A recommended method and device for screening and classified supervision of soil pollution risk enterprises

Abstract

The present invention discloses a method and device for recommending the screening and classified supervision of soil pollution risk enterprises, which relates to the technical field of enterprise soil pollution risk control and supervision. Starting from a full-caliber enterprise list, a list of key industry enterprises that need to be paid attention to is screened and integrated according to specific rules as a screening base; multi-source data such as enterprise production characteristics and environmental distribution characteristics are gathered, and efficient data processing technologies such as knowledge graphs, large language models and spatial analysis are integrated to accurately focus on enterprises that are prone to soil pollution from the full-caliber enterprise list, further pay attention to the impact of pollution spread on the surrounding environment, distinguish enterprise risk types in combination with geographical environmental conditions of pollution spread, automatically calculate and recommend to form a classified supervision list, support relevant management personnel to supplement and improve the supervision list, and promote subsequent risk control and restoration activities.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise soil pollution risk control and supervision, and in particular to a recommended method and device for screening and classified supervision of soil pollution risk enterprises. Background Art

[0002] The emission of toxic and hazardous substances in the production and operation activities of industrial enterprises can easily lead to soil and groundwater pollution. If there is no supervision and risk control, and the pollution is allowed to migrate and spread, it will affect the safety of human settlements and the ecological environment. It is necessary to promptly check the plots with pollution risks and carry out targeted supervision. Relevant management departments are trying various ways to carry out pollution risk investigation and dynamically update the supervision list, such as the list of key soil pollution supervision units and the list of priority supervision plots.

[0003] Soil and groundwater pollution has significant temporal and spatial heterogeneity. Pollution risk investigation requires multi-dimensional data information as a basic support. However, the early management of enterprise land archives is not standardized, basic data and historical information are lacking, and a lot of manpower and material resources are needed for data collection and analysis. The investigation efficiency is low and there is a possibility of omissions. How to quickly and effectively identify enterprises with pollution risks and carry out targeted classification management has become a technical problem that needs to be solved urgently.

[0004] Existing technologies use basic enterprise data, such as company names, industry categories, etc., to conduct preliminary identification of suspected polluted enterprise plots through black-box algorithms such as machine learning. However, the basic enterprise data is not strongly correlated with the indication of soil and groundwater pollution risks of the plots, and key features such as enterprise production characteristics and characteristic pollutants are easily missed. At the same time, it is also impossible to specifically distinguish risk types, affecting the accuracy of identification results and the precision of classification management. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a recommended method and device for screening and classified supervision of soil pollution risk enterprises. The following technical solutions are adopted:

[0006] A recommended method for screening and classified supervision of soil pollution risk enterprises includes the following steps: Step 1, starting from the full-caliber enterprise list, screening and integrating the list of key industry enterprises that need to be paid attention to according to specific rules as the screening base;

[0007] Step 2: Based on the availability and convenience of data, multi-source data on enterprise production characteristics and environmental distribution characteristics are gathered to screen enterprise basic attribute indicators with high correlation indicating the risk of soil and groundwater pollution in plots. The enterprise basic attribute indicators include enterprise production scale, production and operation activity time, three wastes emission management level, enterprise main product information, ground anti-seepage level of enterprise location, soil permeability, groundwater depth and surrounding sensitive point density;

[0008] Step 3: Risk screening focuses on the production characteristics of enterprises. By establishing a knowledge graph, it is possible to infer whether the enterprise is involved in volatile and migratory pollutants from known information. In the inference process, a large language model is combined to achieve efficient data identification and batch matching, and enterprises involved in volatile and migratory pollutants are calculated and screened out;

[0009] Risk screening focuses primarily on the production characteristics of enterprises, screening out enterprises involved in the production, use, storage, disposal or discharge of toxic and hazardous substances, with special attention to volatile and migratory pollutants among toxic and hazardous substances. However, much direct information obtained about enterprises involved in volatile and migratory pollutants is missing, but there is a logical chain of association between the production process, raw materials and products, and pollutants of enterprises. By establishing a knowledge graph, it is possible to infer whether the enterprise is involved in volatile and migratory pollutants from known information; in the inference process, a large language model is further combined to achieve efficient data identification and batch matching, and enterprises involved in volatile and migratory pollutants are calculated and screened out;

