Pollution source monitoring method and system based on neural network

Abstract

The invention discloses a pollution source monitoring method and system based on a neural network, and relates to the technical field of pollution source monitoring, and the method comprises the steps: dividing a preset range of a target drinking water source into a plurality of grid sub-regions through a GIS means, setting water quality monitoring point positions in the grid sub-regions, and obtaining the water quality monitoring data of each point position; constructing a digital twinborn model, judging whether a water quality abnormity alarm signal and a water quality abnormity early warning signal are generated or not according to the water quality monitoring result of each water quality monitoring point, and obtaining a correlation coefficient between the water quality abnormity early warning signal generated by each water quality monitoring point and the scene sub-sequence; adding a pollution source monitoring model through a neural network on the basis of the digital twin model; according to the method, the pollution source monitoring model is learned in real time, real-time monitoring and control are performed by using the sewage treatment whole-process model according to the prediction result of the pollution source monitoring model, effective early warning of sudden water source water quality pollution events such as natural disasters in the drinking water source is realized, and the pollution monitoring timeliness is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of pollution source monitoring, and specifically, to a pollution source monitoring method and system based on a neural network. Background Art

[0002] With the shortage of water resources and the increasing frequency of pollution, the problem of pollution in drinking water source areas has attracted more and more attention, and the pollution prevention and control work in source areas has become a key environmental issue currently concerned. There are many factors causing water source pollution, and natural disasters are one of the more important reasons. Most water source areas are located in mountainous areas where the climate is changeable. Affected by heavy rainfall, flash floods are extremely likely to occur. When flash floods occur, the rapidly rising water flow often carries sediment and stones, with very strong destructive power, easily causing large-area water source pollution. Due to the uncertain factors of natural disasters, the difficulty of monitoring and warning water source pollution increases.

[0003] The comparative document CN115100553A, "Method and System for Detecting and Processing River Surface Pollution Information Based on Convolutional Neural Network", includes the following steps: S1, collecting pictures of river floating objects and colored wastewater discharges to form a data set; S2, data marking to obtain the coordinate information of the rectangular frame and the floating objects and colored wastewater contained in the rectangular frame; S3, expanding the data set; S4, training the improved YOLOv4-tiny neural network model; S5, embedding the improved YOLOv4-tiny neural network model with the best performance into the MCU, and performing video detection on the monitored river or lake; S6, installing the MCU on a drone, and judging the pollution degree of the river or lake to be detected according to the number of detected floating objects and colored wastewater, which can effectively analyze the pollution source and has the characteristics of strong real-time performance and high accuracy.

[0004] The comparative document CN111474307A, "Pollutant Tracing Method, Device, Computer Equipment and Storage Medium", conducts online water quality monitoring on multiple target points in the supervision area. When it is detected that the water quality at a certain target point is abnormal, the certain target point is determined as a pollution monitoring point, and the water quality characteristics and characteristic pollutants of the pollution monitoring point are obtained; the target pollution industry is determined according to the water quality characteristics and the characteristic pollutants; the pollution source object is determined from the pollution tracing range corresponding to the pollution monitoring point according to the target pollution industry, and the pollution source object is the main body causing the water pollution event at the pollution monitoring point. This embodiment can quickly determine the pollution source object and improve the efficiency of pollutant tracing; Emergency monitoring of sudden pollution incidents in water source areas can provide information such as the types of pollutants, concentration distribution, affected areas, and development trends for accident handling decision-making departments. Therefore, in order to protect water source areas and effectively reduce the pollution degree of water source areas caused by natural disasters, it is particularly important to monitor natural disasters and water bodies before natural disasters through emergency monitoring means. Now, a pollution source monitoring method based on neural networks is provided. Summary of the Invention

[0005] In order to solve the above technical problems, the purpose of the present invention is to provide a pollution source monitoring method based on neural networks, including the following steps: Step S1: Divide the preset range of the target drinking water source area into several grid sub-areas through GIS means, set water quality monitoring points in the grid sub-areas, and obtain water quality monitoring data of each point. Step S2: Construct a digital twin model based on the water quality monitoring results of each water quality monitoring point and scenario information, judge whether to generate a water quality anomaly alarm signal and a water quality anomaly early warning signal according to the water quality monitoring results of each water quality monitoring point, and obtain the correlation coefficient between the water quality anomaly early warning signal generated by each water quality monitoring point and the scenario subsequence. Step S3: Add a pollution source monitoring model through a neural network on the basis of the digital twin model; perform real-time learning on the pollution source monitoring model, and perform real-time monitoring and control using the whole process model of sewage treatment according to the prediction results of the pollution source monitoring model.

