices. Example 1

[0020] According to this embodiment, a method embodiment for monitoring and early warning of water environment pollution is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0021] The method embodiment provided in this embodiment can be executed in a mobile terminal, a computer terminal, a server or a similar computing device. Figure 1 The hardware structure block diagram of a computing device for implementing a monitoring and early warning method for water environment pollution is shown. Figure 1 As shown, the computing device may include one or more processors (the processor may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory for storing data, a transmission device for communication functions, and an input / output interface. The memory, the transmission device, and the input / output interface are connected to the processor via a bus. In addition, it may also include: a display, a keyboard, and a cursor control device connected to the input / output interface. A person skilled in the art can understand that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0022] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computing device. As involved in the embodiments of the present disclosure, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0023] The memory can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the monitoring and early warning method for water environment pollution in the embodiment of the present disclosure. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, the monitoring and early warning method for water environment pollution of the above-mentioned application program is realized. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the computing device 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.

[0024] The transmission device is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computing device. In one example, the transmission device includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0025] The display may be, for example, a touch screen liquid crystal display (LCD) that may enable a user to interact with a user interface of the computing device.

[0026] It should be noted that, in some optional embodiments, the above Figure 1 The computing device shown may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of both hardware elements and software elements. Figure 1 This is merely one example of a particular embodiment and is intended to illustrate the types of components that may be present in the computing devices described above.

[0027] Figure 2Schematic diagram of a monitoring and early warning system for water environment pollution according to this embodiment. Figure 2 As shown, the system includes: a plurality of water quality detectors 10 , a monitoring and early warning platform 20 communicatively connected to the plurality of water quality detectors 10 , and a GIS platform 30 .

[0028] The GIS platform 30 is connected to the monitoring and early warning platform 20 for communication, and the GIS platform 30 is used to extract pollution information corresponding to the target area in combination with the remote sensing image of the target area. The pollution information corresponding to the target area includes, for example, river water resource information of the target area, historical pollution source information of the target area, and pollutant development of the target area. It should be noted by those skilled in the art that the above is only an example of pollution information corresponding to the target area, and the actual situation is not limited thereto.

[0029] In addition, the GIS platform 30 is also used to select monitoring points where pollutants may appear based on the extracted pollution information corresponding to the target area, and use the selected monitoring points as target monitoring points.

[0030] When the monitoring and early warning platform 20 receives information related to the target monitoring point sent by the GIS platform 30, it sends the relevant information to the terminal device 40 corresponding to the staff, so that the staff can set the water quality detector 10 at the target monitoring point in the target area based on the information displayed by the terminal device 40.

[0031] Among them, multiple water quality detectors 10 are used to detect the components of various pollutants in the water environment corresponding to the target monitoring point.

[0032] When each water quality detector 10 detects the components of various pollutants in the water environment of the corresponding target monitoring point, the monitoring and early warning platform 20 obtains corresponding data information from each water quality detector 10, and determines the pollution degree corresponding to each target monitoring point based on the collected data information. And when the pollution degree is greater than the preset threshold, the monitoring and early warning platform 20 issues an early warning to the staff through the terminal device 40.

[0033] Figure 3 2 is a modular schematic diagram of the monitoring and early warning platform 20 according to an embodiment of the present application. Figure 3 As shown, the monitoring and early warning platform 20 includes: a data acquisition module, a first graph neural network module, a second graph neural network module, a fusion module, a risk prediction module and an alarm module. Among them, the data acquisition module is used to collect the composition and concentration of various pollutants in the water environment of the target monitoring point from multiple water quality detectors 10, and generate a first graph structure for representing the migration and movement of pollutants and a second graph structure for representing the diffusion and degradation of pollutants.

[0034] The data acquisition module is respectively connected to the first graph neural network module and the second graph neural network for sending the first graph structure to the first graph neural network module and sending the second graph structure to the second graph neural network module.

[0035] The first graph neural network module and the second graph neural network module are respectively connected to the fusion module. The first graph neural network module is used to generate a third graph structure based on the first graph structure, and send the third graph structure to the fusion module. The second graph neural network module is used to generate a fourth graph structure based on the second graph structure, and send the fourth graph structure to the fusion module.

[0036] The fusion module is in communication connection with the risk prediction module. The fusion module is used to extract the second pollutant concentration feature vector from the third graph structure, extract the third pollutant concentration feature vector from the fourth graph structure, and further fuse the second pollutant concentration feature vector and the third pollutant concentration feature vector to generate the first pollutant concentration feature vector.

[0037] The risk prediction module is in communication connection with the alarm module, and is used to predict the pollution degree based on the first pollutant concentration feature vector.

[0038] The alarm module determines whether to issue an alarm to the staff through the terminal device 40 based on the received pollution level prediction result and the preset threshold.

[0039] Under the above operating environment, according to the first aspect of this embodiment, a monitoring and early warning method for water environment pollution is provided. The method comprises: Figure 2 The system implementation shown in . Figure 4 A schematic diagram showing the process of the method is shown in FIG. Figure 4 As shown, the method includes: S402: determining a plurality of target monitoring points pre-set in the target area, and generating a first graph structure for representing the migration and movement of pollutants and a second graph structure for representing the diffusion and degradation of pollutants based on the plurality of target monitoring points; S404: constructing a first graph neural network corresponding to the first graph structure, and inputting the first graph structure into the first graph neural network, thereby outputting a third graph structure; S406: constructing a second graph neural network corresponding to the second graph structure, and inputting the second graph structure into the second graph neural network, thereby outputting a fourth graph structure; S408: extracting and fusing features of the third graph structure and the fourth graph structure, and generating a first pollutant concentration feature vector corresponding to each target monitoring point; and S410: Input the first pollutant concentration feature vector into a pre-trained risk prediction model to determine the pollution degree corresponding to each target monitoring point, and determine whether to issue an early warning based on the pollution degree corresponding to each target monitoring point.

