Marine ranch pollution early warning method and system

Abstract

The invention discloses a marine ranch pollution early warning method and system, and relates to the technical field of pollution early warning. The invention divides the marine ranch and remote sensing images into blocks according to the position of the sewage outlet in the marine ranch, and divides the blocks according to the sewage outlet position, so as to deepen the correlation between each partition and the actual sewage discharge situation of the marine ranch, and extract and identify pixel points according to the remote sensing image, so as to accurately identify the pollutants and pollution values ​​of each partition; by mining the relationship between the historical sewage discharge data of the marine ranch and the pollution value, the sewage discharge pollution value of the marine ranch can be obtained according to the sewage discharge plan of the marine ranch, so that the consideration of the future pollution change of the marine ranch is added to the pollution early warning of the marine ranch, and through the prediction of the current pollution value and the future pollution value, the marine ranch pollution degree with higher accuracy can be obtained comprehensively, so as to better early-warn the marine ranch pollution and improve the working environment of the marine ranch.

Description

Technical Field

[0001] The present invention belongs to the technical field of pollution early warning, and in particular relates to a marine ranch pollution early warning method and system. Background Art

[0002] As a rich and diverse natural resource, marine resources are gradually becoming popular along with the development and utilization of marine resources. Marine ranches, a new type of marine resource development platform, are also becoming more and more popular. However, with the continuous popularization and expansion of marine ranches, the problem of marine ranch pollution has gradually become prominent, becoming an important factor affecting the quality of the marine environment and the sustainable development of marine ranches. The pollution problem will not only reduce the production efficiency and sustainable development of marine ranches, but also aggravate the degradation of the marine environment, thus creating a vicious cycle. Therefore, it is very necessary to warn of marine ranch pollution, so as to provide a better pre-environment for solving the problem of marine ranch pollution.

[0003] At present, the early warning of marine ranch pollution is usually carried out by installing multiple sensors in the marine ranch to monitor the water information in the marine ranch through the sensors to monitor the degree of water pollution. However, this method has a high demand for equipment and cannot be well adapted to marine ranches in different environments. However, due to the complexity and uncertainty of the marine environment, it is impossible to accurately monitor pollution by relying solely on sensor monitoring. Therefore, a new marine ranch pollution early warning method and system is urgently needed to solve the defects of the existing technology. Summary of the invention

[0004] The present invention aims to provide a marine ranch pollution early warning method and system to solve the above-mentioned technical problems, divide the marine ranch into zones to calculate pollution values ​​and sewage pollution values, achieve comprehensive acquisition of marine ranch pollution, and then complete pollution early warning for the marine ranch.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a marine ranch pollution early warning method, comprising:

[0006] Acquire a remote sensing image and a sewage outlet location of the marine ranch; cut the remote sensing image according to the sewage outlet location to obtain a plurality of first subareas and their corresponding first subarea images and first sewage outlets;

[0007] According to the pixel points of each first subarea image, combined with a preset pollutant database, each first subarea image is identified to obtain the pollutants and pollution values ​​of each first subarea;

[0008] Obtaining historical sewage discharge data of each first sewage outlet, extracting pollutants from the historical sewage discharge data, and iteratively training a preset algorithm model in combination with pollutants and pollution values ​​of each first partition to obtain an optimal sewage discharge model;

[0009] According to the sewage discharge plan of each first sewage outlet, combined with the optimal sewage discharge model, the sewage discharge pollution value of each first partition is obtained;

[0010] The central position of the ocean ranch is obtained, and the current pollution level of the ocean ranch is obtained by combining the sewage pollution value and the pollution value of each first subarea; and according to a preset pollution level threshold and the current pollution level, an early warning is issued for pollution of the ocean ranch.

[0011] It can be understood that, compared with the prior art, the present invention divides the marine ranch and remote sensing images into blocks according to the location of the sewage outlet in the marine ranch. The partitioning by the location of the sewage outlet can deepen the correlation between each partition and the actual sewage discharge situation of the marine ranch, and extract and identify pixel points based on the remote sensing image, so as to accurately identify the pollutants and pollution values ​​of each partition; by mining the relationship between the historical sewage discharge data of the marine ranch and the pollution value, the sewage discharge pollution value of the marine ranch can be obtained according to the sewage discharge plan of the marine ranch, so that the future pollution changes of the marine ranch are taken into consideration in the pollution warning of the marine ranch. Through the prediction of the current pollution value and the future pollution value, the marine ranch pollution degree with higher accuracy can be comprehensively obtained, so as to better warn of the marine ranch pollution, reduce the environmental pollution of the marine ranch, and maintain the normal work of the marine ranch.

[0012] As a preferred solution, the remote sensing image is cut according to the position of the sewage outlet to obtain a plurality of first subareas and their corresponding first subarea images and first sewage outlets, specifically including:

[0013] Taking the position of each sewage outlet as the geometric center point of the partition, the marine ranch is cut to obtain a plurality of first partitions and a first sewage outlet corresponding to each first partition;

[0014] The remote sensing image is cut according to the first partitions to obtain a first partition image corresponding to each first partition.

[0015] This preferred solution cuts the marine ranch and remote sensing images by the location of the sewage outlet, so that there is a sewage outlet in each partition, making full use of the sewage discharge characteristics of different sewage outlets in the marine ranch, thereby improving the consistency of pollutants in each partition, and then being able to more accurately obtain pollutants and pollution values ​​in the marine ranch based on the operation of remote sensing images.

[0016] As a preferred solution, the identification of each first subarea image based on the pixel points of each first subarea image and the preset pollutant database to obtain the pollutant data and pollution value of each first subarea specifically includes:

[0017] Performing denoising on the first subarea image to obtain a second subarea image of each first subarea;

[0018] Reconstructing the second subarea images to obtain a stereo remote sensing image of each first subarea;

[0019] Performing image segmentation in a vertical direction on the stereo remote sensing image to obtain a plurality of layered images of each first subarea;

[0020] In combination with a preset pollutant database, pixel extraction and identification are performed on each of the layered images to obtain pollutant data and pollution values ​​in each layered image, and further obtain pollutant data and pollution values ​​of each first partition.

[0021] This preferred solution can filter out abnormal image information by denoising the image, thereby avoiding interference of abnormal information on pollutant identification; by reconstructing the image to establish a three-dimensional remote sensing image, and performing image cutting to obtain a layered image, as well as through the pixel point, which is the basic structure of the remote sensing image, it can not only obtain information on pollutants in different spatial dimensions in the partition, but also fully capture each image information, thereby improving the calculation accuracy of the pollution degree of the marine ranch.

