A chemical enterprise pollution risk factor distribution identification method and device

Abstract

The application provides a chemical enterprise pollution risk element distribution identification method and device, and the method comprises the following steps: inputting a to-be-identified remote sensing image into a chemical building element segmentation model to obtain a chemical building element identification result; performing mask processing on the to-be-identified remote sensing image by using the chemical building element identification result, and retaining non-chemical building elements; inputting the processed to-be-identified remote sensing image into a non-chemical building element identification model to obtain a non-chemical building element identification result; based on the chemical building element identification result and the non-chemical building element identification result, performing reclassification processing on the identified elements by using a spatial analysis method to obtain a chemical enterprise pollution risk element distribution identification result, and the categories of the pollution risk elements after the reclassification processing include potential pollution sources, migration pathways and receptors. Through the application, automatic and efficient identification of the chemical enterprise pollution risk elements with high spatial heterogeneity is realized, and the work efficiency is greatly improved.

Description

Technical Field

[0001] This invention relates to the field of pollutant identification technology, specifically to a method and apparatus for identifying the distribution of pollution risk factors in chemical enterprises. Background Technology

[0002] The high risks and difficulty in precise management within the chemical industry lead to frequent chemical accidents, causing casualties and severe environmental pollution. Currently, the construction and improvement of chemical industrial parks have reduced the dispersion of chemical enterprises, greatly facilitating macro-level control of the industry by management departments. However, efficient and precise risk factor management measures are still lacking for individual chemical enterprises. The distribution of risk factors within each chemical enterprise contains the entire enterprise's safety information. Mastering this information is a crucial link in the major accident prevention and control system, the foundation for environmental supervision and risk prevention in chemical enterprises, and also provides valuable basic data for contaminated site investigation and assessment. Traditional pollution risk factor identification is based on remote sensing images, combined with enterprise production layout and on-site inspections, and completed through manual visual interpretation or experimental analysis. However, this method is time-consuming, labor-intensive, inefficient, and requires multiple identifications for enterprises where pollution risk factors have changed historically. Therefore, there is an urgent need to develop an automatic identification method for the distribution of pollution risk factors in chemical enterprises to improve the work efficiency of relevant functional departments. Summary of the Invention

[0003] Therefore, the technical problem to be solved by the present invention is to overcome the shortcomings of the low efficiency of identifying the distribution of pollution risk factors in chemical enterprises in the prior art, thereby providing a method and device for identifying the distribution of pollution risk factors in chemical enterprises.

[0004] The first aspect of this invention provides a method for identifying the distribution of pollution risk factors in chemical enterprises, comprising: acquiring a remote sensing image to be identified; inputting the remote sensing image to be identified into a pre-trained chemical building element segmentation model to obtain chemical building element identification results; using the chemical building element identification results to perform masking processing on the remote sensing image to be identified, retaining non-chemical building elements in the remote sensing image to be identified; inputting the masked remote sensing image to be identified into a pre-trained non-chemical building element identification model to obtain non-chemical building element identification results; using the chemical building element identification results and the non-chemical building element identification results, employing a spatial analysis method to reclassify the pollution risk factors in the remote sensing image to be identified, thereby obtaining the distribution identification results of pollution risk factors in chemical enterprises. After reclassification, the categories to which the pollution risk factors belong include potential pollution sources, migration pathways, and receptors, and the pollution risk factors include chemical building elements and non-chemical building elements.

[0005] Optionally, the method for identifying the distribution of pollution risk factors in chemical enterprises provided by the present invention further includes: fusing the identification results of the pollution risk factor distribution with the remote sensing image to be identified to obtain a distribution map of pollution risk factors in chemical enterprises.

[0006] Optionally, in the method for identifying the distribution of pollution risk factors in chemical enterprises provided by the present invention, the chemical building element segmentation model is trained through the following steps: acquiring remote sensing images of chemical enterprises; establishing a sample set based on the remote sensing images of chemical enterprises, the sample set including labels of pollution risk factors, the pollution risk factors in the remote sensing images of chemical enterprises including chemical building elements; training the first network model through the sample set to obtain the chemical building element segmentation model.