[0040] Specifically, first, the staff sends a pollution level prediction request related to the target area to the monitoring and early warning platform 20 through the terminal device 40, so that the monitoring and early warning platform 20 responds to the pollution level prediction request sent by the terminal device 40 and sends a target monitoring point determination request to the GIS platform 30.

[0041] Thus, the GIS platform 30 responds to the target monitoring point determination request sent by the monitoring and early warning platform 20, and extracts the pollution information corresponding to the target area from the remote sensing image corresponding to the target area. The pollution information corresponding to the target area includes, for example, river water resource information of the target area, historical pollution source information of the target area, and pollutant development of the target area. Then, based on the pollution information corresponding to the target area, the GIS platform 30 selects monitoring points where pollutants may appear, and uses the selected monitoring points as target monitoring points.

[0042] Furthermore, the GIS platform 30 sends the location information of the target monitoring point to the monitoring and early warning platform 20, and when the monitoring and early warning platform 20 receives the location information of the target monitoring point, the location information of the target monitoring point is further sent to the terminal device 40. Thus, when the staff views the location information of the target monitoring point through the terminal device 40, the water quality detector 10 can be set at the relevant position in the target area according to the displayed location information. Among them, the water quality detector 10 is used to detect the composition and concentration of pollutants at the corresponding position.

[0043] It is worth noting that the pollutant components detected by the water quality detector 10 correspond to the predetermined pollutant indicators. For example, first, the GIS platform 30 collects influencing indicators related to water environment pollution in the target area. Among them, the above-mentioned influencing indicators include: PH, COD, suspended matter, ammonia nitrogen, oils and some heavy metals. Then, the GIS platform 30 sends the above-mentioned influencing indicators to the data acquisition module in the monitoring and early warning platform 20, and when the data acquisition module receives the above-mentioned influencing indicators, the above-mentioned influencing indicators are further screened, so as to select pollutant indicators that can reflect the water environment conditions of the target area.

[0044] The data collection module includes but is not limited to using principal component analysis to screen the above-mentioned influencing indicators, and determining the contribution of each influencing indicator to the water environment pollution in the target area, and selecting the one with the largest contribution as the pollutant indicator. The data collection module can also use SPSS to determine the correlation between the various influencing indicators, and select relatively independent and representative influencing indicators as pollutant indicators.

[0045] Thus, the data acquisition module in the monitoring and early warning module 20 can collect pollutant concentration data corresponding to various pollutant indicators in the current state from multiple water quality detectors 10 set in the target area, such as the concentration corresponding to suspended matter, the concentration corresponding to heavy metals, the concentration corresponding to oils, etc.

[0046] When the data collection module collects pollutant concentration data corresponding to each pollutant index, a first graph structure for representing pollutant migration and a second graph structure for representing pollutant diffusion and degradation are generated based on the pollutant concentration data (S402).

[0047] Specifically, the first graph structure includes multiple nodes corresponding to each target monitoring point ~ And the edges between adjacent nodes ~ In addition, it is worth noting that the movement of pollutants is often related to the direction and speed of water flow. ~ Therefore, the edges between adjacent nodes ~ Used to represent the weight relationship between adjacent nodes ~ and water flow speed ~ There is a directional relationship between adjacent nodes.

[0048] Figure 5 is a schematic diagram of the first graph structure according to an embodiment of the present application. Figure 5 As shown, for example, the node With Node The water flow direction relationship between them is that the water flows from the node Flow to Node , and the node With Node There is a weight relationship between and water flow speed For another example, the node With Node The water flow direction relationship between them is that the water flows from the node Flow to Node , and the node With Node There is a weight relationship between and water flow speed . Thus, due to the node With Node The direction between nodes With Node The directions are not the same, so it can be seen that pollutants do not migrate in just a fixed direction, but may migrate in multiple different directions.

[0049] In addition, it is worth noting that due to space limitations, the above Figure 5 Only nodes are shown ~ , it should be clear to those skilled in the art that there are other nodes ~ .

[0050] Specifically, the second graph structure includes multiple nodes corresponding to each target monitoring point ~ And the edges between adjacent nodes ~ In addition, it is worth noting that since pollutant diffusion and degradation are independent of water flow velocity and direction, they are only related to the distance between adjacent nodes (i.e., adjacent target monitoring points). ~ Therefore, the edges between adjacent nodes ~ Used to represent the weight relationship between adjacent nodes ~ And the distance relationship ~ , and there is no directional relationship between adjacent nodes.

[0051] Figure 6 is a schematic diagram of the second graph structure according to an embodiment of the present application. Figure 6 As shown, for example, the node With Node There is a weight relationship between Relationship with distance For another example, the node With Node There is a weight relationship between Relationship with distance .

[0052] In addition, it is worth noting that due to space limitations, the above Figure 6 Only nodes are shown ~ , it should be clear to those skilled in the art that there are other nodes ~ .

[0053] The above-mentioned specific operation steps for generating the first graph structure and the second graph structure will be described in detail later, so they will not be repeated here.

[0054] Afterwards, the first graph neural network module in the monitoring and early warning platform 20 constructs a first graph neural network corresponding to the first graph structure, and inputs the first graph structure into the first graph neural network, thereby outputting a third graph structure (S404). Specifically, the first graph neural network module constructs the first graph neural network and determines the calculation rules of the first graph neural network. The calculation rules of the first graph neural network are as follows: Pollutant concentration at the nth recursive iteration: ; Among them, reference Figure 5 As can be seen from the above formula, Indicates that in the nth recursive iteration, with the node The corresponding pollutant concentration is Indicates that in the n-1th recursive iteration, with the node The corresponding pollutant concentration is Indicates that in the nth recursive iteration, from node To Node The concentration of pollutants that migrate, Indicates that in the nth recursive iteration, from node To Node The concentration of the migrated pollutants. Where n=1~N, and where N represents the number of times the first graph neural network recursively iterates the first graph structure.