[0022] As a preferred solution, the preset pollutant database is combined to extract and identify pixels of each layered image, obtain pollutant data and pollution value in each layered image, and then obtain pollutants and pollution values ​​of each first partition, specifically including:

[0023] Extracting pixels from each of the layered images respectively to obtain a pixel data set of each layered image; the pixel data set includes: the color of the pixel, the brightness of the pixel and the size of the pixel;

[0024] Constructing a key-value pair for each pixel according to the color and brightness of the pixel;

[0025] Iteratively clustering the pixels in each layered image according to the key-value pairs to obtain a clustering result for each layered image, wherein the clustering result includes: a plurality of cluster centers;

[0026] The preset pollutant database includes: images of pollutants and pollution risk values;

[0027] Extract and identify pixels of pollutant images to obtain pixel color and pixel brightness of each pollutant in the pollutant database;

[0028] According to the pixel color and pixel brightness of the cluster center, the pixel color and pixel brightness of each pollutant in the pollutant database are matched to obtain the pollutant corresponding to each cluster center, and then the pollutant of each layered image is obtained;

[0029] According to the pollutants of each layered image, the pollutants of each first partition are obtained;

[0030] According to the pollution risk value of the pollutant and the size of the pixel point, in combination with the clustering result, the pollution value of each layered image is calculated, and then the pollution value of each first partition is obtained.

[0031] This preferred solution clusters the basic information of pixels in the remote sensing image, and based on the image pixel features of each pollutant in the pollutant database, can accurately and comprehensively obtain the types of pollutants in the partition; and through the risk value of each pollutant, the accuracy of the obtained partition pollution value is improved.

[0032] As a preferred solution, the pollution risk value of each layered image is calculated based on the pollution risk value of the pollutant and the size of the pixel point, combined with the clustering result, to obtain the pollution value of each first partition, specifically including:

[0033] The clustering result includes: the cluster corresponding to the cluster center and the number of pixels in each cluster;

[0034] According to the size of the pixels and the number of pixels in each cluster, the area of ​​each cluster is calculated;

[0035] According to the area of ​​each cluster and the pollution risk value of the pollutant, the pollution value of each pollutant in the layered image is calculated;

[0036] According to the pollution value of each pollutant, the pollution value of each layered image is obtained, and then the pollution value of each first partition is obtained.

[0037] This preferred solution calculates the area of ​​each cluster by the number and size of pixel points, thereby obtaining the area without pollutants, and finally obtaining the pollution value in each layered image, and finally obtaining the pollution value of the partition; calculating the pollution value by the basic unit pixel points in the image can avoid omissions in the calculation of pollutants, thereby improving the calculation accuracy of the partition pollution value.

[0038] As a preferred solution, the historical sewage discharge data of each first sewage outlet is obtained, pollutants are extracted from the historical sewage discharge data, and the preset algorithm model is iteratively trained in combination with the pollutants and pollution values ​​of each first partition to obtain the optimal sewage discharge model, which specifically includes:

[0039] Acquire historical sewage discharge data of each first sewage discharge outlet, perform pollutant extraction on the historical sewage discharge data, and obtain historical pollutant data of each first sewage discharge outlet;

[0040] Constructing an objective function of the algorithm model according to the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition;

[0041] Decomposing the historical pollutant data according to a variational mode decomposition algorithm to obtain a plurality of pollutant mode functions;

[0042] The pollutant modal function, the pollutants and pollution values ​​of the first partition are input into the algorithm model, and combined with the objective function, the algorithm model is iteratively trained until the value of the objective function meets the preset threshold requirement, the training of the algorithm model is completed, and the optimal pollution discharge model is obtained.

[0043] This preferred solution trains the algorithm model through historical pollution discharge data and current marine ranch pollution data, which can fully explore the relationship between marine ranch pollution discharge and pollution, and thus predict the future pollution value of the marine ranch based on the trained optimal pollution discharge model, so that accurate early warning of marine ranch pollution can be made based on both current and future perspectives.

[0044] As a preferred solution, the objective function and pollution discharge constraints of the algorithm model are constructed based on the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition, specifically including:

[0045] According to the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition, the objective function of the algorithm model is constructed, wherein the objective function is:

[0046] ,

[0047] in, Indicates the type of pollutant, represents the set of pollutant types, The pollution factor representing the pollutant can be obtained by querying the pollutant database; Indicates the proportion of the pollutant; Indicates the total amount of sewage discharged from the first sewage outlet. Indicates the first sewage outlet The discharge time of each pollutant; Represents the pollution value of the first partition.

[0048] This preferred solution fully explores the relationship between marine ranch discharge and pollution by constructing the objective function of the algorithm model, so that the future pollution value of the marine ranch can be predicted based on the trained optimal discharge model, so that accurate early warning of marine ranch pollution can be made based on both current and future perspectives.

[0049] As a preferred solution, the central position of the ocean ranch is obtained, and the current pollution level of the ocean ranch is obtained by combining the pollution level prediction value, the sewage pollution value and the pollution value of each first subarea; and according to a preset pollution level threshold, combined with the current pollution level, an early warning of pollution of the ocean ranch is issued, specifically including:

[0050] Obtaining the central position of the ocean ranch, and calculating the sewage discharge proportion of each first partition according to the central position and the position of the first sewage outlet;

[0051] Calculate the current pollution degree of the marine ranch according to the pollution discharge proportion, pollution value and pollution discharge pollution value of each of the first sub-areas;

[0052] When the current pollution level exceeds a preset pollution level threshold, an early warning of marine ranch pollution is issued so that the marine ranch staff can adjust the sewage discharge plan of each first sewage outlet according to the current pollution level.

[0053] This preferred solution obtains the discharge proportion through the location of each sewage outlet, and then realizes marine ranch pollution warning based on the combination of the present and the future according to the current pollution value and the predicted pollution value, thereby improving the accuracy of marine ranch pollution warning from two pollution perspectives.

[0054] Accordingly, an embodiment of the present invention provides a marine ranch pollution early warning system, including: a partition module, a pollution value acquisition module, a pollution model acquisition module, a pollution pollution value acquisition module and a pollution early warning module;

[0055] The partition module is used to obtain the remote sensing image and the location of the sewage outlet of the marine ranch; according to the location of the sewage outlet, the remote sensing image is cut to obtain a plurality of first partitions and their corresponding first partition images and first sewage outlets;

[0056] The pollution value acquisition module is used to identify each first subarea image according to the pixel points of each first subarea image in combination with a preset pollutant database to obtain the pollutants and pollution values ​​of each first subarea;

[0057] The sewage discharge model acquisition module is used to obtain the historical sewage discharge data of each first sewage discharge outlet, extract pollutants from the historical sewage discharge data, and iteratively train the preset algorithm model in combination with the pollutants and pollution values ​​of each first partition to obtain the optimal sewage discharge model;

[0058] The sewage pollution value acquisition module is used to obtain the sewage pollution value of each first partition according to the sewage discharge plan of each first sewage outlet in combination with the optimal sewage discharge model;

[0059] The pollution warning module is used to obtain the central position of the ocean ranch, and combine the pollution prediction value, sewage pollution value and pollution value of each first partition to obtain the current pollution level of the ocean ranch; according to the preset pollution level threshold, combined with the current pollution level, an early warning is issued for ocean ranch pollution.