[0007] Optionally, in the method for identifying the distribution of pollution risk factors in chemical enterprises provided by the present invention, a chemical building element segmentation model is obtained by training a first network model with a sample set, including: enhancing the sample set by rotation and / or random flipping to obtain an enhanced dataset; inputting the enhanced dataset into the first network model to train the first network model to obtain the chemical building element segmentation model.

[0008] Optionally, in the method for identifying the distribution of pollution risk factors in chemical enterprises provided by the present invention, the first network model is a U-Net network.

[0009] Optionally, in the method for identifying the distribution of pollution risk factors in chemical enterprises provided by the present invention, the non-chemical building element identification model is trained through the following steps: acquiring remote sensing images of chemical enterprises; establishing a sample set based on the remote sensing images of chemical enterprises, the sample set including labels for pollution risk factors, the pollution risk factors in the remote sensing images of chemical enterprises including chemical building elements and non-chemical building elements; performing masking processing on the sample set to retain the non-chemical building elements in the sample set; inputting the masked sample set into a second network model to train the second network model to obtain the non-chemical building element identification model.

[0010] Optionally, in the method for identifying the distribution of pollution risk factors in chemical enterprises provided by the present invention, the second network model is a Spectral-Spatial-CNN network.

[0011] The second aspect of this invention provides a device for identifying the distribution of pollution risk factors in chemical enterprises, comprising: an image acquisition module for acquiring a remote sensing image to be identified; a first identification module for inputting the remote sensing image to be identified into a pre-trained chemical building element segmentation model to obtain chemical building element identification results; a masking module for masking the remote sensing image to be identified using the chemical building element identification results, retaining non-chemical building elements in the remote sensing image to be identified; a second identification module for inputting the masked remote sensing image to be identified into a pre-trained non-chemical building element identification model to obtain non-chemical building element identification results; and a risk factor distribution identification module for using the chemical building element identification results and the non-chemical building element identification results to perform spatial analysis to reclassify the pollution risk factors in the remote sensing image to be identified into potential pollution sources, migration pathways, and receptors, thereby obtaining the pollution risk factor distribution identification results for chemical enterprises. After reclassification, the categories to which the pollution risk factors belong include potential pollution sources, migration pathways, and receptors, and the pollution risk factors include chemical building elements and non-chemical building elements.

[0012] A third aspect of the present invention provides a computer device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to perform the method for identifying the distribution of pollution risk factors in chemical enterprises as provided in the first aspect of the present invention.

[0013] A fourth aspect of the present invention provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores computer instructions for causing a computer to execute the method for identifying the distribution of pollution risk factors in chemical enterprises as provided in the first aspect of the present invention.

[0014] The technical solution of this invention has the following advantages:

[0015] The present invention provides a method and apparatus for identifying the distribution of pollution risk factors in chemical enterprises. After acquiring the remote sensing image to be identified, it adopts different deep learning network models to segment chemical building elements and classify non-chemical building elements, respectively, to address the spatial heterogeneity issues of chemical building elements and non-chemical building elements. Through model integration, it achieves automatic and efficient identification of pollution risk factors in chemical enterprises with high spatial heterogeneity. Compared with traditional manual visual interpretation and identification methods, the work efficiency is greatly improved. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a specific example of a method for identifying the distribution of pollution risk factors in chemical enterprises, as described in this invention.

[0018] Figure 2 The image to be identified in this embodiment of the invention;

[0019] Figure 3 This is a distribution map of pollution risk factors in an embodiment of the present invention;

[0020] Figure 4 This is a schematic diagram illustrating a specific example of a device for identifying the distribution of pollution risk factors in chemical enterprises, as described in this invention.

[0021] Figure 5 This is a schematic diagram illustrating a specific example of a computer device in an embodiment of the present invention. Detailed Implementation

[0022] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] In the description of this invention, it should be noted that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0024] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0025] With the rapid development of remote sensing technology, the spatiotemporal resolution of remote sensing images has gradually improved, and they contain rich information about ground features. Artificial intelligence technologies, represented by deep learning, are increasingly being applied to remote sensing image ground feature classification and feature recognition, greatly promoting the automatic and efficient processing of massive amounts of remote sensing image data. In recent years, deep learning methods have made significant progress in semantic segmentation, object detection, and image classification, and have been widely applied in urban functional area identification, building land extraction, and mining area ground feature extraction. Regarding pollution risk factor identification, due to the sparsity of potential pollution sources distributed over large areas, developing automatic pollution risk factor identification technology is costly and inefficient, and the diversity of scenarios results in the lack of universally applicable solutions. However, for chemical industrial parks or chemical enterprises within parks, pollution risk factor distribution maps have high value density and high application management value. They are of significant practical importance for understanding the distribution of pollution risk factors in chemical enterprises, detecting changes in pollution risk factors, guiding the construction planning of chemical enterprises, and combining historical remote sensing imagery to assist in the investigation of contaminated sites.