[0055] For nodes Pollutant concentration at the nth recursive iteration: ; Among them, reference Figure 5 As can be seen from the above formula, Indicates that in the nth recursive iteration, with the node The corresponding pollutant concentration is Indicates that in the n-1th recursive iteration, from node To Node The concentration of pollutants that migrate, Indicates that in the n-1th recursive iteration, from node To Node The concentration of pollutants that migrate, Indicates that in the n-1th recursive iteration, from node To Node The concentration of the migrated pollutants. Where n=1~N, and where N represents the number of times the first graph neural network recursively iterates the first graph structure.

[0056] For nodes Pollutant concentration at the nth recursive iteration: ; Among them, reference Figure 5 As can be seen from the above formula, Indicates that in the nth recursive iteration, with the node The corresponding pollutant concentration is Indicates that in the n-1th recursive iteration, from node To Node The concentration of pollutants that migrate, Indicates that in the n-1th recursive iteration, from node To Node The concentration of pollutants that migrate, Indicates that in the n-1th recursive iteration, from node To Node The concentration of pollutants that migrate, Indicates that in the n-1th recursive iteration, from node To Node The concentration of the migrated pollutants. Where n=1~N, and where N represents the number of times the first graph neural network recursively iterates the first graph structure.

[0057] For nodes Pollutant concentration at the nth recursive iteration: ; Among them, reference Figure 5 As can be seen from the above formula, Indicates that in the nth recursive iteration, with the node The corresponding pollutant concentration is Indicates that in the n-1th recursive iteration, from node To Node The concentration of pollutants that migrate, Indicates that in the n-1th recursive iteration, from node To Node The concentration of pollutants that migrate, Indicates that in the n-1th recursive iteration, from node To Node The concentration of the migrated pollutants. Where n=1~N, and where N represents the number of times the first graph neural network recursively iterates the first graph structure.

[0058] Similarly, the above calculation rules can be used to determine the The pollutant concentration at the nth recursive iteration. It is worth noting that since a water quality detector 10 is set at each target monitoring point corresponding to each node, the initial concentration can be determined based on the pollutant components detected by the water quality detector 10 and the concentrations corresponding to the corresponding components. That is, in this embodiment, when n=1, the node Corresponding pollutant concentration ,node Corresponding pollutant concentration ,...,node Corresponding pollutant concentration It can be determined based on the water quality detector 10.

[0059] Thus, when the first graph structure is recursively iterated N times using the first graph neural network, a third graph structure is output. And in the third graph structure, The corresponding pollutant concentration is , and the node The corresponding pollutant concentration is , and the node The corresponding pollutant concentration is , ..., with nodes The corresponding pollutant concentration is .

[0060] Afterwards, the second graph neural network module in the monitoring and early warning platform 20 constructs a second graph neural network corresponding to the second graph structure, and inputs the second graph structure into the second graph neural network, thereby outputting a fourth graph structure (S406). Specifically, the second graph neural network module constructs the second graph neural network and determines the calculation rules of the second graph neural network. The calculation rules of the second graph neural network are as follows: Pollutant concentration at the nth recursive iteration: ; Among them, reference Figure 6 As can be seen from the above formula, Indicates that in the nth recursive iteration, with the node The corresponding pollutant concentration is Indicates that in the n-1th recursive iteration, with the node The corresponding pollutant concentration is Indicates that in the n-1th recursive iteration, the node The concentration of pollutants to be degraded, Indicates that in the n-1th recursive iteration, the node To Node The concentration of diffuse pollutants, Indicates that in the n-1th recursive iteration, the node To Node The concentration of diffused pollutants. Where n=1~N, and where N represents the number of times the second graph neural network recursively iterates the second graph structure.

[0061] Among them, it is worth noting that since the degradation process of pollutants corresponding to each node is only related to its own pollutant concentration, but has nothing to do with the pollutant concentrations of other nodes, when determining the calculation rules corresponding to the first graph neural network, only the degradation coefficient corresponding to the node and the pollutant concentration of the node at the last recursive iteration are used to calculate the pollutant concentration corresponding to the node under the current number of recursive iterations.

[0062] For nodes Pollutant concentration at the nth recursive iteration: ; Among them, reference Figure 6 As can be seen from the above formula, Indicates that in the nth recursive iteration, with the node The corresponding pollutant concentration is Indicates that in the n-1th recursive iteration, the node To Node The concentration of diffuse pollutants, Indicates that in the n-1th recursive iteration, the node The concentration of pollutants to be degraded, Indicates that in the n-1th recursive iteration, the node To Node The concentration of diffuse pollutants, Indicates that in the n-1th recursive iteration, the node To Node The concentration of diffused pollutants. Where n=1~N, and where N represents the number of times the second graph neural network recursively iterates the second graph structure.

[0063] For nodes Pollutant concentration at the nth recursive iteration: ; Among them, reference Figure 6 As can be seen from the above formula, Indicates that in the nth recursive iteration, with the node The corresponding pollutant concentration is Indicates that in the n-1th recursive iteration, the node To Node The concentration of diffuse pollutants, Indicates that in the n-1th recursive iteration, the node To Node The concentration of diffuse pollutants, Indicates that in the n-1th recursive iteration, the node The concentration of pollutants to be degraded, Indicates that in the n-1th recursive iteration, the node To Node The concentration of diffuse pollutants, Indicates that in the n-1th recursive iteration, the node To Node The concentration of diffused pollutants. Where n=1~N, and where N represents the number of times the second graph neural network recursively iterates the second graph structure.

[0064] For nodes Pollutant concentration at the nth recursive iteration: ; Among them, reference Figure 6 As can be seen from the above formula, Indicates that in the nth recursive iteration, with the node The corresponding pollutant concentration is Indicates that in the n-1th recursive iteration, the node To Node The concentration of diffuse pollutants, Indicates that in the n-1th recursive iteration, the node To Node The concentration of diffuse pollutants, Indicates that in the n-1th recursive iteration, the node The concentration of pollutants to be degraded, Indicates that in the n-1th recursive iteration, the node To Node The concentration of diffused pollutants. Where n=1~N, and where N represents the number of times the second graph neural network recursively iterates the second graph structure.