[0060] As a preferred solution, the pollution value acquisition module includes: a pollution value acquisition unit;

[0061] The pollution value acquisition unit is used to perform denoising on the first subarea image to obtain a second subarea image of each first subarea;

[0062] Reconstructing the second subarea images to obtain a stereo remote sensing image of each first subarea;

[0063] Performing image segmentation in a vertical direction on the stereo remote sensing image to obtain a plurality of layered images of each first subarea;

[0064] In combination with a preset pollutant database, pixel extraction and identification are performed on each of the layered images to obtain pollutant data and pollution values ​​in each layered image, and further obtain pollutant data and pollution values ​​of each first partition.

[0065] It can be understood that, compared with the existing technology, this system divides the marine ranch and remote sensing images into blocks according to the location of the sewage outlet in the marine ranch. The partitioning based on the location of the sewage outlet can deepen the correlation between each partition and the actual sewage discharge situation of the marine ranch, and extract and identify pixels based on the remote sensing image, so as to accurately identify the pollutants and pollution values ​​of each partition; by mining the relationship between the historical sewage discharge data of the marine ranch and the pollution value, the sewage discharge pollution value of the marine ranch can be obtained according to the sewage discharge plan of the marine ranch, so that the future pollution changes of the marine ranch are taken into consideration in the pollution warning of the marine ranch. Through the prediction of the current pollution value and the future pollution value, the marine ranch pollution degree with higher accuracy can be obtained comprehensively, so as to better warn of the marine ranch pollution, reduce the environmental pollution of the marine ranch, and maintain the normal work of the marine ranch. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 : A flow chart of the steps of a marine ranch pollution early warning method provided by an embodiment of the present invention;

[0067] Figure 2 : A structural diagram of a marine ranch pollution early warning system provided by an embodiment of the present invention;

[0068] Among them, 201: partition module; 202: pollution value acquisition module; 203: pollution model acquisition module; 204: pollution value acquisition module; 205: pollution warning module. DETAILED DESCRIPTION

[0069] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] The quality of the marine environment, as the basic component of the development and utilization of marine resources, has a profound impact on the efficiency of marine resource development. Among them, marine ranches, as an emerging marine resource development platform, combine the advantages of marine aquaculture and land ranches, and can achieve efficient development of marine resources. Therefore, the scale of marine ranches is increasing, which also brings about the problem of marine ranch pollution.

[0071] Since most marine ranches are engaged in aquaculture, although marine ranches carry out filtration and disinfection when discharging pollutants, they still cannot completely prevent pollutants from being discharged into the ocean. Over time, this has had a bad impact on the marine environment and polluted the marine environment. The pollution of the marine environment will affect the efficient output of marine ranches. For example, during the aquaculture process, marine ranches will discharge wastewater containing high nutrients. After these nutrients enter the ocean, they will promote the over-reproduction of phytoplankton such as algae, thereby causing eutrophication of water bodies and deterioration of water quality. In addition, pesticides and antibiotics used on aquaculture organisms during the aquaculture process may also enter the ocean, thereby causing damage to the marine environment.

[0072] At present, the early warning of marine ranches usually focuses on using sensors to regularly detect the quality of marine water bodies, and judging the current pollution level by analyzing the water content, so as to further improve the marine ranch's sewage discharge plan. However, this method is only limited to the current marine ranch pollution level, and does not deeply explore the relationship between marine ranch sewage discharge and water pollution, and cannot further accurately improve the pollution of marine ranches. Moreover, water analysis through sensors has higher requirements for the installation and number of sensors, and its implementation is difficult. Therefore, there is an urgent need for a simple, accurate and complete method for early warning of marine ranch pollution to solve the defects of the existing technology.

[0073] Embodiment 1

[0074] Please refer to Figure 1 , is a flow chart of the steps of a marine ranch pollution early warning method provided by an embodiment of the present invention, including steps S101-S105:

[0075] Step S101: obtaining a remote sensing image of the ocean ranch and a sewage outlet location; cutting the remote sensing image according to the sewage outlet location to obtain a plurality of first partitions and their corresponding first partition images and first sewage outlets.

[0076] In this embodiment, the remote sensing image is cut according to the position of the sewage outlet to obtain a plurality of first subareas and their corresponding first subarea images and first sewage outlets, specifically including:

[0077] Taking the position of each sewage outlet as the geometric center point of the partition, the marine ranch is cut to obtain a plurality of first partitions and a first sewage outlet corresponding to each first partition;

[0078] The remote sensing image is cut according to the first partitions to obtain a first partition image corresponding to each first partition.

[0079] The marine ranch and remote sensing images are cut according to the location of the sewage outlet, so that there is a sewage outlet in each partition, which makes full use of the sewage discharge characteristics of different sewage outlets in the marine ranch, thereby improving the consistency of pollutants in each partition, and then more accurately obtaining pollutants and pollution values ​​in the marine ranch based on the operation of remote sensing images.

[0080] It should be noted that the position of each sewage outlet is used as the geometric center point of the partition to cut the marine ranch. The size and shape of each first partition obtained by cutting should be kept consistent, and there should be only one sewage outlet in each first partition; thereby, further in-depth step-by-step judgment on water pollution can be made based on the sewage discharge from the marine ranch.

[0081] Step S102: identifying each first subarea image according to the pixel points of each first subarea image in combination with a preset pollutant database to obtain the pollutants and pollution values ​​of each first subarea.

[0082] In this embodiment, the identification of each first subarea image based on the pixel points of each first subarea image and the preset pollutant database to obtain the pollutant data and pollution value of each first subarea specifically includes:

[0083] Performing denoising on the first subarea image to obtain a second subarea image of each first subarea;

[0084] Reconstructing the second subarea images to obtain a stereo remote sensing image of each first subarea;

[0085] Performing image segmentation in a vertical direction on the stereo remote sensing image to obtain a plurality of layered images of each first subarea;

[0086] In combination with a preset pollutant database, pixel points of each layered image are extracted and identified respectively to obtain pollutant data and pollution values ​​in each layered image, and further obtain pollutant data and pollution values ​​of each first partition.