[0010] Step 4: Determine the possibility of pollutants from enterprises entering the soil. Based on the pollution characteristic statistics of regulated enterprises and the characteristic thresholds of various pollution indicators, select enterprises that are prone to soil pollution from those involving volatile and migratory pollutants;

[0011] Step 5: Distinguish the degree of impact of enterprise pollution migration and diffusion on the surrounding environment again. Based on the hydrogeological conditions of pollution diffusion and the surrounding environmental distribution characteristic data, and combining prior knowledge to set characteristic thresholds, select enterprises that are prone to pollution diffusion, enterprises with high sensitivity in the surrounding areas, and enterprises that are subject to key risk screening from enterprises that are prone to soil pollution, and form a classified supervision list; and identify key areas of concern based on the spatial distribution characteristics of various types of enterprises.

[0012] Optionally, the specific method of step 1 is: batch obtain full-caliber yellow page enterprise and industrial and commercial enterprise data, form a list of key industry enterprises according to the screening rules of key soil pollution industries, obtain the spatial coordinates of enterprises based on the map service API interface, and merge duplicate and redundant enterprises;

[0013] Compare the list of enterprises that have been included in supervision by the management department, mark the supervision status of each enterprise, and distinguish between those that have been included in supervision and those that have not.

[0014] Optionally, in step 2, the basic attribute indicators of the enterprise are standardized by the following methods: for production scale, numerical grading is performed according to the number of employees and annual turnover distribution of the enterprise; for production and operation activity time, the interval between the registration and closure of the enterprise is calculated; for the three wastes emission management level, the distribution density of neighboring regulated enterprises is calculated; for the main product information of the enterprise, batch structured recognition is performed through a large language model;

[0015] The enterprise spatial data indicators are standardized using the following method: the enterprise distribution layer is superimposed with the surface impermeable surface layer, soil layer property layer, groundwater depth layer, and sensitive point distribution layer after spatial information coordinate conversion, and the spatial characteristic data corresponding to the enterprise location are extracted, including the ground anti-seepage ratio, soil permeability, groundwater depth, and surrounding sensitive point density.

[0016] Optionally, in step 4, based on knowledge graph reasoning, the probability of the relationship between the company's products and volatile and migratory pollutants is calculated. When the product-related data of the company to be evaluated is input, the probability of the company being involved in volatile and migratory pollutants is output immediately.

[0017] Optionally, the specific method for calculating the probability of the relationship between the enterprise's products and the volatile and migratory pollutants is: based on the purpose of knowledge reasoning, combining the expert industry knowledge of the enterprise's production, pollution production and emission characteristics and pollutant attributes to define the knowledge graph ontology, including entities, entity attributes and relationships between entities; designing a triple structure data table to store the above entities, entity attributes and relationships between entities;

[0018] Corresponding to entities, entity attributes, and relationship fields between entities, using the pollution discharge permit registration data of representative sample enterprises, extract entity data and its attributes and relationships one by one, store them in a triple structure data table, and carry out data cleaning and integration to achieve mapping to the knowledge graph ontology;

[0019] Analyze the relationship probability between entities to assign weights to key node relationships, determine the connection condition probability of each edge in the knowledge graph, and fill this conditional probability into the relationship attribute of the relationship table in the triple table;

[0020] A visual graph database is established based on the populated triple structure data table, and the graph database is used for storage; when new enterprise triple data is loaded and updated, the entities, entity attributes, entity relationships, and corresponding entity relationship conditional probabilities in the relevant triple structure data table are also updated;

[0021] Based on the conditional probability between entity relationships in the graph database, the probability of the enterprise to be evaluated involving volatile and migrating pollutants is calculated according to the inference formula; if the product-related data of the enterprise to be evaluated is semi-structured, the large language model is called to perform batch structured recognition of product entities, and batch match and reasoning calculation are performed with the corresponding entities in the existing knowledge graph library;

[0022] Set a probability threshold for the inference results and screen out enterprises involved in volatile and migrating pollutants from key industry enterprises.

[0023] Optionally, the comprehensive probability of volatile and migratory pollutants generated by the product can be calculated using the following formula:

[0024]

[0025] Where i is the path number, j is the number of the edge on the i path, and m is the number of edges on path i. If the product of the enterprise to be evaluated, the production process or raw and auxiliary materials information used by the product are available at the same time, when calculating the above comprehensive probability, only the input edge weight of the production process or raw and auxiliary materials needs to be set to 1, and the input edge weights of the remaining production processes or raw and auxiliary materials need to be set to 0.