[0026] This invention provides a method for identifying the distribution of pollution risk factors in chemical enterprises, such as... Figure 1 As shown, it includes:

[0027] Step S11: Acquire the remote sensing image to be identified.

[0028] In one alternative embodiment, the remote sensing image to be identified is a remote sensing image of an area where chemical enterprises are located.

[0029] In one optional embodiment, a 0.6-meter resolution Google Earth image (a true-color remote sensing image product resulting from the fusion of multi-source remote sensing images) is used to extract the remote sensing image to be identified through spatial analysis.

[0030] Step S12: Input the remote sensing image to be identified into the pre-trained chemical building element segmentation model to obtain the chemical building element identification result.

[0031] Step S13: Combine the chemical building element identification results to perform masking processing on the remote sensing image to be identified, and retain non-chemical building elements in the remote sensing image to be identified.

[0032] In this embodiment of the invention, when performing masking processing on the remote sensing image to be identified, chemical building elements in the remote sensing image to be identified can be masked, thereby retaining non-chemical building elements.

[0033] Step S14: Input the masked remote sensing image to be identified into the pre-trained non-chemical building element recognition model to obtain the non-chemical building element recognition result.

[0034] In one optional embodiment, the pollution risk factors of chemical enterprises include two categories: chemical building elements and non-chemical building elements. Among them, chemical building elements are taller and have obvious geometric features. Since the features of chemical building elements and non-chemical building elements in remote sensing images are quite different, this embodiment of the invention uses two different models to identify chemical building elements and non-chemical building elements in the remote sensing images to be identified.

[0035] In one alternative embodiment, chemical building elements include geometrically distinctive and tall office areas and production plants.

[0036] In one alternative embodiment, non-chemical building elements include special features such as open-air production equipment, storage tank areas, open-air cargo storage, open-air pipelines, chimneys, cooling towers, water reservoirs, paved roads, bare land, natural water bodies, and vegetation.

[0037] In one optional embodiment, when inputting the remote sensing image to be identified into the chemical building element segmentation model and the non-chemical building element identification model, the sliding window and sliding step size are adjusted according to the spatial resolution, spatial range and other characteristics of the remote sensing image to be identified, combined with the minimum coverage area of ​​different pollution risk elements, in order to optimize the identification effect.

[0038] Step S15: Using the identification results of chemical building elements and non-chemical building elements, the pollution risk elements in the remote sensing image to be identified are reclassified using spatial analysis method to obtain the distribution identification results of pollution risk elements of chemical enterprises. After reclassification, the categories to which the pollution risk elements belong include potential pollution sources, migration pathways and receptors. The pollution risk elements include chemical building elements and non-chemical building elements.

[0039] In one optional embodiment, the reclassification of pollution risk factors refers to reclassifying the various chemical and non-chemical building elements identified into three categories: potential pollution sources, migration pathways, and receptors, which serve as the final pollution risk factor identification result.

[0040] To strengthen environmental risk management of construction projects, environmental risk assessments are required for projects involving production, use, and storage (including pipeline transportation). Risk identification involves assessing the hazards of substances (raw materials, fuels, pollutants, etc.), the hazards of production systems (production equipment, storage and transportation facilities, auxiliary production facilities, etc.), and the transfer pathways of hazardous substances, and analyzing potential environmentally sensitive targets. Therefore, in the task of identifying pollution risk factors for chemical enterprises, considering the site layout, potential pollution sources within the chemical enterprise, their possible migration pathways, and the receptors that potential pollution sources may affect, pollution risk factors can be identified from three aspects: potential pollution sources, migration pathways, and receptors.