[0087] By denoising the image, abnormal image information can be filtered out, avoiding the interference of abnormal information on pollutant identification; by reconstructing the image to establish a three-dimensional remote sensing image, and performing image cutting to obtain a layered image, as well as through the pixel point, which is the basic structure of the remote sensing image, not only can the information of pollutants in different spatial dimensions in the partition be obtained, but also each image information can be fully captured, thereby improving the calculation accuracy of the pollution degree of the marine ranch.

[0088] In an optional embodiment, reconstructing the second subarea image to obtain a stereoscopic remote sensing image of each first subarea specifically includes:

[0089] Extract key features from each second subarea image of each first subarea, and extract features such as terrain features and shape and color of the sewage outlet;

[0090] Based on the extracted key features, three-dimensional software is used to perform modeling to obtain a three-dimensional remote sensing image of each first subarea;

[0091] The three-dimensional remote sensing image is texture mapped and rendered to obtain a stereoscopic remote sensing image of each first subarea.

[0092] This optional embodiment takes into account the distribution characteristics of pollutants in the marine ranch. Since pollutants are distributed at different heights in the same area, the two-dimensional second partition image is three-dimensionalized to obtain a three-dimensional stereo remote sensing image. By extracting key features, a partitioned stereo image can be obtained while retaining the original image features. This can obtain more complete marine environmental information and improve the accuracy of marine ranch pollution warning.

[0093] In this embodiment, the preset pollutant database is combined to extract and identify pixels of each layered image, obtain pollutant data and pollution value in each layered image, and then obtain pollutants and pollution values ​​of each first partition, specifically including:

[0094] Extracting pixels from each of the layered images respectively to obtain a pixel data set of each layered image; the pixel data set includes: the color of the pixel, the brightness of the pixel and the size of the pixel;

[0095] Constructing a key-value pair for each pixel according to the color and brightness of the pixel;

[0096] Iteratively clustering the pixels in each layered image according to the key-value pairs to obtain a clustering result for each layered image, wherein the clustering result includes: a plurality of cluster centers;

[0097] The preset pollutant database includes: images of pollutants and pollution risk values;

[0098] Extract and identify pixels of pollutant images to obtain pixel color and pixel brightness of each pollutant in the pollutant database;

[0099] According to the pixel color and pixel brightness of the cluster center, the pixel color and pixel brightness of each pollutant in the pollutant database are matched to obtain the pollutant corresponding to each cluster center, and then the pollutant of each layered image is obtained;

[0100] According to the pollutants of each layered image, the pollutants of each first partition are obtained;

[0101] According to the pollution risk value of the pollutant and the size of the pixel point, in combination with the clustering result, the pollution value of each layered image is calculated, and then the pollution value of each first partition is obtained.

[0102] By clustering the basic information of pixels in remote sensing images and based on the image pixel features of each pollutant in the pollutant database, the types of pollutants in the partition can be accurately and comprehensively obtained; and through the risk value of each pollutant, the accuracy of the obtained partition pollution value is improved.

[0103] In an optional embodiment, the size of the pixel point can be determined according to the acquired remote sensing image, and the key-value pair of the pixel point is in the form of <color, brightness>.

[0104] In an optional embodiment, the pixel points in each layered image are iteratively clustered according to the key-value pairs to obtain a clustering result for each layered image, wherein the clustering result includes: a plurality of cluster centers, specifically including:

[0105] According to the key-value pair of each pixel, the pixels in the layered image are randomly divided into k groups, and k key-value pairs are randomly selected as the initial clustering centers. The distance between each key-value pair and each initial clustering center is calculated. According to the distance calculation result, each key-value pair is assigned to the initial clustering center with the smallest distance to form new k clusters, completing a clustering calculation.

[0106] According to the clustering calculation, the clustering center is continuously iterated until the clustering center no longer changes, and the clustering iteration is completed to obtain the clustering result of each layered image, and the clustering result includes: a plurality of clustering centers.

[0107] It should be noted that the calculation of the distance between each key-value pair and each initial cluster center described in the embodiment of the present invention is based on the similarity of color and brightness in the key-value pair; the clustering of key-value pairs described in the present invention can use clustering algorithms such as K-means clustering algorithm, DBSCAN algorithm or hierarchical clustering algorithm (Hierarchical Clustering) to calculate the distance between pixels.

[0108] In an optional embodiment, the pixel color and pixel brightness of the cluster center are matched with the pixel color and pixel brightness of each pollutant in the pollutant database to obtain the pollutant corresponding to each cluster center, and then obtain the pollutant of each layered image, specifically including:

[0109] Get the pixel color and pixel brightness corresponding to the cluster center;

[0110] Match the pixel color of the cluster center with the pixel color of each pollutant in the pollution database, calculate the color similarity, and select several candidate pollutants according to the preset color similarity threshold;

[0111] Match the brightness of the pixels at the cluster center with the brightness of the pixels of the candidate pollutants, calculate the brightness similarity, select the candidate pollutant with the highest brightness similarity as the matching pollutant, and obtain the pollutant corresponding to each cluster center;

[0112] The pollutants corresponding to each cluster center are counted to obtain the pollutants of the layered image.

[0113] In this embodiment, the pollution risk value of each layered image is calculated based on the pollution risk value of the pollutant and the size of the pixel point, combined with the clustering result, to obtain the pollution value of each first partition, specifically including:

[0114] The clustering result includes: the cluster corresponding to the cluster center and the number of pixels in each cluster;

[0115] According to the size of the pixels and the number of pixels in each cluster, the area of ​​each cluster is calculated;

[0116] According to the area of ​​each cluster and the pollution risk value of the pollutant, the pollution value of each pollutant in the layered image is calculated;

[0117] According to the pollution value of each pollutant, the pollution value of each layered image is obtained, and then the pollution value of each first partition is obtained.

[0118] The area of ​​each cluster is calculated by the number and size of pixels, so as to obtain the area without pollution, and finally the pollution value in each layered image is obtained, and finally the pollution value of the partition is obtained; calculating the pollution value by the basic unit pixel points in the image can avoid omission of pollutants, thereby improving the calculation accuracy of the partition pollution value.