[0026] Optionally, in step 5, the pollution index characteristics of the regulated soil polluting enterprises are statistically analyzed, and the pollution index data of the enterprises to be evaluated are screened accordingly, so that each indicator value is not lower than the lower quartile value of the corresponding indicator of the regulated enterprises, and the enterprises that are prone to soil pollution are comprehensively screened out.

[0027] Optionally, in step 5, indicator thresholds are set hierarchically according to expert prior knowledge, and based on pollution diffusion conditions, enterprises with underground soil permeability higher than a specific value and groundwater depth lower than a specific value are screened; based on the impact of pollution on the surrounding environment, enterprises with a density of sensitive points around their locations higher than a specific value are screened, thereby achieving rapid screening of specific types of enterprises.

[0028] A recommendation device for the screening and classification supervision of soil pollution risk enterprises, comprising an integrated data loading and storage module, an attribute data processing module, a spatial data processing module, a knowledge graph reasoning module, a pollution feature statistics module, a classification screening and list recommendation module and a spatial display module, for implementing the method described in any one of claims 1 to 6 when executing the calculation program;

[0029] The data loading and storage module is used to classify, mark and store the acquired multi-source data, and supports data import and update;

[0030] The attribute data processing module is used to integrate and clean inventory attribute data, perform standardized processing according to rules, and support batch export;

[0031] The spatial data processing module is used to use ArcGIS mapping analysis software to perform standardized processing according to rules and support batch export;

[0032] The knowledge graph reasoning module is used to load the populated triple structure data table, establish a graph database and support data update; call the large language model to perform batch structured processing on the main product information of the evaluated enterprise, calculate the probability of its involvement in volatile and migratory pollutants according to the rules, and screen and export enterprises involved in volatile and migratory pollutants after setting the probability threshold;

[0033] The pollution characteristic statistics module is used to automatically identify and count the relevant pollution index data of regulated enterprises and provide pollution index screening thresholds;

[0034] The hierarchical screening and list recommendation module is used to automatically export the recommended list according to the default indicator screening thresholds, and supports manual update of each indicator screening threshold;

[0035] The spatial display module is used to classify and display the spatial distribution of various recommended enterprises, distinguishing them by color and legend markings, and supporting the identification of key areas of focus.

[0036] Optionally, the integrated data loading and storage module, attribute data processing module, spatial data processing module, knowledge graph reasoning module, pollution characteristic statistics module, classification screening and inventory recommendation module and spatial display module are implemented based on a computer server.

[0037] In summary, the present invention includes at least one of the following beneficial technical effects:

[0038] The present invention can provide a recommended method and device for the investigation and classified supervision of enterprises with soil pollution risks, which gathers multi-source data such as enterprise production characteristics and environmental distribution characteristics, integrates efficient data processing technologies such as knowledge graphs, large language models and spatial analysis, accurately focuses on enterprises that are prone to soil pollution from a full-caliber enterprise list, further pays attention to the impact of pollution spread on the surrounding environment, distinguishes enterprise risk types in combination with the geographical environmental conditions of pollution spread, automatically calculates and recommends a classified supervision list, supports relevant managers to supplement and improve the supervision list, and promotes subsequent risk control and remediation activities. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flow chart of a recommended method for screening and classified supervision of soil pollution risk enterprises of the present invention;

[0040] Figure 2 It is a schematic diagram of the process of integrating the list of key soil pollution industries and standardizing multi-source data in the present invention;

[0041] Figure 3 It is a schematic diagram of the process of establishing the knowledge graph of enterprise production and pollution identification and knowledge reasoning of the present invention;

[0042] Figure 4 It is a schematic diagram of the knowledge graph ontology structure of the present invention;

[0043] Figure 5 It is a schematic diagram of a recommended device for screening and classified supervision of soil pollution risk enterprises according to the present invention;

[0044] Figure 6 It is a spatial classification display effect diagram specifically implemented by the present invention. DETAILED DESCRIPTION

[0045] The present invention is further described in detail below in conjunction with the accompanying drawings.