[0065] Similarly, the above calculation rules can be used to determine the The pollutant concentration at the nth recursive iteration. It is worth noting that since water quality detection 10 is set at the target monitoring point corresponding to each node, the initial concentration can be determined based on the pollutant components detected by the water quality detection instrument 10 and the concentration corresponding to the corresponding components. That is, in this embodiment, when n=1, the node Corresponding pollutant concentration ,node Corresponding pollutant concentration ,...,node The corresponding pollutant concentration It can be determined based on the water quality detector 10.

[0066] Thus, when the second graph structure is recursively iterated N times using the second graph neural network, a fourth graph structure is output. And in the fourth graph structure, The corresponding pollutant concentration is , and the node The corresponding pollutant concentration is , and the node The corresponding pollutant concentration is , ..., with nodes The corresponding pollutant concentration is .

[0067] Afterwards, when the third graph structure is outputted based on the first graph structure by using the first graph neural network module, and the fourth graph structure is outputted based on the second graph structure by using the second graph neural network module, the fusion module further extracts and fuses the features of the third graph structure and the fourth graph structure, and generates a first pollutant concentration feature vector corresponding to each target monitoring point (S408). First, the third graph structure is feature extracted using a pre-trained neural network model to generate a second pollutant concentration feature vector corresponding to each target monitoring point. Then, the fourth graph structure is feature extracted using a pre-trained neural network model to generate a third pollutant concentration feature vector corresponding to each target monitoring point. Finally, the second pollutant concentration feature vector and the third pollutant concentration feature vector are fused using MLP to generate a first pollutant concentration feature vector. The above content will be described in detail later, so it will not be repeated here.

[0068] Finally, when the fusion module outputs the first pollutant concentration feature vector, the risk prediction module inputs the first pollutant concentration feature vector into a pre-trained risk prediction model (wherein the risk prediction model includes, for example, an RNN model, a fully connected layer, and multiple sofmax classifiers) to determine the pollution degree corresponding to each target monitoring point. The pollution degree corresponding to each target monitoring point can be, for example, the probability of pollution occurring at the target monitoring point and the probability of no pollution occurring. The higher the probability of pollution occurring, the greater the possibility of pollution occurring at the target monitoring point, and the more serious the pollution.

[0069] Thus, when the risk prediction module outputs the pollution degree corresponding to each target monitoring point, the early warning module determines whether an early warning is needed based on the pollution degree corresponding to each target monitoring point. For example, when the early warning module determines that an early warning is needed based on any one of the multiple target monitoring points, an early warning is issued to the staff through the terminal device 40.

[0070] As described in the background technology, although it is possible to determine the main pollutants in the water environment of the target area by investigating and evaluating the pollution sources, it is impossible to locate the pollution sources, and it is impossible to make an accurate assessment and prediction of the water quality of the water environment in the target area. For another example, although it is possible to determine the degree of pollution in the water environment of the target area by entrusting a third-party agency to conduct sampling and analysis, the water quality monitoring equipment can only collect water quality data from discrete sites and cannot obtain more comprehensive water quality data. For another example, since the water environment is open and complex, pollutants will diffuse and migrate, and the effects between the various pollutants are complex and there are many coupling factors, so how to accurately predict the changes in the degree of water environment pollution and monitor and warn based on the prediction results is a problem that needs to be solved urgently.

[0071] In view of this, referring to the above-mentioned contents, it can be known that the present application generates a first graph structure for representing the migration and movement of pollutants and a second graph structure for representing the diffusion and degradation of pollutants. And because the migration and movement of pollutants between two adjacent target monitoring points will affect each other, when the first graph neural network is used to process the data of the first graph structure, the association relationship between the two adjacent nodes (corresponding to the target monitoring points) (i.e., the edge between the two adjacent nodes) can be retained, so that the third graph structure finally outputted not only retains the data information of each node, but also retains the structural information between the two adjacent nodes.

[0072] Similarly, since the diffusion of pollutants between two adjacent target monitoring points will affect each other, when the second graph neural network is used to process the data of the second graph structure, the association relationship between the two adjacent nodes (corresponding to the target monitoring points) (that is, the edge between the two adjacent nodes) can be retained, so that the fourth graph structure finally output not only retains the data information of each node, but also retains the structural information between the two adjacent nodes.

[0073] Therefore, when the third graph structure and the fourth graph structure are fused by MLP, the first pollutant concentration feature vector corresponding to each target monitoring point finally generated contains relatively comprehensive feature information. And when the first pollutant concentration feature vector is input into the pre-trained risk prediction model, the risk prediction model can accurately predict the pollution degree corresponding to each target monitoring point. Finally, based on the pollution degree corresponding to each target monitoring point, it is selected whether to issue an early warning.

[0074] This solves the technical problem in the prior art that it is currently impossible to accurately predict changes in the degree of water pollution and to perform monitoring and early warning based on the prediction results.

[0075] Optionally, the first pollutant concentration characteristic vector is input into a pre-trained risk prediction model to determine the pollution degree corresponding to each target monitoring point, and based on the pollution degree corresponding to each target monitoring point, it is determined whether to issue an early warning. The operation includes: inputting the first pollutant concentration characteristic vector into a pre-trained risk prediction model and outputting the pollution degree probability corresponding to each target monitoring node; judging whether the pollution degree probability corresponding to each target monitoring node is greater than a pre-set pollution degree threshold; and issuing an early warning when the pollution degree probability is greater than the pollution degree threshold.

[0076] Specifically, Figure 7 Schematic diagram of the risk prediction model according to the embodiment of the present application. Figure 7 As shown, the risk prediction model includes, for example, an RNN model, a fully connected layer, and multiple softmax classifiers. Among them, the fully connected layer includes 2f neurons. Therefore, when the first pollutant concentration feature vector is input into the risk prediction model, the fully connected layer can output a vector for representing the degree of pollution and a vector for the degree of non-pollution for each target monitoring point. Therefore, when the fully connected layer inputs two vectors corresponding to each target monitoring point (i.e., a vector for representing the degree of pollution and a vector for the degree of non-pollution) into the corresponding softmax classifier, the softmax classifier can output the probability of the degree of pollution corresponding to each target monitoring point based on the two vectors corresponding to each target monitoring point.