[0119] In an optional embodiment, the calculation of the pollution value of each layered image based on the pollution risk value of the pollutant and the size of the pixel point in combination with the clustering result, and then obtaining the pollution value of each first partition, specifically includes:

[0120] The clustering result includes: the cluster corresponding to the cluster center and the number of pixels in each cluster;

[0121] Based on the size of the pixels and the number of pixels in each cluster, the area of ​​each cluster is calculated:

[0122] ,

[0123] in, For the The area of ​​the cluster, i.e. The area of ​​pollutants, For each The number of pixels in the cluster, is the size of the pixel;

[0124] According to the area of ​​each cluster and the pollution risk value of the pollutant, the pollution value of each pollutant in the layered image is calculated:

[0125] ,

[0126] in, For the The pollution value of each pollutant, For the The pollution risk value of each pollutant is For the The area of ​​the cluster, i.e. The area of ​​each pollutant;

[0127] According to the pollution value of each pollutant, the pollution value of each layered image is obtained, and then the pollution value of each first partition is obtained.

[0128] It should be noted that obtaining the pollution value of each layered image according to the pollution value of each pollutant, and then obtaining the pollution value of each first partition, specifically includes:

[0129] Adding the pollution values ​​of all pollutants in each layered image to obtain the pollution value of each layered image;

[0130] The pollution values ​​of all layered images in each first partition are added together to obtain the pollution value of each first partition.

[0131] Step S103: Obtain historical sewage discharge data of each first sewage outlet, extract pollutants from the historical sewage discharge data, and iteratively train the preset algorithm model in combination with the pollutants and pollution values ​​of each first partition to obtain the optimal sewage discharge model.

[0132] In this embodiment, the historical sewage discharge data of each first sewage outlet is obtained, pollutants are extracted from the historical sewage discharge data, and the preset algorithm model is iteratively trained in combination with the pollutants and pollution values ​​of each first partition to obtain the optimal sewage discharge model, which specifically includes:

[0133] Acquire historical sewage discharge data of each first sewage discharge outlet, perform pollutant extraction on the historical sewage discharge data, and obtain historical pollutant data of each first sewage discharge outlet;

[0134] Constructing an objective function of the algorithm model according to the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition;

[0135] Decomposing the historical pollutant data according to a variational mode decomposition algorithm to obtain a plurality of pollutant mode functions;

[0136] The pollutant modal function, the pollutants and pollution values ​​of the first partition are input into the algorithm model, and combined with the objective function, the algorithm model is iteratively trained until the value of the objective function meets the preset threshold requirement, the training of the algorithm model is completed, and the optimal pollution discharge model is obtained.

[0137] By training the algorithm model with historical pollution discharge data and current marine ranch pollution data, we can fully explore the relationship between marine ranch pollution discharge and pollution, and thus predict the future pollution values ​​of the marine ranch based on the trained optimal pollution discharge model, so that we can accurately warn of marine ranch pollution from both current and future perspectives.

[0138] In this embodiment, the objective function and pollution discharge constraints of the algorithm model are constructed according to the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition, specifically including:

[0139] According to the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition, the objective function of the algorithm model is constructed, wherein the objective function is:

[0140] ,

[0141] in, Indicates the type of pollutant, represents the set of pollutant types, The pollution factor representing the pollutant can be obtained by querying the pollutant database; Indicates the proportion of the pollutant; Indicates the total amount of sewage discharged from the first sewage outlet. Indicates the first sewage outlet The discharge time of each pollutant; Represents the pollution value of the first partition.

[0142] By constructing the objective function of the algorithm model, the relationship between marine ranch discharge and pollution can be fully explored, so that the future pollution value of the marine ranch can be predicted based on the trained optimal discharge model, making it possible to accurately warn of marine ranch pollution from both current and future perspectives.

[0143] In an optional embodiment, the preset algorithm model is specifically a CNN-LSTM model, that is, a deep learning model composed of a convolutional neural network (CNN) and a long short-term memory network (LSTM); the CNN model includes three convolution layers and two pooling layers; wherein the first one-dimensional convolution layer is connected to the first maximum pooling layer for dimensionality reduction sampling of the input feature map; the first maximum pooling layer, the second one-dimensional convolution layer and the second maximum pooling layer are connected in sequence for keeping the size of the feature map unchanged; the third one-dimensional convolution layer is connected to the second maximum pooling layer for weighted summation of the feature map and then outputting it to the LSTM model; the LSTM model includes a forget gate, an input gate and an output gate for capturing long-term dependencies in time series data; the CNN-LSTM model also includes a fully connected layer, which is used to reconstruct the features output by the LSTM model to obtain a mapping relationship between the historical discharge data of the ocean ranch sewage outlet and the pollution values ​​in the corresponding partitions.

[0144] Step S104: according to the sewage discharge plan of each first sewage outlet, in combination with the optimal sewage discharge model, the sewage discharge pollution value of each first partition is obtained.

[0145] In an optional embodiment, the sewage discharge plan of each first sewage outlet is combined with the optimal sewage discharge model to obtain the sewage discharge pollution value of each first partition, which specifically includes:

[0146] Perform variational mode decomposition on the sewage discharge plan of each first sewage outlet to obtain several mode functions;

[0147] The modal function is input into the optimal pollution discharge model to obtain the pollution discharge value of each first partition.

[0148] Step S105: Obtain the central position of the ocean ranch, and combine the sewage pollution value and the pollution value of each first partition to obtain the current pollution level of the ocean ranch; and issue an early warning for ocean ranch pollution based on a preset pollution level threshold and the current pollution level.

[0149] In this embodiment, the central position of the ocean ranch is obtained, and the current pollution level of the ocean ranch is obtained by combining the pollution level prediction value, the sewage pollution value and the pollution value of each first subarea; and according to the preset pollution level threshold, combined with the current pollution level, an early warning of pollution of the ocean ranch is issued, specifically including:

[0150] Obtaining the central position of the ocean ranch, and calculating the sewage discharge proportion of each first partition according to the central position and the position of the first sewage outlet;

[0151] Calculate the current pollution degree of the marine ranch according to the pollution discharge proportion, pollution value and pollution discharge pollution value of each of the first sub-areas;

[0152] When the current pollution level exceeds a preset pollution level threshold, an early warning of marine ranch pollution is issued so that the marine ranch staff can adjust the sewage discharge plan of each first sewage outlet according to the current pollution level.

[0153] In an optional embodiment, the current pollution degree of the ocean ranch is calculated according to the pollution discharge proportion, pollution value and pollution discharge pollution value of each of the first subareas, specifically including:

[0154] Multiplying the pollutant discharge weight of each first partition by the pollutant discharge pollution value to obtain a theoretical pollutant discharge pollution value of each first partition;

[0155] The theoretical pollution value of each first partition is added to the pollution value to obtain the current pollution degree of each first partition;

[0156] The current pollution degree of each first partition is added up to obtain the current pollution degree of the ocean ranch.