[0041] In one optional embodiment, environmental risk factor analysis is performed on each chemical building element and non-chemical building element according to potential pollution sources, migration pathways, and receptors. Potential pollution source elements with a high probability of potential pollution include production plants, open-air production equipment, storage tank areas, open-air cargo storage, open-air pipelines, chimneys, and water storage ponds; migration pathway elements that may cause pollutant migration include bare land and natural water bodies; receptor elements that may be affected by potential pollution source elements include office areas and vegetation.

[0042] The method for identifying the distribution of pollution risk factors in chemical enterprises provided in this invention, after acquiring the remote sensing image to be identified, uses different deep learning network models to segment chemical building elements and classify non-chemical building elements to address the spatial heterogeneity issues of chemical building elements and non-chemical building elements. Through model integration, it achieves automatic and efficient identification of pollution risk factors in chemical enterprises with high spatial heterogeneity, which greatly improves work efficiency compared to traditional manual visual interpretation and identification methods.

[0043] In an optional embodiment, the method for identifying the distribution of pollution risk factors in chemical enterprises provided by this invention further includes, after performing step S15, the following:

[0044] By fusing the results of pollution risk factor distribution identification with the remote sensing image to be identified, a pollution risk factor distribution map of chemical enterprises is obtained.

[0045] like Figure 2 The image shown is a remote sensing image to be identified in a specific embodiment. Figure 3 To and Figure 2 The corresponding pollution risk factor distribution map can intuitively show the distribution of potential pollution sources, migration pathways, and receptors, thereby facilitating pollution management.

[0046] In an alternative embodiment, the optimal visual effect can be achieved by adjusting the weight parameters of image fusion to meet management requirements.

[0047] In an optional embodiment, the chemical building element segmentation model is trained through the following steps:

[0048] First, remote sensing images of chemical enterprises are acquired. Detailed information regarding these images can be found in the description of the remote sensing images to be identified in the above embodiments, and will not be repeated here.

[0049] Then, a sample set is established based on remote sensing images of chemical enterprises. The sample set includes labels for pollution risk factors, which include chemical building elements in the remote sensing images of chemical enterprises.

[0050] Finally, the first network model was trained using the sample set to obtain the chemical building element segmentation model.

[0051] In an optional embodiment, the step of training the first network model using a sample set to obtain a chemical building element segmentation model specifically includes:

[0052] First, the sample set is augmented by rotation and / or random flipping to obtain an augmented dataset.

[0053] Then, the augmented dataset is input into the first network model to train the first network model and obtain the chemical building element segmentation model.

[0054] In an alternative embodiment, the first network model is a U-Net network. The U-Net network algorithm can better identify building outlines.

[0055] In an optional embodiment, the first network model is trained using stochastic gradient descent (SGD). For the multi-classification problem of chemical building elements, a softmax function is used as the output, and a cross-entropy function is used as the loss function. The loss function is calculated as follows:

[0056]

[0057] For each pixel, the label is l:Ω→{1,…,K}. A pixel weight ω(X) is introduced, giving higher weights to pixels closer to building boundaries in the image, where p l(X) (X) is the soft-max loss function.

[0058] The weight of each pixel in the loss function is pre-calculated, allowing the network to focus more on learning the edges of interconnected chemical structures. Morphological calculation of the segmentation boundaries is used, and the specific calculation method is as follows:

[0059]

[0060] Where ω c (X) represents the weights balancing the proportions of building categories, d1(X) represents the distance to the boundary of the nearest chemical building feature, d2(X) represents the distance to the boundary of the second nearest chemical building feature, and ω0 and σ are constant values. The initial constant values ​​ω0 and σ are adjusted so that all feature maps have approximately unit variance. The standard deviation is used as... The distribution is initialized using a Gaussian distribution, where N is the number of input nodes for each neuron.

[0061] In one optional embodiment, when training the first network model, the network is adjusted according to the recognition results to extract important features such as shape and texture of various chemical building elements and output a chemical building element segmentation model.

[0062] In an optional embodiment, the non-chemical building element recognition model is trained through the following steps:

[0063] First, remote sensing images of chemical enterprises are acquired. Detailed information regarding these images can be found in the description of the remote sensing images to be identified in the above embodiments, and will not be repeated here.

[0064] Secondly, a sample set is established based on remote sensing images of chemical enterprises. The sample set includes labels for pollution risk factors, which include chemical building elements and non-chemical building elements in the remote sensing images of chemical enterprises.