[0006] Further, the process of dividing the preset range of the target drinking water source area into several grid sub-areas through GIS means, setting water quality monitoring points in the grid sub-areas, and obtaining water quality monitoring data of each point includes: Obtain GIS geographical data within the preset range of the target drinking water source area through GIS means, obtain the topographic features of the water source area according to the GIS geographical data, identify the water source area type of the water source area according to the topographic features of the water source area, and divide the preset range of the target drinking water source area into several grid sub-areas according to the water source area type; Set water quality monitoring points in each grid sub-area, and obtain the water quality monitoring indicators of each water quality monitoring point by using data retrieval according to the water source area type of the corresponding grid sub-area; The water quality monitoring points obtain water quality data in real time according to the water quality monitoring indicators, mark the monitoring time, and set the monitoring period.

[0007] Further, the process of constructing a digital twin model based on the water quality monitoring results of each water quality monitoring point and scenario information includes: Construct an initial virtual scene model of the digital twin, and obtain the AR scene information of the current geographical location through GIS means according to the geographical locations of the water quality monitoring points within the preset range of the target drinking water source, and process the AR scene information of each water quality monitoring point into a scene subsequence; Obtain the physical entities of each water quality monitoring point in the physical space within the preset range of the target drinking water source, and obtain the water quality data of each water quality monitoring point at each moment within the preset monitoring period; Perform three-dimensional modeling on the physical entities of each water quality monitoring point and map them to the initial virtual scene model of the digital twin. Set API interfaces on the three-dimensional models of each water quality monitoring point, convert the water quality data of each water quality monitoring point at each moment within the preset monitoring period into twin data, input the twin data of each water quality monitoring point at each moment and the corresponding scene subsequence into the initial virtual scene model of the digital twin, and adjust the AR scene information and the three-dimensional models of the water quality monitoring points according to the positional relationship to generate the initial virtual scene model of the digital twin at each moment; Obtain the digital twin model within the current monitoring period by dynamically combining the initial virtual scene models of the digital twin at each moment.

[0008] Furthermore, the process of judging whether to generate a water quality anomaly alarm signal and a water quality anomaly early warning signal according to the water quality monitoring results of each water quality monitoring point includes: Obtain the water quality data monitoring results of each water quality monitoring point within the current monitoring period, obtain the monitoring index thresholds of each water quality monitoring point, and compare the water quality data monitoring results of each water quality monitoring point with the corresponding monitoring index thresholds; If there is a water quality monitoring index in the water quality data monitoring results that is greater than the monitoring index threshold, generate a water quality anomaly alarm signal; If there is no water quality monitoring index in the water quality data monitoring results that is greater than the monitoring index threshold, then obtain the change range of each water quality monitoring index according to the water quality data monitoring results at the current moment and the water quality data monitoring results at the previous moment in the current monitoring period, preset a change range threshold, and compare the change range of each water quality monitoring index with the change range threshold; If there is a water quality monitoring index with a change range greater than the change range threshold, generate a water quality anomaly early warning signal.

[0009] Furthermore, the process of obtaining the correlation coefficient between the water quality anomaly early warning signal generated by each water quality monitoring point and the scene subsequence includes: Obtain the average historical occurrence probability of generating water quality anomaly warning signals at each water quality monitoring point under different scenario subsequences, construct an evaluation criterion for the correlation between water quality anomaly warning signals and scenario subsequences based on the average historical occurrence probability of generating water quality anomaly warning signals at each water quality monitoring point under different scenario subsequences, and construct a membership degree matrix representing the fuzzy relationship between the correlation degree between water quality anomaly warning signals and scenario subsequences and the evaluation criterion through fuzzy comprehensive evaluation; After obtaining the water quality anomaly warning signals generated at each water quality monitoring point under different scenario subsequences, obtain the historical probability of generating water quality anomaly alarm signals within a preset time period. Use the historical probability of generating water quality anomaly alarm signals within a preset time period after generating water quality anomaly warning signals at each water quality monitoring point under different scenario subsequences as the weight coefficient to construct a weight matrix for different scenario subsequences; Obtain the correlation coefficient between the water quality anomaly warning signals generated at each water quality monitoring point and the scenario subsequences according to the membership degree matrix and the weight matrix.