[0077] For example, based on the first pollutant concentration feature vector, the risk prediction model can output the pollution degree probability corresponding to each target monitoring point. And wherein, the risk prediction model can output two corresponding probabilities for each target monitoring point. Figure 7 As shown, the target monitoring point Corresponding probability and probability (i.e., probability of contamination degree). Among them, probability Used to indicate target monitoring points Probability of contamination level Used to indicate the probability of target monitoring point The probability of uncontaminated level. And among them, the probability and probability The sum is 100%.

[0078] Therefore, when the risk prediction module outputs the pollution degree probability corresponding to each target monitoring node, the pollution degree probability is further input into the alarm module.

[0079] The alarm module then determines whether the probability of the pollution level corresponding to each target monitoring point is greater than a preset threshold. Corresponding probability and probability (i.e., pollution degree probability), and the pollution degree threshold is pre-set in the alarm module The alarm module will Pollution level threshold For comparison, if the probability Greater than the preset pollution level threshold , it means that the target monitoring point The corresponding location may be contaminated by pollutants, so the alarm module sends a first-level alarm to the staff through the terminal device 40.

[0080] In addition, it is worth noting that if the probability The larger the probability The smaller, that is, the probability With probability The larger the difference between the two, the closer it is to the target monitoring point. The pollution situation at the corresponding location will be more serious, so the alarm module sends a secondary alarm to the staff through the terminal device 40.

[0081] If more than half of the target monitoring points in the target area are shown to be potentially contaminated, it means that most locations in the target area are likely to be contaminated by pollutants, so the alarm module sends a third-level alarm to the staff through the terminal device 40.

[0082] Therefore, by using the pre-set risk prediction model to predict the probability of pollution level at each target monitoring point, and based on the prediction results, judging whether it is necessary to alert the staff through the terminal equipment, the technical effect of being able to promptly reflect possible pollution situations to the staff and remind the staff to deal with them as soon as possible, thereby avoiding pollution of the water environment in the target area.

[0083] Optionally, the operations of extracting features from and fusing the third graph structure and the fourth graph structure, and generating a first pollutant concentration feature vector corresponding to each target monitoring point, include: extracting features from the third graph structure, and generating a second pollutant concentration feature vector corresponding to each target monitoring point; extracting features from the fourth graph structure, and generating a third pollutant concentration feature vector corresponding to each target monitoring point; and fusing the second pollutant concentration feature vector and the third pollutant concentration feature vector using MLP to generate a first pollutant concentration feature vector.

[0084] Specifically, when the fusion module receives the third graph structure, the pre-trained neural network model is further used to extract features from the third graph structure. The third graph structure is used to indicate the graph structure generated after the first graph neural network module performs N recursive iterations on the first graph structure. The corresponding pollutant concentration is , and the node The corresponding pollutant concentration is , and the node The corresponding pollutant concentration is , ..., with nodes The corresponding pollutant concentration is And the pollutant concentration corresponding to each target monitoring point after N recursive iterations can be determined based on the calculation rules corresponding to the first graph neural network.

[0085] Therefore, after the pre-trained neural network model extracts features from the third graph structure, the second pollutant concentration feature vector G corresponding to the multiple target monitoring points can be obtained. .

[0086] Similarly, when the fusion module receives the fourth graph structure, it further uses the pre-trained neural network model to extract features from the fourth graph structure. The fourth graph structure is used to indicate the graph structure generated after the second graph neural network module performs N recursive iterations on the second graph structure. The corresponding pollutant concentration is , and the node The corresponding pollutant concentration is , and the node The corresponding pollutant concentration is , ..., with nodes The corresponding pollutant concentration is And the pollutant concentration corresponding to each target monitoring point after N recursive iterations can be determined based on the calculation rules corresponding to the second graph neural network.

[0087] Therefore, after the pre-trained neural network model extracts features from the fourth graph structure, the third pollutant concentration feature vector H corresponding to the multiple target monitoring points can be obtained. .

[0088] Finally, MLP is used to analyze the second pollutant concentration feature vector G and the third pollutant concentration characteristic vector H Fusion is performed to generate the first pollutant concentration feature vector C The first pollutant concentration characteristic vector includes pollutant migration information corresponding to the second pollutant concentration characteristic vector and pollutant diffusion and degradation information corresponding to the third pollutant concentration characteristic vector.

[0089] Thus, after the fusion module extracts features from the third graph structure and the fourth graph structure respectively and generates the corresponding second pollutant concentration feature vector and the third pollutant concentration feature vector, the MLP can be further used to fuse the second pollutant concentration feature vector and the third pollutant concentration feature vector, thereby generating a first pollutant feature vector that includes both pollutant migration information and pollutant diffusion and degradation information in the target area. That is, since the first pollutant feature vector contains more comprehensive pollutant information, the prediction result generated based on the first pollutant feature vector is more accurate.

[0090] Optionally, the operation of generating a first graph structure for representing the migration movement of pollutants based on multiple target monitoring points includes: determining multiple nodes for generating the first graph structure based on multiple pre-set target monitoring points; determining edges between each adjacent node based on a first association relationship between each adjacent node, wherein the first association relationship is used to indicate a weight relationship and a water flow velocity relationship between each adjacent node; and generating the first graph structure when determining multiple nodes and edges between each adjacent node.

[0091] Specifically, referring to the above content, it can be known that since multiple target monitoring points in the target area are predetermined, each target monitoring point can be determined as a plurality of nodes in the first graph structure. With the nodes in the first graph structure Corresponding, target monitoring point With the nodes in the first graph structure Corresponding to,..., target monitoring point With the nodes in the first graph structure correspond.