[0157] In an optional embodiment, the obtaining of the central position of the ocean ranch and calculating the sewage discharge proportion of each first partition according to the central position and the position of the first sewage outlet specifically include:

[0158] Get the central location of the ocean ranch;

[0159] Calculate the geometric distance between each first sewage outlet and the center position to obtain the geometric distance of the first sewage outlet;

[0160] The geometric distances of the first sewage outlet are sorted from far to near to obtain a sorting result;

[0161] The pollution discharge weight is assigned according to the sorting result to obtain the pollution discharge weight of each first partition.

[0162] In an optional embodiment, when the current pollution level exceeds a preset pollution level threshold, an early warning is issued for marine ranch pollution, so that the marine ranch staff can adjust the sewage discharge plan of each first sewage outlet according to the current pollution level, specifically including:

[0163] When the current pollution level exceeds a preset pollution level threshold, detecting the pollution value and the discharge pollution value of each first partition;

[0164] When the pollution value of the first partition exceeds the preset partition pollution threshold, the pollutants of the first partition are cleaned so that the pollution value of the corresponding first partition meets the standard;

[0165] When the pollution value of the first zone exceeds the preset pollution value threshold, the pollution plan of the first zone is adjusted to obtain an adjusted pollution plan, until the adjusted pollution value recalculated according to the adjusted pollution plan meets the standard.

[0166] The discharge proportion is obtained through the location of each sewage outlet, and then according to the current pollution value and the predicted pollution value, the marine ranch pollution warning based on the combination of the present and the future is realized. From the two perspectives of pollution, the accuracy of the marine ranch pollution warning is improved.

[0167] This embodiment divides the marine ranch and remote sensing images into blocks according to the locations of sewage outlets in the marine ranch. The blocks are divided according to the locations of sewage outlets, which can deepen the correlation between each block and the actual sewage discharge situation of the marine ranch, and extract and identify pixels based on remote sensing images, so as to accurately identify the pollutants and pollution values ​​of each block; by mining the relationship between the historical sewage discharge data of the marine ranch and the pollution value, the sewage discharge pollution value of the marine ranch can be obtained according to the sewage discharge plan of the marine ranch, so that the future pollution changes of the marine ranch are taken into consideration in the pollution warning of the marine ranch. Through the prediction of current pollution values ​​and future pollution values, the marine ranch pollution degree with higher accuracy can be comprehensively obtained, so as to better warn of marine ranch pollution, reduce environmental pollution of the marine ranch, and maintain the normal operation of the marine ranch.

[0168] Embodiment 2

[0169] Please refer to Figure 2 , which is a structural diagram of a marine ranch pollution warning system provided in an embodiment of the present invention, including: a partition module 201, a pollution value acquisition module 202, a pollution model acquisition module 203, a pollution value acquisition module 204 and a pollution warning module 205.

[0170] The partitioning module 201 is used to obtain the remote sensing image and sewage outlet location of the marine ranch; according to the sewage outlet location, the remote sensing image is cut to obtain a plurality of first partitions and their corresponding first partition images and first sewage outlets.

[0171] In this embodiment, the partition module 201 includes: a partition unit;

[0172] The partitioning unit is used to use the position of each sewage outlet as the geometric center point of the partition, and to cut the marine ranch to obtain a plurality of first partitions and a first sewage outlet corresponding to each first partition;

[0173] The remote sensing image is cut according to the first partitions to obtain a first partition image corresponding to each first partition.

[0174] The pollution value acquisition module 202 is used to identify each first subarea image according to the pixel points of each first subarea image in combination with a preset pollutant database to obtain the pollutants and pollution values ​​of each first subarea.

[0175] In this embodiment, the pollution value acquisition module 202 includes: a pollution value acquisition unit;

[0176] The pollution value acquisition unit is used to perform denoising on the first subarea image to obtain a second subarea image of each first subarea;

[0177] Reconstructing the second subarea images to obtain a stereo remote sensing image of each first subarea;

[0178] Performing image segmentation in a vertical direction on the stereo remote sensing image to obtain a plurality of layered images of each first subarea;

[0179] In combination with a preset pollutant database, pixel extraction and identification are performed on each of the layered images to obtain pollutant data and pollution values ​​in each layered image, and further obtain pollutant data and pollution values ​​of each first partition.

[0180] In this embodiment, the pollution value acquisition unit includes: a pollution value acquisition subunit;

[0181] The pollution value acquisition subunit is used to extract pixels from each of the layered images to obtain a pixel data set of each layered image; the pixel data set includes: the color of the pixel, the brightness of the pixel and the size of the pixel;

[0182] Constructing a key-value pair for each pixel according to the color and brightness of the pixel;

[0183] Iteratively clustering the pixels in each layered image according to the key-value pairs to obtain a clustering result for each layered image, wherein the clustering result includes: a plurality of cluster centers;

[0184] The preset pollutant database includes: images of pollutants and pollution risk values;

[0185] Extract and identify pixels of pollutant images to obtain pixel color and pixel brightness of each pollutant in the pollutant database;

[0186] According to the pixel color and pixel brightness of the cluster center, the pixel color and pixel brightness of each pollutant in the pollutant database are matched to obtain the pollutant corresponding to each cluster center, and then the pollutant of each layered image is obtained;

[0187] According to the pollutants of each layered image, the pollutants of each first partition are obtained;

[0188] According to the pollution risk value of the pollutant and the size of the pixel point, in combination with the clustering result, the pollution value of each layered image is calculated, and then the pollution value of each first partition is obtained.

[0189] In this embodiment, the pollution value acquisition subunit includes: a pollution value acquisition component;

[0190] The pollution value acquisition component is used for the clustering result including: the cluster corresponding to the cluster center and the number of pixels in each cluster;

[0191] According to the size of the pixels and the number of pixels in each cluster, the area of ​​each cluster is calculated;

[0192] According to the area of ​​each cluster and the pollution risk value of the pollutant, the pollution value of each pollutant in the layered image is calculated;

[0193] According to the pollution value of each pollutant, the pollution value of each layered image is obtained, and then the pollution value of each first partition is obtained.

[0194] The pollution model acquisition module 203 is used to obtain the historical pollution data of each first sewage outlet, extract pollutants from the historical pollution data, and iteratively train the preset algorithm model in combination with the pollutants and pollution values ​​of each first partition to obtain the optimal pollution model.

[0195] In this embodiment, the pollution model acquisition module 203 includes: a pollution model acquisition unit;

[0196] The sewage model acquisition unit is used to acquire the historical sewage discharge data of each first sewage outlet, perform pollutant extraction on the historical sewage discharge data, and obtain the historical pollutant data of each first sewage outlet;

[0197] Constructing an objective function of the algorithm model according to the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition;

[0198] Decomposing the historical pollutant data according to a variational mode decomposition algorithm to obtain a plurality of pollutant mode functions;

[0199] The pollutant modal function, the pollutants and pollution values ​​of the first partition are input into the algorithm model, and combined with the objective function, the algorithm model is iteratively trained until the value of the objective function meets the preset threshold requirement, the training of the algorithm model is completed, and the optimal pollution discharge model is obtained.