[0010] Furthermore, add a pollution source monitoring model through a neural network based on the digital twin model; the process of real-time learning of the pollution source monitoring model includes: Construct a pollution source monitoring model based on a neural network, and link the pollution source monitoring model through the 3D model API interface of each water quality monitoring point in the digital twin model; Construct a historical data set with the scenario subsequences at each moment and the water quality data monitoring results at each moment in each historical monitoring period of the water quality monitoring points in the digital twin model; Divide the historical data set into a training set and a test set, perform real-time learning and training on the pollution source monitoring model through the training set, and then verify the similarity of the output data matrix of the pollution source monitoring model after training through the test set to obtain a pollution source monitoring model verified by the test set.

[0011] Furthermore, the process of real-time monitoring and control using the whole sewage treatment process model according to the prediction results of the pollution source monitoring model includes: Obtain the time series sequence of the scenario subsequences of the water quality monitoring points and the time series sequence of the water quality data monitoring results in the current monitoring period, and use the time series sequence of the scenario subsequences and the time series sequence of the water quality data monitoring results as the input data set; Input the input data set into the pollution source monitoring model to generate the prediction results of the scenario subsequences and the water quality data monitoring of the water quality monitoring points in the remaining time of the current monitoring period; Judge whether the water quality monitoring points generate water quality anomaly alarm signals in the remaining time of the current monitoring period according to the water quality data monitoring prediction results of the water quality monitoring points; If a water quality anomaly alarm signal is generated at the water quality monitoring point within the remaining time of the current monitoring period, pollution emergency measures are taken for the grid sub-region where the water quality monitoring point is located; If a water quality anomaly alarm signal will not be generated at the water quality monitoring point within the remaining time of the current monitoring period, the occurrence probability of generating a water quality anomaly warning signal at the water quality monitoring point within the remaining time of the current monitoring period is obtained according to the correlation coefficient between the water quality anomaly warning signal generated by the water quality monitoring point and the scenario subsequence and the prediction result of the scenario subsequence of the water quality monitoring point; A preset occurrence probability threshold is set. If the occurrence probability of generating a water quality anomaly warning signal at the water quality monitoring point within the remaining time of the current monitoring period is greater than the occurrence probability threshold, pollution prevention measures are taken for the grid sub-region where the water quality monitoring point is located.

[0012] A pollution source monitoring system based on a neural network includes a monitoring center, and the monitoring center is communicatively connected with a data acquisition module, a model construction module, a data processing module, and a data analysis module; The data acquisition module is used to divide the preset range of the target drinking water source into several grid sub-regions by means of GIS, set water quality monitoring points in the grid sub-regions, and obtain the water quality monitoring data of each point; The model construction module is used to construct a digital twin model according to the water quality monitoring results of each water quality monitoring point and the scenario information; The data processing module is used to judge whether a water quality anomaly alarm signal and a water quality anomaly warning signal are generated according to the water quality monitoring results of each water quality monitoring point, and obtain the correlation coefficient between the water quality anomaly warning signal generated by each water quality monitoring point and the scenario subsequence; The data analysis module is used to add a pollution source monitoring model through a neural network on the basis of the digital twin model; perform real-time learning on the pollution source monitoring model, and perform real-time monitoring and control on the basis of the prediction result of the pollution source monitoring model by using the whole process model of sewage treatment.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention generates the scenario subsequence prediction results and water quality data monitoring prediction results of water quality monitoring points within the remaining time of the current monitoring cycle according to the pollution source monitoring model, and determines whether a water quality anomaly alarm signal is generated at the water quality monitoring points within the remaining time of the current monitoring cycle according to the water quality data monitoring prediction results of the water quality monitoring points. If no water quality anomaly alarm signal is generated at the water quality monitoring points within the remaining time of the current monitoring cycle, the occurrence probability of generating a water quality anomaly warning signal at the water quality monitoring points within the remaining time of the current monitoring cycle is obtained according to the correlation coefficient between the water quality anomaly warning signal and the scenario subsequence generated by the water quality monitoring points and the scenario subsequence prediction results of the water quality monitoring points. The effective warning of sudden water source pollution events such as natural disasters in the drinking water source area is realized, and the temporal dynamic characteristics of pollution information are obtained, ensuring the timeliness of pollution monitoring. Description of the Drawings