[0092] When multiple nodes in the first graph structure are determined, the edges between the adjacent nodes can be determined based on the weight relationship between the adjacent nodes and the water flow velocity relationship. Figure 5 As shown, since the migration of pollutants is often related to the water flow speed and water flow direction, the migration of pollutants between adjacent nodes is related to the water flow speed. In addition, since a node is often adjacent to multiple nodes, there are corresponding weight coefficients between adjacent nodes. It is worth noting that in the first graph structure, the weight relationship corresponds to the weight coefficient.

[0093] For example, the node Respectively with nodes and nodes adjacent, so the nodes With Node There is a relationship between water velocity And weight relationship , and the node With Node The relationship between water velocity And weight relationship Composition Node With Node The edges between nodes With Node There is a relationship between water velocity And weight relationship , and the node With Node The relationship between water velocity And weight relationship Composition Node With Node The edge between.

[0094] And so on, other nodes ~ There will also be edges between adjacent nodes in . Thus, when multiple nodes and edges between adjacent nodes are determined, a first graph structure can be generated.

[0095] Optionally, the operation of generating a second graph structure for representing the diffusion and degradation of pollutants based on multiple target monitoring points includes: determining multiple nodes for generating the second graph structure based on multiple pre-set target monitoring points; determining edges between each adjacent node based on a second association relationship between each adjacent node, wherein the second association relationship is used to indicate a weight relationship and a distance between each adjacent node; and generating the second graph structure when the edges between multiple nodes and each adjacent node are determined.

[0096] Specifically, referring to the above content, it can be known that since multiple target monitoring points in the target area are predetermined, each target monitoring point can be determined as a plurality of nodes in the first graph structure. With the nodes in the first graph structure Corresponding, target monitoring point With the nodes in the first graph structure Corresponding to,..., target monitoring point With the nodes in the first graph structure correspond.

[0097] When determining multiple nodes in the second graph structure, the edges between the adjacent nodes can be determined based on the weight relationship and distance relationship between the adjacent nodes. Figure 6 As shown, since the diffusion movement of pollutants is often related to the distance between adjacent target monitoring points, the diffusion movement of pollutants between each adjacent node is related to the distance. In addition, since a node is often adjacent to multiple nodes, there is a weight relationship between each adjacent node. It is worth noting that in the second graph structure, the weight relationship includes not only the weight coefficient between two adjacent nodes, but also the diffusion coefficient and the degradation coefficient. And among them, the diffusion coefficient includes, for example, the molecular diffusion coefficient, the turbulent diffusion coefficient and the dispersion coefficient.

[0098] For example, the node Respectively with nodes and nodes adjacent, so the nodes With Node There is a distance relationship between And weight relationship , and the node With Node The distance relationship between And weight relationship Composition Node With Node The edges between nodes With Node There is a distance relationship between And weight relationship , and the node With Node The distance relationship between And weight relationship Composition Node With Node Between the edges.

[0099] And so on, other nodes ~ There will also be edges between adjacent nodes in . Thus, when multiple nodes and edges between adjacent nodes are determined, a second graph structure can be generated.

[0100] Optionally, the operation of determining multiple target monitoring points pre-set in the target area includes: collecting remote sensing images corresponding to the target area, and extracting pollution information corresponding to the target area from the remote sensing images using a GIS system; and using the GIS system and based on the pollution information, selecting monitoring points where pollutants appear, and using the selected monitoring points as target monitoring points.

[0101] Specifically, first, when the monitoring and early warning platform 20 responds to the pollution degree detection request sent by the terminal device 40, it sends a target monitoring point determination request to the GIS platform 30. The GIS platform 30 responds to the request and collects a remote sensing image corresponding to the target area. The GIS platform 30 is provided with a GIS system, and the remote sensing image can be pre-stored in the GIS system, for example.

[0102] Furthermore, the GIS platform 30 extracts pollution information corresponding to the target area from the remote sensing image corresponding to the target area, wherein the pollution information includes, for example, river water resource information of the target area, historical pollution source information of the target area, and pollutant development status of the target area.

[0103] The GIS platform 30 then selects monitoring points where pollutants may appear or where pollutants have appeared before based on the pollution information corresponding to the target area, and uses the selected monitoring points as target monitoring points.

[0104] Thus, by utilizing the GIS platform to determine the operation of the target monitoring points, the technical effect of ensuring the generation of the first map structure and the second map structure and providing the necessary basis for subsequent prediction of the degree of pollution in the target area is achieved.

[0105] Therefore, according to the first aspect of this embodiment, the technical effect of being able to accurately predict changes in the degree of water environment pollution and to perform monitoring and early warning based on the prediction results is achieved.

[0106] In addition, reference Figure 1As shown, according to the second aspect of this embodiment, a storage medium is provided, wherein the storage medium includes a stored program, wherein when the program is run, a processor executes any one of the above methods.

[0107] Therefore, according to this embodiment, the technical effect of being able to accurately predict changes in the degree of water environment pollution and perform monitoring and early warning based on the prediction results is achieved.

[0108] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0109] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention. Example 2

[0110] Figure 8 FIG. 8 shows a monitoring and early warning device 800 for water environment pollution according to this embodiment, and the device 800 corresponds to the method according to embodiment 1. Figure 8As shown, the device 800 includes: a graph structure generation module 810, which is used to determine multiple target monitoring points pre-set in the target area, and based on the multiple target monitoring points, respectively generate a first graph structure for representing the migration movement of pollutants and a second graph structure for representing the diffusion and degradation of pollutants; a first graph structure output module 820, which is used to construct a first graph neural network corresponding to the first graph structure, and input the first graph structure into the first graph neural network, thereby outputting a third graph structure; a second graph structure output module 830, which is used to construct a second graph neural network corresponding to the second graph structure, and input the second graph structure into the second graph neural network, thereby outputting a fourth graph structure; a fusion module 840, which is used to extract and fuse features of the third graph structure and the fourth graph structure, and generate a first pollutant concentration feature vector corresponding to each target monitoring point; and an early warning module 850, which is used to input the first pollutant concentration feature vector into a pre-trained risk prediction model, determine the pollution degree corresponding to each target monitoring point, and determine whether to issue an early warning based on the pollution degree corresponding to each target monitoring point.