[0065] Then, the sample set is masked to retain the non-chemical building elements in the sample set.

[0066] Finally, the masked sample set is input into the second network model to train the second network model and obtain the non-chemical building element recognition model.

[0067] In one optional embodiment, non-chemical building elements include special features such as open-air production equipment, storage tank areas, open-air cargo storage, open-air pipelines, chimneys, cooling towers, water reservoirs, paved roads, bare land, natural water bodies, and vegetation. For individual elements, the main characteristics of open-air cargo storage areas, open-air pipelines, and paved roads are shape and texture features; the main characteristics of water reservoirs are color and shape features; and the main characteristics of bare land, natural water bodies, and vegetation are color features.

[0068] Non-chemical building elements within chemical plant sites exhibit unique spatial relationships and significant spatial variability. Identifying pollution risk factors from open-air production equipment, chimneys, and cooling towers is particularly challenging. Open-air production equipment includes large heating furnaces and distillation towers, medium-sized reactors and fans, as well as small machinery, each possessing unique textural features. Chimneys and cooling towers are relatively tall, casting noticeable shadows, and remote sensing imagery is significantly affected by projection angles. Therefore, to address the diversity and imbalance of the sample set, this invention first optimizes the sample set and then augments the data through rotation and random flipping. Finally, the augmented dataset is input into a second network model for training. The resulting non-chemical building element identification model can accurately identify non-chemical building elements within remote sensing images of chemical plants by combining textural and shadow features.

[0069] In one optional embodiment, to adapt to the spatial variability within the chemical enterprise site and to effectively identify various non-chemical building elements within the chemical enterprise site, a Spectral-Spatial-CNN network is used as the second network model.

[0070] In one alternative embodiment, an autoencoder is employed for spatial feature learning, using end-to-end supervised learning to adapt to the task of classifying non-chemical building elements. Fewer independent connection weights are used, thus enabling it to adapt to tasks such as identifying non-chemical building elements like reservoirs where labeled data is scarce. The AdaGrad parameter update algorithm is used for optimization, automatically adjusting the learning rate based on changes in the network training samples. Pixels Xi in the image are used as the network input, representing a p×p neighborhood of p×p×N representing spatial context information. c It exists in volume form, where N c Given the number of channels in the input image (3 channels in RGB), output the predicted label of Xi.

[0071] In one alternative embodiment, when training the second network model, the model is adjusted based on the iterative training results to output a non-chemical building element recognition model.

[0072] In an optional embodiment, before training the first network model and the second network model, the remote sensing image features of each pollution risk element were statistically analyzed. This data served as the basis for distinguishing between building elements and non-building elements, as well as the classification rules for reclassifying them into potential pollution sources, migration pathways, and receptors. For example, the results are shown in the table below:

[0073]

[0074] U-Net is a commonly used network algorithm in building segmentation, capable of effectively identifying building outlines. In land cover multi-classification tasks, the Spectral-Spatial-CNN network demonstrates high performance and advantages. Unlike general remote sensing imagery-based feature recognition tasks, chemical plant sites are unique, with numerous and complex man-made features and strong spatial heterogeneity. Using a single deep learning network often fails to achieve high-precision identification of multiple pollution risk factors. Natural features within chemical plant sites are scarce and sparsely distributed, with vegetation and bare land exhibiting unique boundary characteristics—a characteristic of man-made industrial sites. Due to the specific location of man-made landscapes, vegetation within chemical plant sites acts as receptors for potential pollution sources, while bare land and natural water bodies serve as migration pathways for these sources, eliminating the need for distance limitations as in typical scenarios. Furthermore, unlike natural water bodies, man-made reservoirs, due to their production and construction functions, significantly increase the risk of pollution, making them potential pollution sources. Other man-made features are complex, but their geometric characteristics are relatively obvious and their textures are intricate. The office area is the primary recipient of pollution within the site, with high population density and thus posing a significant risk. Other features, besides cooling towers and paved roads, are potential sources of pollution, closely related to the storage and use of chemicals. Due to the complexity of intra-class features—such as open-air production equipment including large heating furnaces and distillation towers, medium-sized reactors and fans, and small machinery, each possessing unique textures and other characteristics, and the significant height and shadow characteristics of chimneys and cooling towers, coupled with the substantial influence of projection angle on remote sensing imagery—identification becomes challenging. However, due to the specificity of chemical building elements and their widespread distribution within chemical enterprise sites, their identification effectiveness plays a decisive role in the final identification of pollution risk factors in chemical enterprises.