[0046] The embodiment of the present invention discloses a recommended method and device for screening and classified supervision of soil pollution risk enterprises.

[0047] Reference Figure 1-Figure 6 First, starting from the full-caliber enterprise list, we will screen and integrate the list of key industry enterprises that need attention according to specific rules as the basis for investigation.

[0048] Taking a certain city as an example, we obtained more than 60,000 yellow pages companies and more than 400,000 industrial and commercial enterprise list data in batches, cleaned the data and integrated the fields, and eliminated non-key industry companies according to the key industry screening rules in Appendix B of the Technical Guidelines for the Investigation of Soil Pollution Status of Construction Land (HJ 25.1-2019) to form a list of key industry companies; based on the map service API interface, we obtained the spatial coordinates of the companies and merged the companies with duplicate addresses. After screening and integration, we obtained more than 10,000 key industry companies in the city as the base number for this investigation.

[0049] The comprehensive enterprise name, spatial coordinates, and industry category information are compared with the list of enterprises that are already supervised by the management department, and the supervision status of each enterprise in the list is marked to distinguish between supervised and unsupervised enterprises.

[0050] Secondly, based on the availability and convenience of data, we screened out indicators with high correlation to indicate the risk of soil and groundwater pollution in plots, including (1) basic attribute data: the industry category to which the enterprise belongs, production scale, production and operation activity time, "three wastes" emission management level, and information on the enterprise's main products; (2) spatial data: the ground anti-seepage level of the enterprise's location, soil permeability, groundwater depth, and density of surrounding sensitive points.

[0051] In the embodiment, the corresponding indicator data in the list of key industry enterprises in the city are batch collected and standardized.

[0052] In the basic attribute data, for the industry category indicators, the industry category data in the list are uniformly processed in accordance with the standard format of the National Economic Industry Classification (GB / T4754-2017); for the production scale indicators, the number of employees and annual turnover of the enterprise are comprehensively counted, and the data are divided into 13 levels from 1 to 13, and the highest level of the two types of data is taken as the standardized data, which is a discrete numerical field; for the production and operation activity time indicators, the annual interval from the registration to the closure of the enterprise is calculated as the standardized data, which is a discrete numerical field; for the "three wastes" emission management level indicators, ArcGIS is used to conduct kernel density analysis on the supervised plots, and the kernel density values ​​of the supervised plots at the locations of each enterprise are used as standardized data, which is a continuous numerical field; for the information on the main products of the enterprise, batch structured recognition is performed through a large language model (detailed in the third aspect).

[0053] In the spatial data, for the ground anti-seepage level, soil permeability, groundwater depth, and surrounding sensitive point density indicators, the impervious surface raster data (accuracy 10m), soil property raster data (1:100,000), groundwater depth point data, and human settlement sensitive point POI data (schools, residential areas, hospitals, public management and public service places) of the city are collected in batches on the corresponding public source data website. The enterprise distribution layer is superimposed on the above layers after spatial information coordinate conversion, and the spatial feature data of the corresponding enterprise location are extracted, including: for the ground anti-seepage level, the ground anti-seepage ratio corresponding to each enterprise distribution point is classified as 1 or 0, which is a discrete numerical field; for the soil permeability, the distribution point of each enterprise is The corresponding soil layer property classification (1 heavy clay, 2 silty clay, 3 clay, 4 silty clay loam, 5 clay loam, 6 silt, 7 silt loam, 8 sandy clay, 9 loam, 10 sandy clay loam, 11 sandy loam, 12 loamy sand, 13 sand) is used as standardized attribute data, which is a discrete numerical field; for the groundwater depth, ArcGIS inverse distance interpolation is used to first rasterize the point data, and the depth value corresponding to each enterprise distribution point is used as the standardized attribute data, which is a continuous numerical field; for the density of surrounding sensitive points, ArcGIS is used to perform kernel density analysis on the distribution of sensitive points, and the density of sensitive points corresponding to each enterprise distribution point is normalized as standardized attribute data, which is a continuous numerical field.