[0111] Optionally, the early warning module 850 includes: a pollution degree probability output module, which is used to input the pollutant concentration feature vector into a pre-trained risk prediction model and output the pollution degree probability corresponding to each target monitoring node; a judgment module, which is used to judge whether the pollution degree probability corresponding to each target monitoring node is greater than a pre-set pollution degree threshold; and an early warning sub-module, which is used to issue an early warning when the pollution degree probability is greater than the pollution degree threshold.

[0112] Optionally, the fusion module 840 includes: a first feature extraction module, used to extract features from the third graph structure and generate a second pollutant concentration feature vector corresponding to each target monitoring point; a second feature extraction module, used to extract features from the fourth graph structure and generate a third pollutant concentration feature vector corresponding to each target monitoring point; and a fusion submodule, used to use MLP to fuse the second pollutant concentration feature vector and the third pollutant concentration feature vector to generate a first pollutant concentration feature vector.

[0113] Optionally, the graph structure generation module 810 includes: a first node determination module, used to determine multiple nodes for generating a first graph structure based on multiple pre-set target monitoring points; a first edge determination module, used to determine the edges between each adjacent node based on a first association relationship between each adjacent node, wherein the first association relationship is used to indicate the weight relationship and water flow velocity relationship between each adjacent node; and a first graph structure generation module, used to generate a first graph structure when determining multiple nodes and edges between each adjacent node.

[0114] Optionally, the graph structure generation module 810 includes: a second node generation module, used to determine multiple nodes for generating a second graph structure based on multiple pre-set target monitoring points; a second edge generation module, used to determine the edges between each adjacent node based on the second association relationship between each adjacent node, wherein the second association relationship is used to indicate the weight relationship and distance between each adjacent node; and a second graph structure generation module, used to generate a second graph structure when determining multiple nodes and the edges between each adjacent node.

[0115] Optionally, the graph structure generation module 810 includes: a pollution information extraction module, which is used to collect remote sensing images corresponding to the target area, and use the GIS system to extract pollution information corresponding to the target area from the remote sensing images; and a target monitoring point determination module, which is used to use the GIS system and based on the pollution information to determine the target monitoring points.

[0116] Therefore, according to this embodiment, the technical effect of being able to accurately predict changes in the degree of water environment pollution and perform monitoring and early warning based on the prediction results is achieved. Example 3

[0117] Fig. 9 FIG. 9 shows a monitoring and early warning device 900 for water environment pollution according to this embodiment, and the device 900 corresponds to the method according to embodiment 1. Fig. 9 As shown, the device 900 includes: a processor 910; and a memory 920, which is connected to the processor 910 and is used to provide the processor 910 with instructions for processing the following processing steps: determining multiple target monitoring points pre-set in the target area, and based on the multiple target monitoring points, respectively generating a first graph structure for representing the migration movement of pollutants and a second graph structure for representing the diffusion and degradation of pollutants; constructing a first graph neural network corresponding to the first graph structure, and inputting the first graph structure into the first graph neural network, thereby outputting a third graph structure; constructing a second graph neural network corresponding to the second graph structure, and inputting the second graph structure into the second graph neural network, thereby outputting a fourth graph structure; performing feature extraction and fusion on the third graph structure and the fourth graph structure, and generating a first pollutant concentration feature vector corresponding to each target monitoring point; and inputting the first pollutant concentration feature vector into a pre-trained risk prediction model to determine the degree of pollution corresponding to each target monitoring point, and based on the degree of pollution corresponding to each target monitoring point, determining whether to issue an early warning.

[0118] Optionally, the first pollutant concentration characteristic vector is input into a pre-trained risk prediction model to determine the pollution degree corresponding to each target monitoring point, and based on the pollution degree corresponding to each target monitoring point, it is determined whether to issue an early warning, including: inputting the first pollutant concentration characteristic vector into a pre-trained risk prediction model, and outputting the pollution degree probability corresponding to each target monitoring node; judging whether the pollution degree probability corresponding to each target monitoring node is greater than a pre-set pollution degree threshold; and issuing an early warning when the pollution degree probability is greater than the pollution degree threshold.

[0119] Optionally, the operations of extracting features from and fusing the third graph structure and the fourth graph structure, and generating a first pollutant concentration feature vector corresponding to each target monitoring point, include: extracting features from the third graph structure, and generating a second pollutant concentration feature vector corresponding to each target monitoring point; extracting features from the fourth graph structure, and generating a third pollutant concentration feature vector corresponding to each target monitoring point; and fusing the second pollutant concentration feature vector and the third pollutant concentration feature vector using MLP to generate a first pollutant concentration feature vector.

[0120] Optionally, the operation of generating a first graph structure for representing the migration movement of pollutants based on multiple target monitoring points includes: determining multiple nodes for generating the first graph structure based on multiple pre-set target monitoring points; determining edges between each adjacent node based on a first association relationship between each adjacent node, wherein the first association relationship is used to indicate a weight relationship and a water flow velocity relationship between each adjacent node; and generating the first graph structure when determining multiple nodes and edges between each adjacent node.

[0121] Optionally, the operation of generating a second graph structure for representing the diffusion and degradation of pollutants based on multiple target monitoring points includes: determining multiple nodes for generating the second graph structure based on multiple pre-set target monitoring points; determining edges between each adjacent node based on a second association relationship between each adjacent node, wherein the second association relationship is used to indicate a weight relationship and a distance between each adjacent node; and generating the second graph structure when the edges between multiple nodes and each adjacent node are determined.

[0122] Optionally, the operation of determining multiple pre-set target monitoring points in the target area includes: collecting remote sensing images corresponding to the target area, and extracting pollution information corresponding to the target area from the remote sensing images using a GIS system; and determining the target monitoring points based on the pollution information using the GIS system.

[0123] Therefore, according to this embodiment, the technical effect of being able to accurately predict changes in the degree of water environment pollution and perform monitoring and early warning based on the prediction results is achieved.