[0014] Figure 1 It is a schematic diagram of a pollution source monitoring method based on a neural network according to an embodiment of the present application.

[0015] Figure 2 It is a schematic diagram of a pollution source monitoring system based on a neural network according to an embodiment of the present application. Detailed Embodiments

[0016] Next, in combination with the drawings in the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of the present application.

[0017] As Figure 1 shown, a pollution source monitoring method based on a neural network includes the following steps: Step S1: Divide the preset range of the target drinking water source area into several grid sub-regions by means of GIS, set water quality monitoring points in the grid sub-regions, and obtain the water quality monitoring data of each point. Step S2: Construct a digital twin model according to the water quality monitoring results and scenario information of each water quality monitoring point, determine whether a water quality anomaly alarm signal and a water quality anomaly warning signal are generated according to the water quality monitoring results of each water quality monitoring point, and obtain the correlation coefficient between the water quality anomaly warning signal generated by each water quality monitoring point and the scenario subsequence. Step S3: Add a pollution source monitoring model through a neural network on the basis of the digital twin model; perform real-time learning on the pollution source monitoring model, and perform real-time monitoring and control using the whole process model of sewage treatment according to the prediction results of the pollution source monitoring model.

[0018] It should be further noted that in the specific implementation process, the process of dividing the preset range of the target drinking water source into several grid sub - regions by means of GIS, setting water quality monitoring points in the grid sub - regions, and obtaining water quality monitoring data for each point includes: Obtaining GIS geographical data within the preset range of the target drinking water source by means of GIS, obtaining the topographic features of the water source area based on the GIS geographical data, identifying the type of the water source area according to the topographic features of the water source area, and dividing the preset range of the target drinking water source into several grid sub - regions according to the type of the water source area; Setting water quality monitoring points in each grid sub - region, and obtaining water quality monitoring indicators for each water quality monitoring point by using data retrieval according to the type of the water source area of the corresponding grid sub - region; The water quality monitoring points obtain water quality data in real time according to the water quality monitoring indicators, mark the monitoring time, and set the monitoring period.

[0019] It should be further noted that in the specific implementation process, the process of constructing a digital twin model according to the water quality monitoring results and scenario information of each water quality monitoring point includes: Constructing an initial digital twin virtual scene model, and obtaining AR scene information of the current geographical location by means of GIS according to the geographical location where the water quality monitoring points are located within the preset range of the target drinking water source, and processing the AR scene information of each water quality monitoring point into a scene subsequence; Obtaining the physical entities of each water quality monitoring point in the physical space within the preset range of the target drinking water source, and obtaining the water quality data of each water quality monitoring point at each moment within the preset monitoring period; Performing three - dimensional modeling processing on the physical entities of each water quality monitoring point and mapping them into the initial digital twin virtual scene model, setting API interfaces on the three - dimensional models of each water quality monitoring point, converting the water quality data of each water quality monitoring point at each moment within the preset monitoring period into twin data, inputting the twin data of each water quality monitoring point at each moment and the corresponding scene subsequence into the initial digital twin virtual scene model, and adjusting the AR scene information and the three - dimensional models of the water quality monitoring points according to the position relationship to generate the initial digital twin virtual scene model at each moment; Obtaining the digital twin model within the current monitoring period by dynamically combining the initial digital twin virtual scene models at each moment.