[0200] In this embodiment, the pollution model acquisition unit includes: an objective function construction subunit;

[0201] The objective function construction subunit is used to construct the objective function of the algorithm model according to the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition, wherein the objective function is:

[0202] ,

[0203] in, Indicates the type of pollutant, represents the set of pollutant types, The pollution factor representing the pollutant can be obtained by querying the pollutant database; Indicates the proportion of the pollutant; Indicates the total amount of sewage discharged from the first sewage outlet. Indicates the first sewage outlet The discharge time of each pollutant; Represents the pollution value of the first partition.

[0204] The sewage pollution value acquisition module 204 is used to obtain the sewage pollution value of each first partition according to the sewage discharge plan of each first sewage outlet in combination with the optimal sewage discharge model.

[0205] The pollution warning module 205 is used to obtain the central position of the ocean ranch, and combine the pollution prediction value, sewage pollution value and pollution value of each first partition to obtain the current pollution level of the ocean ranch; according to the preset pollution level threshold, combined with the current pollution level, an early warning is issued for ocean ranch pollution.

[0206] In this embodiment, the pollution warning module 205 includes: a pollution warning unit;

[0207] The pollution early warning unit is used to obtain the central position of the ocean ranch, and calculate the sewage discharge proportion of each first partition according to the central position and the position of the first sewage outlet;

[0208] Calculate the current pollution degree of the marine ranch according to the pollution discharge proportion, pollution value and pollution discharge pollution value of each of the first sub-areas;

[0209] When the current pollution level exceeds a preset pollution level threshold, an early warning of marine ranch pollution is issued so that the marine ranch staff can adjust the sewage discharge plan of each first sewage outlet according to the current pollution level.

[0210] This embodiment divides the marine ranch and remote sensing images into blocks according to the locations of sewage outlets in the marine ranch. The blocks are divided according to the locations of sewage outlets, which can deepen the correlation between each block and the actual sewage discharge situation of the marine ranch, and extract and identify pixels based on remote sensing images, so as to accurately identify the pollutants and pollution values ​​of each block; by mining the relationship between the historical sewage discharge data of the marine ranch and the pollution value, the sewage discharge pollution value of the marine ranch can be obtained according to the sewage discharge plan of the marine ranch, so that the future pollution changes of the marine ranch are taken into consideration in the pollution warning of the marine ranch. Through the prediction of current pollution values ​​and future pollution values, the marine ranch pollution degree with higher accuracy can be comprehensively obtained, so as to better warn of marine ranch pollution, reduce environmental pollution of the marine ranch, and maintain the normal operation of the marine ranch.

[0211] In summary, the embodiment of the present invention divides the marine ranch and remote sensing images into blocks according to the location of the sewage outlet in the marine ranch, and divides the blocks according to the location of the sewage outlet, so as to deepen the correlation between each block and the actual sewage discharge situation of the marine ranch, and extracts and identifies pixels based on the remote sensing image, so as to accurately identify the pollutants and pollution values ​​of each block; by mining the relationship between the historical sewage discharge data of the marine ranch and the pollution value, the sewage discharge pollution value of the marine ranch can be obtained according to the sewage discharge plan of the marine ranch, so that the future pollution changes of the marine ranch are taken into consideration in the pollution warning of the marine ranch, and through the prediction of the current pollution value and the future pollution value, the marine ranch pollution degree with higher accuracy can be comprehensively obtained, so as to better warn of the marine ranch pollution, reduce the environmental pollution of the marine ranch, and maintain the normal work of the marine ranch.

[0212] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A marine ranch pollution early warning method, characterized in that: include: Obtain remote sensing images of the marine ranch and the location of sewage outfalls; According to the position of the sewage outlet, the remote sensing image is cut to obtain a plurality of first subareas and their corresponding first subarea images and a first sewage outlet; According to the pixel points of each first subarea image, combined with a preset pollutant database, each first subarea image is identified to obtain the pollutants and pollution values ​​of each first subarea; Obtaining historical sewage discharge data of each first sewage outlet, extracting pollutants from the historical sewage discharge data, and iteratively training a preset algorithm model in combination with pollutants and pollution values ​​of each first partition to obtain an optimal sewage discharge model; According to the sewage discharge plan of each first sewage outlet, combined with the optimal sewage discharge model, the sewage discharge pollution value of each first partition is obtained; Obtaining the central position of the ocean ranch, and combining the sewage pollution value and the pollution value of each first subarea to obtain the current pollution degree of the ocean ranch; According to the preset pollution threshold, combined with the current pollution level, early warning of marine ranch pollution is issued; The step of identifying each first subarea image according to the pixel points of each first subarea image and combining with a preset pollutant database to obtain pollutant data and pollution value of each first subarea specifically includes: Performing denoising on the first subarea image to obtain a second subarea image of each first subarea; Reconstructing the second subarea images to obtain a stereo remote sensing image of each first subarea; Performing image segmentation in a vertical direction on the stereo remote sensing image to obtain a plurality of layered images of each first subarea; In combination with a preset pollutant database, pixel points of each layered image are extracted and identified to obtain pollutant data and pollution values ​​in each layered image, and then the pollutant data and pollution values ​​of each first partition are obtained; The obtaining of historical sewage discharge data of each first sewage outlet, extracting pollutants from the historical sewage discharge data, and iteratively training a preset algorithm model in combination with pollutants and pollution values ​​of each first partition to obtain an optimal sewage discharge model specifically includes: Acquire historical sewage discharge data of each first sewage discharge outlet, perform pollutant extraction on the historical sewage discharge data, and obtain historical pollutant data of each first sewage discharge outlet; Constructing an objective function of the algorithm model according to the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition; Decomposing the historical pollutant data according to a variational mode decomposition algorithm to obtain a plurality of pollutant mode functions; Inputting the pollutant modal function, the pollutants and the pollution value of the first partition into the algorithm model, combining the objective function, iteratively training the algorithm model until the value of the objective function meets the preset threshold requirement, completing the training of the algorithm model, and obtaining the optimal pollution discharge model; The objective function and pollution discharge constraints of the algorithm model are constructed according to the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition, specifically including: According to the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition, the objective function of the algorithm model is constructed, wherein the objective function is: ， in, Indicates the type of pollutant, represents the set of pollutant types, Indicates The pollution factors of the pollutants can be obtained by querying the pollutant database; Indicates The proportion of pollutants of this type; Indicates the total amount of sewage discharged from the first sewage outlet. Indicates the first sewage outlet The discharge time of each pollutant; Represents the pollution value of the first partition.