[0124] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0125] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0126] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0127] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0128] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0129] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.

[0130] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A monitoring and early warning method for water environment pollution, characterized in that: include: Determine a plurality of target monitoring points pre-set in the target area, and based on the plurality of target monitoring points, respectively generate a first graph structure for representing the movement of pollutants and a second graph structure for representing the diffusion and degradation of pollutants; Constructing a first graph neural network corresponding to the first graph structure, and inputting the first graph structure into the first graph neural network, thereby outputting a third graph structure; constructing a second graph neural network corresponding to the second graph structure, and inputting the second graph structure into the second graph neural network, thereby outputting a fourth graph structure; Extracting and fusing the third graph structure and the fourth graph structure, and generating a first pollutant concentration feature vector corresponding to each target monitoring point; as well as The first pollutant concentration feature vector is input into a pre-trained risk prediction model to determine the pollution degree corresponding to each target monitoring point, and based on the pollution degree corresponding to each target monitoring point, it is determined whether to issue an early warning.

2. The method according to claim 1, characterized in that: Inputting the first pollutant concentration feature vector into a pre-trained risk prediction model, determining the pollution degree corresponding to each target monitoring point, and determining whether to issue an early warning based on the pollution degree corresponding to each target monitoring point, includes: Inputting the first pollutant concentration feature vector into a pre-trained risk prediction model, and outputting the pollution degree probability corresponding to each target monitoring node; Determining whether the pollution degree probability corresponding to each target monitoring node is greater than a preset pollution degree threshold; and When the pollution level probability is greater than the pollution level threshold, an early warning is issued.

3. The method according to claim 2, characterized in that The operation of extracting and fusing the features of the third graph structure and the fourth graph structure and generating a first pollutant concentration feature vector corresponding to each target monitoring point includes: Extracting features from the third graph structure and generating second pollutant concentration feature vectors corresponding to each target monitoring point; Extracting features from the fourth graph structure and generating third pollutant concentration feature vectors corresponding to the respective target monitoring points; and The second pollutant concentration feature vector and the third pollutant concentration feature vector are fused using MLP to generate the first pollutant concentration feature vector.

4. The method according to claim 3, characterized in that The operation of generating a first graph structure for representing the migration movement of pollutants based on multiple target monitoring points includes: Based on a plurality of preset target monitoring points, determining a plurality of nodes for generating the first graph structure; Determine the edge between each adjacent node based on the first association relationship between each adjacent node, wherein the first association relationship is used to indicate the weight relationship and the water flow velocity relationship between each adjacent node; and The first graph structure is generated when edges between the plurality of nodes and the respective adjacent nodes are determined.

5. The method according to claim 3, characterized in that: Based on multiple target monitoring points, an operation of generating a second graph structure for representing pollutant diffusion and degradation includes: Based on a plurality of preset target monitoring points, determining a plurality of nodes for generating the second graph structure; Determine the edge between each adjacent node based on the second association relationship between each adjacent node, wherein the second association relationship is used to indicate the weight relationship and the distance between each adjacent node; and The second graph structure is generated when the edges between the plurality of nodes and the respective adjacent nodes are determined.

6. The method according to claim 5, characterized in that Determine the operation of multiple target monitoring points pre-set in the target area, including: Collecting remote sensing images corresponding to the target area, and extracting pollution information corresponding to the target area from the remote sensing images using a GIS system; and The target monitoring point is determined by utilizing the GIS system and based on the pollution information.

7. A storage medium, characterized in that: The storage medium includes a stored program, wherein when the program is run, the processor executes the method according to any one of claims 1 to 6.

8. A monitoring and early warning device for water environment pollution, characterized in that: include: A graph structure generation module, used to determine a plurality of target monitoring points pre-set in a target area, and based on the plurality of target monitoring points, respectively generate a first graph structure for representing the movement of pollutants and a second graph structure for representing the diffusion and degradation of pollutants; a first graph structure output module, configured to construct a first graph neural network corresponding to the first graph structure, and input the first graph structure into the first graph neural network, thereby outputting a third graph structure; a second graph structure output module, used to construct a second graph neural network corresponding to the second graph structure, and input the second graph structure into the second graph neural network, thereby outputting a fourth graph structure; A fusion module, used for extracting and fusing features of the third graph structure and the fourth graph structure, and generating a first pollutant concentration feature vector corresponding to each target monitoring point; as well as The early warning module is used to input the first pollutant concentration feature vector into a pre-trained risk prediction model, determine the pollution degree corresponding to each target monitoring point, and determine whether to issue an early warning based on the pollution degree corresponding to each target monitoring point.

9. The device according to claim 8, characterized in that The early warning module comprises: A pollution degree probability output module, used to input the first pollutant concentration feature vector into a pre-trained risk prediction model, and output the pollution degree probability corresponding to each target monitoring node; A judgment module, used to judge whether the pollution degree probability corresponding to each target monitoring node is greater than a preset pollution degree threshold; The early warning submodule is used to issue an early warning when the probability of the pollution level is greater than the pollution level threshold.

10. A monitoring and early warning device for water environment pollution, characterized in that: include: processor; as well as A memory, connected to the processor, configured to provide the processor with instructions for processing the following processing steps: Determine a plurality of target monitoring points pre-set in the target area, and based on the plurality of target monitoring points, respectively generate a first graph structure for representing the movement of pollutants and a second graph structure for representing the diffusion and degradation of pollutants; Constructing a first graph neural network corresponding to the first graph structure, and inputting the first graph structure into the first graph neural network, thereby outputting a third graph structure; Constructing a second graph neural network corresponding to the second graph structure, and inputting the second graph structure into the second graph neural network, thereby outputting a fourth graph structure; Extracting and fusing the third graph structure and the fourth graph structure, and generating a first pollutant concentration feature vector corresponding to each target monitoring point; as well as The first pollutant concentration feature vector is input into a pre-trained risk prediction model to determine the pollution degree corresponding to each target monitoring point, and based on the pollution degree corresponding to each target monitoring point, it is determined whether to issue an early warning.

Images