[0020] It should be further noted that in the specific implementation process, the types of the water source area include groundwater sources, surface water sources, decentralized water sources, centralized water sources, etc.; The described processing scenario information includes urban domestic sewage discharge, industrial sewage discharge, agricultural sewage discharge, pollution discharge caused by transportation accidents, debris flows caused by climate mutations, certain minerals in natural geological structures such as copper, chromium, selenium, and lead that may dissolve into groundwater, resulting in natural water body pollution, the rapid growth of naturally occurring microorganisms such as bacteria and algae in lakes, rivers, and other water bodies, forming water blooms (such as algal blooms), releasing toxic substances, polluting water quality, earthquakes or landslides that can cause surface substances to directly enter water source areas, thereby causing pollution, and water body pollution caused by anglers' baiting, etc.

[0021] It should be further noted that in the specific implementation process, the process of judging whether to generate a water quality anomaly alarm signal and a water quality anomaly early warning signal based on the water quality monitoring results of each water quality monitoring point includes: Obtain the water quality data monitoring results within the current monitoring period of each water quality monitoring point, obtain the monitoring index thresholds of each water quality monitoring point, and compare the water quality data monitoring results of each water quality monitoring point with the corresponding monitoring index thresholds; If there is a water quality monitoring index in the water quality data monitoring results that is greater than the monitoring index threshold, generate a water quality anomaly alarm signal; If there is no water quality monitoring index in the water quality data monitoring results that is greater than the monitoring index threshold, then obtain the change range of each water quality monitoring index based on the water quality data monitoring results at the current moment and the water quality data monitoring results at the previous moment in the current monitoring period, preset a change range threshold, and compare the change range of each water quality monitoring index with the change range threshold; If there is a water quality monitoring index with a change range greater than the change range threshold, generate a water quality anomaly early warning signal.

[0022] It should be further noted that in the specific implementation process, the calculation formula for obtaining the change range of each water quality monitoring index based on the water quality data monitoring results at the current moment and the water quality data monitoring results at the previous moment in the current monitoring period is Si = (Pti - Yti) / Yti × 100%, where Si is the change range of the i - type water quality monitoring index; Pti is the water quality data monitoring result of the i - type water quality monitoring index at the current moment; Yti is the water quality data monitoring result of the i - type water quality monitoring index at the previous moment.

[0023] It should be further noted that in the specific implementation process, the process of obtaining the correlation coefficient between the water quality anomaly early warning signal generated by each water quality monitoring point and the scenario subsequence includes: Obtain the average historical occurrence probability of generating water quality anomaly warning signals at each water quality monitoring point under different scenario subsequences. Construct an evaluation criterion for the correlation degree between water quality anomaly warning signals and scenario subsequences based on the average historical occurrence probability of generating water quality anomaly warning signals at each water quality monitoring point under different scenario subsequences, and construct a membership degree matrix representing the fuzzy relationship between the correlation degree between water quality anomaly warning signals and scenario subsequences and the evaluation criterion through fuzzy comprehensive evaluation; After obtaining the water quality anomaly warning signals generated at each water quality monitoring point under different scenario subsequences, obtain the historical probability of generating water quality anomaly alarm signals within a preset time period. Use the historical probability of generating water quality anomaly alarm signals within a preset time period after generating water quality anomaly warning signals at each water quality monitoring point under different scenario subsequences as the weight coefficient to construct a weight matrix for different scenario subsequences; Obtain the correlation coefficients between the water quality anomaly warning signals generated at each water quality monitoring point and the scenario subsequences according to the membership degree matrix and the weight matrix.

[0024] It should be further noted that in the specific implementation process, the process of obtaining the correlation coefficients between the water quality anomaly warning signals generated at each water quality monitoring point and the scenario subsequences according to the membership degree matrix and the weight matrix includes: Fuse the weight matrix and the membership degree matrix to obtain the fuzzy comprehensive evaluation result vector M of the correlation coefficients between the water quality anomaly warning signals generated at each water quality monitoring point and the scenario subsequences Fuse the weight matrix B and the membership degree matrix E through the following formula to obtain the fuzzy comprehensive evaluation matrix Z of each item of data, and obtain the fuzzy comprehensive evaluation result vector M corresponding to each scenario subsequence from the fuzzy comprehensive evaluation matrix Z; Among them, the formula is: ; Among them, " " represents the multiplication of the elements at the corresponding positions of the weight matrix and the membership degree matrix, is a weighted parameter used to control the balance between the weight matrix and the membership degree matrix in the fuzzy comprehensive evaluation matrix of each item of data.