2. A marine ranch pollution early warning method as claimed in claim 1, characterized in that: The step of cutting the remote sensing image according to the position of the sewage outlet to obtain a plurality of first subareas and their corresponding first subarea images and the first sewage outlet specifically includes: Taking the position of each sewage outlet as the geometric center point of the partition, the marine ranch is cut to obtain a plurality of first partitions and a first sewage outlet corresponding to each first partition; The remote sensing image is cut according to the first partitions to obtain a first partition image corresponding to each first partition.

3. The marine ranch pollution early warning method according to claim 1, characterized in that: The method of extracting and identifying pixels of each layered image in combination with a preset pollutant database to obtain pollutant data and pollution values ​​in each layered image, and then obtaining pollutants and pollution values ​​in each first partition, specifically includes: Extracting pixels from each of the layered images respectively to obtain a pixel data set of each layered image; the pixel data set includes: the color of the pixel, the brightness of the pixel and the size of the pixel; Constructing a key-value pair for each pixel according to the color and brightness of the pixel; Iteratively clustering the pixels in each layered image according to the key-value pairs to obtain a clustering result for each layered image, wherein the clustering result includes: a plurality of cluster centers; The preset pollutant database includes: images of pollutants and pollution risk values; Extract and identify pixels of pollutant images to obtain pixel color and pixel brightness of each pollutant in the pollutant database; According to the pixel color and pixel brightness of the cluster center, the pixel color and pixel brightness of each pollutant in the pollutant database are matched to obtain the pollutant corresponding to each cluster center, and then the pollutant of each layered image is obtained; According to the pollutants of each layered image, the pollutants of each first partition are obtained; According to the pollution risk value of the pollutant and the size of the pixel point, in combination with the clustering result, the pollution value of each layered image is calculated, and then the pollution value of each first partition is obtained.

4. A marine ranch pollution early warning method as claimed in claim 3, characterized in that: The step of calculating the pollution value of each layered image according to the pollution risk value of the pollutant and the size of the pixel point in combination with the clustering result, and then obtaining the pollution value of each first partition, specifically includes: The clustering result includes: the cluster corresponding to the cluster center and the number of pixels in each cluster; According to the size of the pixels and the number of pixels in each cluster, the area of ​​each cluster is calculated; According to the area of ​​each cluster and the pollution risk value of the pollutant, the pollution value of each pollutant in the layered image is calculated; According to the pollution value of each pollutant, the pollution value of each layered image is obtained, and then the pollution value of each first partition is obtained.

5. The marine ranch pollution early warning method according to claim 1, characterized in that: The central position of the ocean ranch is obtained, and the current pollution degree of the ocean ranch is obtained by combining the pollution degree prediction value, the sewage pollution value and the pollution value of each first subarea; According to the preset pollution threshold, combined with the current pollution level, early warning of marine ranch pollution is issued, including: Obtaining the central position of the ocean ranch, and calculating the sewage discharge proportion of each first partition according to the central position and the position of the first sewage outlet; Calculate the current pollution degree of the marine ranch according to the pollution discharge proportion, pollution value and pollution discharge pollution value of each of the first sub-areas; When the current pollution level exceeds a preset pollution level threshold, an early warning of marine ranch pollution is issued so that the marine ranch staff can adjust the sewage discharge plan of each first sewage outlet according to the current pollution level.

6. A marine ranch pollution early warning system, characterized in that: include: Partitioning module, pollution value acquisition module, pollution model acquisition module, pollution value acquisition module and pollution early warning module; The partition module is used to obtain the remote sensing image and the location of the sewage outlet of the marine ranch; according to the location of the sewage outlet, the remote sensing image is cut to obtain a plurality of first partitions and their corresponding first partition images and first sewage outlets; The pollution value acquisition module is used to identify each first subarea image according to the pixel points of each first subarea image in combination with a preset pollutant database to obtain the pollutants and pollution values ​​of each first subarea; The sewage discharge model acquisition module is used to obtain the historical sewage discharge data of each first sewage discharge outlet, extract pollutants from the historical sewage discharge data, and iteratively train the preset algorithm model in combination with the pollutants and pollution values ​​of each first partition to obtain the optimal sewage discharge model; The sewage pollution value acquisition module is used to obtain the sewage pollution value of each first partition according to the sewage discharge plan of each first sewage outlet in combination with the optimal sewage discharge model; The pollution warning module is used to obtain the central position of the marine ranch, and combine the pollution degree prediction value, the sewage pollution value and the pollution value of each first subarea to obtain the current pollution degree of the marine ranch; according to the preset pollution degree threshold, combined with the current pollution degree, to warn of marine ranch pollution; The pollution value acquisition module includes: a pollution value acquisition unit; The pollution value acquisition unit is used to perform denoising on the first subarea image to obtain a second subarea image of each first subarea; Reconstructing the second subarea images to obtain a stereo remote sensing image of each first subarea; Performing image segmentation in a vertical direction on the stereo remote sensing image to obtain a plurality of layered images of each first subarea; In combination with a preset pollutant database, pixel points of each layered image are extracted and identified to obtain pollutant data and pollution values ​​in each layered image, and then the pollutant data and pollution values ​​of each first partition are obtained; The pollution model acquisition module includes: a pollution model acquisition unit; The sewage model acquisition unit is used to acquire the historical sewage discharge data of each first sewage outlet, perform pollutant extraction on the historical sewage discharge data, and obtain the historical pollutant data of each first sewage outlet; Constructing an objective function of the algorithm model according to the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition; Decomposing the historical pollutant data according to a variational mode decomposition algorithm to obtain a plurality of pollutant mode functions; Inputting the pollutant modal function, the pollutants and the pollution value of the first partition into the algorithm model, combining the objective function, iteratively training the algorithm model until the value of the objective function meets the preset threshold requirement, completing the training of the algorithm model, and obtaining the optimal pollution discharge model; The sewage model acquisition unit includes: an objective function construction subunit; The objective function construction subunit is used to construct the objective function of the algorithm model according to the historical pollutant data of each first sewage outlet and the pollutants and pollution values ​​of each first partition, wherein the objective function is: ， in, Indicates the type of pollutant, represents the set of pollutant types, Indicates The pollution factors of the pollutants can be obtained by querying the pollutant database; Indicates The proportion of pollutants of this type; Indicates the total amount of sewage discharged from the first sewage outlet. Indicates the first sewage outlet The discharge time of each pollutant; Represents the pollution value of the first partition.

Images