[0075] Based on the above analysis, and considering the characteristics of pollution risk factors in chemical enterprises, the method provided in the above embodiments analyzes the features of features within the chemical enterprise site, uses the U-Net network algorithm to segment chemical building elements, and then uses the Spectral-Spatial-CNN network algorithm to identify non-chemical building elements. Different training methods are employed, and the models are integrated to obtain an integrated model for identifying pollution risk factors in chemical enterprises. This model not only efficiently extracts various features within the chemical enterprise site but also effectively addresses the spatial heterogeneity of pollution risk factors, enabling automatic and efficient interpretation of refined pollution risk factors. This meets the need for efficient identification of various pollution risk factors within chemical enterprises and facilitates automatic mapping of the distribution of pollution risk factors.

[0076] This invention provides a device for identifying the distribution of pollution risk factors in chemical enterprises, such as... Figure 4 As shown, it includes:

[0077] The image acquisition module 21 is used to acquire the remote sensing image to be identified. For details, please refer to the description in step S11 above, which will not be repeated here.

[0078] The first identification module 22 is used to input the remote sensing image to be identified into a pre-trained chemical building element segmentation model to obtain the chemical building element identification result. For details, please refer to the description in step S12 above, which will not be repeated here.

[0079] The masking module 23 is used to perform masking on the remote sensing image to be identified using the chemical building element identification results, and to retain non-chemical building elements in the remote sensing image to be identified. For details, please refer to the description in step S13 above, which will not be repeated here.

[0080] The second recognition module 24 is used to input the masked remote sensing image to be recognized into a pre-trained non-chemical building element recognition model to obtain the non-chemical building element recognition result. For details, please refer to the description in step S14 above, which will not be repeated here.

[0081] The risk factor distribution identification module 25 is used to reclassify the pollution risk factors in the remote sensing image to be identified based on the identification results of chemical building elements and non-chemical building elements, using spatial analysis methods. The reclassification results are divided into potential pollution sources, migration pathways and receptors, thereby obtaining the pollution risk factor distribution identification results of chemical enterprises. For details, please refer to the description in step S15 above, which will not be repeated here.

[0082] This invention provides a computer device, such as... Figure 5 As shown, the computer device mainly includes one or more processors 31 and a memory 32. Figure 5 Take a processor 31 as an example.

[0083] The computer device may also include an input device 33 and an output device 34.

[0084] The processor 31, memory 32, input device 33, and output device 34 can be connected via a bus or other means. Figure 5 Taking the example of a connection between China and Israel via a bus.

[0085] Processor 31 can be a Central Processing Unit (CPU). Processor 31 can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or combinations thereof. The general-purpose processor can be a microprocessor or any conventional processor. Memory 32 can include a program storage area and a data storage area. The program storage area can store the operating system and at least one application program required for a function; the data storage area can store data created based on the use of the chemical enterprise pollution risk factor distribution identification device. Furthermore, memory 32 can include high-speed random access memory and non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory 32 may optionally include memory remotely located relative to processor 31, and this remote memory can be connected to the chemical enterprise pollution risk factor distribution identification device via a network. Input device 33 can receive calculation requests (or other numerical or character information) input by the user, as well as generate key signal inputs related to the chemical enterprise pollution risk factor distribution identification device. Output device 34 may include a display screen or other display device for outputting calculation results.

[0086] This invention provides a computer-readable storage medium that stores computer instructions. The computer storage medium stores computer-executable instructions that can execute the method for identifying the distribution of pollution risk factors in chemical enterprises as described in any of the above method embodiments. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc.; the storage medium may also include combinations of the above types of memory.