[0025] It should be further noted that in the specific implementation process, add a pollution source monitoring model through a neural network on the basis of the digital twin model; the process of real-time learning of the pollution source monitoring model includes: Construct a pollution source monitoring model based on a neural network, and link the pollution source monitoring model through the 3D model API interface of each water quality monitoring point in the digital twin model; Construct a historical dataset from the scene subsequences and water quality data monitoring results at each moment in each historical monitoring period of the water quality monitoring points in the digital twin model; Divide the historical dataset into a training set and a test set. Use the training set to perform real-time learning and training on the pollution source monitoring model. Then, use the test set to verify the similarity of the output data matrix of the trained pollution source monitoring model to obtain a pollution source monitoring model that passes the test set verification.

[0026] It should be further noted that in the specific implementation process, the process of using the whole sewage treatment process model for real-time monitoring and control according to the prediction results of the pollution source monitoring model includes: Obtain the time series of the scene subsequence and the time series of the water quality data monitoring results of the water quality monitoring points in the current monitoring period, and use the time series of the scene subsequence and the time series of the water quality data monitoring results as the input dataset; Input the input dataset into the pollution source monitoring model to generate the prediction results of the scene subsequence and the water quality data monitoring of the water quality monitoring points in the remaining time of the current monitoring period; Judge whether a water quality anomaly alarm signal is generated at the water quality monitoring point in the remaining time of the current monitoring period according to the prediction result of the water quality data monitoring of the water quality monitoring point; If a water quality anomaly alarm signal is generated at the water quality monitoring point in the remaining time of the current monitoring period, take pollution emergency measures for the grid sub-region where the water quality monitoring point is located; If a water quality anomaly alarm signal is not generated at the water quality monitoring point in the remaining time of the current monitoring period, obtain the occurrence probability of generating a water quality anomaly warning signal at the water quality monitoring point in the remaining time of the current monitoring period according to the correlation coefficient between the generated water quality anomaly warning signal and the scene subsequence of the water quality monitoring point and the prediction result of the scene subsequence of the water quality monitoring point; Preset an occurrence probability threshold. If the occurrence probability of generating a water quality anomaly warning signal at the water quality monitoring point in the remaining time of the current monitoring period is greater than the occurrence probability threshold, take pollution prevention measures for the grid sub-region where the water quality monitoring point is located.

[0027] As Figure 2 shown, a pollution source monitoring system based on a neural network includes a monitoring center, and the monitoring center is communicatively connected to a data acquisition module, a model construction module, a data processing module, and a data analysis module; The data acquisition module is used to divide the preset range of the target drinking water source into several grid sub-regions by means of GIS, set water quality monitoring points in the grid sub-regions, and obtain the water quality monitoring data of each point; The model construction module is used to construct a digital twin model based on the water quality monitoring results and scenario information of each water quality monitoring point; The data processing module is used to judge whether to generate a water quality anomaly alarm signal and a water quality anomaly early warning signal according to the water quality monitoring results of each water quality monitoring point, and obtain the correlation coefficient between the generation of the water quality anomaly early warning signal and the scenario subsequence at each water quality monitoring point; The data analysis module is used to add a pollution source monitoring model through a neural network on the basis of the digital twin model; perform real-time learning on the pollution source monitoring model, and perform real-time monitoring and control using the whole process model of sewage treatment according to the prediction results of the pollution source monitoring model.

[0028] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A pollution source monitoring method based on a neural network, characterized in that, It includes the following steps: Step S1: Divide the preset range of the target drinking water source area into several grid sub-areas by means of GIS, set water quality monitoring points in the grid sub-areas, and obtain water quality monitoring data of each point; Step S2: Construct a digital twin model based on the water quality monitoring results of each water quality monitoring point and the scenario information, judge whether to generate a water quality anomaly alarm signal and a water quality anomaly early warning signal according to the water quality monitoring results of each water quality monitoring point, and obtain the correlation coefficient between the water quality anomaly early warning signal generated by each water quality monitoring point and the scenario subsequence; Step S3: Add a pollution source monitoring model through a neural network on the basis of the digital twin model; perform real-time learning on the pollution source monitoring model, and perform real-time monitoring and control using the whole process model of sewage treatment according to the prediction results of the pollution source monitoring model.

2. The pollution source monitoring method based on a neural network according to claim 1, wherein The process of dividing the preset range of the target drinking water source area into several grid sub-areas by means of GIS, setting water quality monitoring points in the grid sub-areas, and obtaining water quality monitoring data of each point includes: Obtain the GIS geographical data within the preset range of the target drinking water source area by means of GIS, obtain the topographic features of the water source area according to the GIS geographical data, identify the water source type of the water source area according to the topographic features of the water source area, and divide the preset range of the target drinking water source area into grid sub-areas according to the water source type; Set water quality monitoring points in each grid sub-area, and retrieve the water quality monitoring indicators of each water quality monitoring point according to the water source type of the corresponding grid sub-area; The water quality monitoring points obtain water quality data in real time according to the water quality monitoring indicators, mark the monitoring time, and set the monitoring period.

3. The pollution source monitoring method based on a neural network according to claim 2, wherein, The process of constructing a digital twin model based on the water quality monitoring results of each water quality monitoring point and the scenario information includes: Construct a digital twin initial virtual scene model, obtain the AR scene information of the current geographical location by means of GIS according to the geographical location of the water quality monitoring points within the preset range of the target drinking water source area, and process the AR scene information of each water quality monitoring point into a scenario subsequence; Obtain the physical entities of each water quality monitoring point in the physical space within the preset range of the target drinking water source area, and obtain the water quality data of each water quality monitoring point at each moment within the preset monitoring period; Perform three-dimensional modeling processing on the physical entities of each water quality monitoring point and map them to the digital twin initial virtual scene model. Set an API interface on the three-dimensional model of each water quality monitoring point, convert the water quality data of each water quality monitoring point at each moment within the preset monitoring period into twin data, input the twin data of each water quality monitoring point at each moment and the corresponding scenario subsequence into the digital twin initial virtual scene model, and adjust the AR scene information and the three-dimensional model of the water quality monitoring points according to the position relationship to generate the digital twin initial virtual scene model at each moment; Obtain the digital twin model within the current monitoring period by dynamically combining the digital twin initial virtual scene models at each moment.

4. The method for monitoring pollution sources based on a neural network according to claim 3, wherein The process of determining whether to generate a water quality anomaly alarm signal and a water quality anomaly early warning signal based on the water quality monitoring results of each water quality monitoring point includes: Obtain the water quality data monitoring results within the current monitoring period of each water quality monitoring point, obtain the monitoring index thresholds of each water quality monitoring point, and compare the water quality data monitoring results of each water quality monitoring point with the corresponding monitoring index thresholds; If there is a water quality monitoring index in the water quality data monitoring results that is greater than the monitoring index threshold, generate a water quality anomaly alarm signal; If there is no water quality monitoring index in the water quality data monitoring results that is greater than the monitoring index threshold, then obtain the change range of each water quality monitoring index based on the water quality data monitoring results at the current moment and the water quality data monitoring results at the previous moment in the current monitoring period, preset a change range threshold, and compare the change range of each water quality monitoring index with the change range threshold; If there is a water quality monitoring index with a change range greater than the change range threshold, generate a water quality anomaly early warning signal.

5. The method for monitoring pollution sources based on a neural network according to claim 4, wherein, The process of obtaining the correlation coefficient between the water quality anomaly early warning signal generated by each water quality monitoring point and the scenario subsequence includes: Obtain the average historical occurrence probability of generating a water quality anomaly early warning signal by each water quality monitoring point under different scenario subsequences, construct an evaluation criterion for the correlation between the water quality anomaly early warning signal and the scenario subsequence based on the average historical occurrence probability of generating a water quality anomaly early warning signal by each water quality monitoring point under different scenario subsequences, and construct a membership matrix representing the fuzzy relationship between the correlation degree between the water quality anomaly early warning signal and the scenario subsequence and the evaluation criterion through fuzzy comprehensive evaluation; After obtaining the water quality anomaly early warning signal generated by each water quality monitoring point under different scenario subsequences, obtain the historical probability of generating a water quality anomaly alarm signal within a preset time period, and use the historical probability of generating a water quality anomaly alarm signal within a preset time period after generating the water quality anomaly early warning signal by each water quality monitoring point under different scenario subsequences as the weight coefficient to construct a weight matrix for different scenario subsequences; Obtain the correlation coefficient between the water quality anomaly early warning signal generated by each water quality monitoring point and the scenario subsequence according to the membership matrix and the weight matrix.