[0087] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for identifying the distribution of pollution risk factors in chemical enterprises, characterized in that, include: The remote sensing image to be identified is acquired. The pollution risk elements in the remote sensing image to be identified include chemical building elements and non-chemical building elements. Chemical building elements include: office areas and production plants with obvious geometric features and high height; non-chemical building elements include: open-air production equipment, storage tank areas, open-air stockpiled goods, open-air pipelines, chimneys, cooling towers, water storage tanks, hardened roads, bare land, natural water bodies, and vegetation. The remote sensing image to be identified is input into a pre-trained chemical building element segmentation model to obtain the chemical building element identification results. The chemical building element identification results are used to perform masking processing on the remote sensing image to be identified, so that non-chemical building elements are retained in the remote sensing image to be identified. The masked remote sensing image to be identified is input into a pre-trained non-chemical building element recognition model to obtain the non-chemical building element recognition result. Using the identification results of chemical building elements and the identification results of non-chemical building elements, the pollution risk elements in the remote sensing image to be identified are reclassified using spatial analysis method to obtain the distribution identification results of chemical pollution risk elements. After reclassification, the categories to which the pollution risk elements belong include potential pollution sources, migration pathways and receptors.

2. The method for identifying the distribution of pollution risk factors in chemical enterprises according to claim 1, characterized in that, Also includes: The pollution risk factor distribution identification results are fused with the remote sensing image to be identified to obtain a pollution risk factor distribution map of chemical enterprises.

3. The method for identifying the distribution of pollution risk factors in chemical enterprises according to claim 1, characterized in that, The chemical building element segmentation model is trained using the following steps: Acquire remote sensing images of chemical enterprises; A sample set is established based on the remote sensing images of the chemical enterprises. The sample set includes labels for pollution risk elements, and the pollution risk elements in the remote sensing images of the chemical enterprises include chemical building elements. The first network model is trained using the sample set to obtain a chemical building element segmentation model.

4. The method for identifying the distribution of pollution risk factors in chemical enterprises according to claim 3, characterized in that, The first network model is trained using the aforementioned sample set to obtain a chemical building element segmentation model, including: The sample set is enhanced by rotation and / or random flipping to obtain an enhanced dataset; The augmented dataset is input into the first network model, and the first network model is trained to obtain the chemical building element segmentation model.

5. The method for identifying the distribution of pollution risk factors in chemical enterprises according to claim 3 or 4, characterized in that, The first network model is the U-Net network.

6. The method for identifying the distribution of pollution risk factors in chemical enterprises according to claim 1, characterized in that, The non-chemical building element recognition model is trained using the following steps: Acquire remote sensing images of chemical enterprises; A sample set is established based on the remote sensing images of the chemical enterprises. The sample set includes labels for pollution risk elements. The pollution risk elements in the remote sensing images of the chemical enterprises include chemical building elements and non-chemical building elements. The sample set is masked to retain non-chemical building elements in the sample set; The masked sample set is input into the second network model, and the second network model is trained to obtain the non-chemical building element recognition model.

7. A device for identifying the distribution of pollution risk factors in chemical enterprises, characterized in that, include: The image acquisition module is used to acquire remote sensing images to be identified. The pollution risk elements in the remote sensing images to be identified include chemical building elements and non-chemical building elements. Chemical building elements include: office areas and production plants with obvious geometric features and high height; non-chemical building elements include: open-air production equipment, storage tank areas, open-air stockpiled goods, open-air pipelines, chimneys, cooling towers, water storage tanks, hardened roads, bare land, natural water bodies, and vegetation. The first identification module is used to input the remote sensing image to be identified into a pre-trained chemical building element segmentation model to obtain the chemical building element identification result. The masking module is used to perform masking processing on the remote sensing image to be identified using the identification results of the chemical building elements, and to retain non-chemical building elements in the remote sensing image to be identified. The second recognition module is used to input the masked remote sensing image to be recognized into a pre-trained non-chemical building element recognition model to obtain the non-chemical building element recognition result. The risk factor distribution identification module is used to reclassify the elements in the remote sensing image to be identified based on the identification results of the chemical building elements and the identification results of the non-chemical building elements, using spatial analysis methods to obtain the chemical pollution risk factor distribution identification results. After reclassification, the categories to which the pollution risk factors belong include potential pollution sources, migration pathways, and receptors.

8. A computer device, characterized in that, include: At least one processor; The system also includes a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which are executed by the at least one processor to perform the method for identifying the distribution of pollution risk factors in chemical enterprises as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the method for identifying the distribution of pollution risk factors in chemical enterprises as described in any one of claims 1-6.

Images