6. The method for monitoring pollution sources based on a neural network according to claim 5, characterized in that, Adding a pollution source monitoring model through a neural network based on the digital twin model; the process of real-time learning of the pollution source monitoring model includes: Construct a pollution source monitoring model based on a neural network, and link the pollution source monitoring model through the 3D model API interface of each water quality monitoring point in the digital twin model; Construct a historical data set from the scenario subsequence and the water quality data monitoring results at each moment in each historical monitoring period of the water quality monitoring points in the digital twin model; Divide the historical data set into a training set and a test set, perform real-time learning and training on the pollution source monitoring model through the training set, and then verify the similarity of the output data matrix of the pollution source monitoring model after training through the test set to obtain a pollution source monitoring model verified by the test set.

7. The method for monitoring pollution sources based on a neural network according to claim 6, characterized in that, The process of real-time monitoring and control using the whole sewage treatment process model according to the prediction results of the pollution source monitoring model includes: Obtain the time series of the scenario subsequences of the water quality monitoring points and the time series of the water quality data monitoring results within the current monitoring period, and use the time series of the scenario subsequences and the time series of the water quality data monitoring results as the input data set; Input the input data set into the pollution source monitoring model to generate the prediction results of the scenario subsequences of the water quality monitoring points and the prediction results of the water quality data monitoring during the remaining time of the current monitoring period; Judge whether a water quality anomaly alarm signal is generated at the water quality monitoring point during the remaining time of the current monitoring period according to the prediction result of the water quality data monitoring at the water quality monitoring point; If a water quality anomaly alarm signal is generated at the water quality monitoring point during the remaining time of the current monitoring period, take pollution emergency measures for the grid sub-region where the water quality monitoring point is located; If a water quality anomaly alarm signal is not generated at the water quality monitoring point during the remaining time of the current monitoring period, obtain the occurrence probability of generating a water quality anomaly warning signal at the water quality monitoring point during the remaining time of the current monitoring period according to the correlation coefficient between the water quality anomaly warning signal generated at the water quality monitoring point and the scenario subsequence and the prediction result of the scenario subsequence of the water quality monitoring point; Preset an occurrence probability threshold. If the occurrence probability of generating a water quality anomaly warning signal at the water quality monitoring point during the remaining time of the current monitoring period is greater than the occurrence probability threshold, take pollution prevention measures for the grid sub-region where the water quality monitoring point is located.

8. A pollution source monitoring system based on a neural network, specifically applied to a pollution source monitoring method based on a neural network according to any one of claims 1 to 7, including a monitoring center, characterized in that, The monitoring center is communicatively connected to a data acquisition module, a model construction module, a data processing module, and a data analysis module; The data acquisition module is used to divide the preset range of the target drinking water source into several grid sub-regions by means of GIS, set water quality monitoring points in the grid sub-regions, and obtain the water quality monitoring data of each point; The model construction module is used to construct a digital twin model according to the water quality monitoring results and scenario information of each water quality monitoring point; The data processing module is used to judge whether a water quality anomaly alarm signal and a water quality anomaly warning signal are generated according to the water quality monitoring results of each water quality monitoring point, and obtain the correlation coefficient between the water quality anomaly warning signal generated at each water quality monitoring point and the scenario subsequence; The data analysis module is used to add a pollution source monitoring model through a neural network on the basis of the digital twin model; perform real-time learning on the pollution source monitoring model, and perform real-time monitoring and control using the whole sewage treatment process model according to the prediction results of the pollution source monitoring model.

Images