[0054] On the third aspect, risk investigation should first focus on the production characteristics of enterprises, screen out enterprises involved in the production, use, storage, disposal or discharge of toxic and hazardous substances, and pay special attention to volatile and migratory pollutants among toxic and hazardous substances. For the fields in the enterprise list after cleaning and integration, the data coverage varies. For example, the enterprise product information is relatively comprehensive, while the information on the enterprise's volatile and migratory pollutants is relatively missing, which affects the accuracy of enterprise screening. However, there is a logical chain of association between the enterprise's production process, raw materials and products, and the fields involving volatile and migratory pollutants. Through complex relationship deductive reasoning, the probability of the relationship between the enterprise's products and the volatile and migratory pollutants can be sorted out. Based on the above foundation, the present invention establishes a method for reasoning whether an enterprise is involved in volatile and migratory pollutants based on a knowledge graph.

[0055] The general framework is that, based on the purpose of this knowledge graph deductive reasoning, a top-down approach is used to first define the knowledge graph ontology, then map the entities, entity attributes, and relationships between entities in the enterprise sample data to the defined ontology, and assign weights to key node relationships based on the parsed probability of relationships between sample entities, construct a knowledge graph, and on this basis calculate and infer the probability of the enterprise to be evaluated involving volatile and migrating pollutants. The specific implementation process includes:

[0056] First, based on expert and industry knowledge such as enterprise production, pollution emission characteristics, and pollutant attributes, the knowledge graph ontology is defined, including entities, entity attributes, and relationships between entities. Figure 4 As shown, the entities include eight categories: enterprise, product, industry, production process, raw and auxiliary materials, intermediate products, toxic and harmful pollutants, volatile and migratory pollutants, which are represented by nodes in the knowledge graph; entity attributes refer to the properties of the entity itself, that is, node attributes, such as the attributes of raw and auxiliary materials can include pollutants; the relationship between entities refers to the mapping relationship between entities, and the relationship between entity nodes is represented by edges in the knowledge graph, such as the production process node and the intermediate product node are connected by the "produce" edge.

[0057] A triple structure data table is set to store the above entities, entity attributes and relationships between entities, and each row label is used to reflect the relationship between the entities. In the embodiment, a triple structure data table (Table 1) between entity 1 enterprise and entity 2 product is given, including the enterprise name, 3 products produced by the enterprise and the main attributes of the product; in another embodiment, a triple structure data table (Table 2) between the same enterprise entity 3 production process and entity 4 raw and auxiliary materials is given, including 3 types of production processes involved in the enterprise, raw and auxiliary materials corresponding to each production process and their pollutant attributes. In this example, there is a one-to-many relationship between the entities, and all are one-to-one correspondence through row labels; the triple structure data tables between other entities are similar to the above structure, see Table 1 and Table 2.

[0058] Table 1 Triple structure data table example 1

[0059]

[0060] Table 2 Triple structure data table example 2

[0061]

[0062]

[0063] Secondly, based on the entities and entity attributes as well as the relationships between entities in the ontology, Python scripts are used to directly extract entity data, their attributes and relationships from the pollutant discharge permit registration data of representative sample enterprises in corresponding fields one by one, and store them in a triple structure data table to achieve mapping to the defined ontology.

[0064] In the embodiment, the pollution discharge permit registration forms of 50 representative enterprises in the metal surface treatment and heat treatment industry are selected, and the entity data of enterprises, products, production processes, raw materials, intermediate products and their attributes and corresponding relationships are extracted one by one using Python scripts, and automatically filled into the corresponding triple structure data table; for the toxic and hazardous substance entities and the volatile and migrating pollutant entity attributes, the pollutant property parameter dictionary library that is currently more mature in related industries is automatically loaded, and the style can refer to Appendix B of HJ25.3.

[0065] The automatically extracted and filled data of each enterprise may contain duplication, redundancy, errors or relationship contradictions. In order to ensure the accuracy of the representative sample analysis data, manual review and correction are carried out on data such as "product contaminant attributes", "raw material and auxiliary material contaminant attributes" and "intermediate product contaminant attributes". For example, pollutant attributes with synonyms or near-synonyms in the same entity are eliminated, thereby completing data cleaning and integration to form a relatively standardized and unified structural data table, so as to improve the accuracy of subsequent relationship weighting and probability reasoning.

[0066] Thirdly, the probability of the relationship between entities is analyzed to assign weights to the key node relationships, the connection condition probability of each edge in the knowledge graph is determined, and this conditional probability is filled into the relationship attribute of the relationship table in the triple table. The present invention mainly targets three types of entity relationships: product and production process, production process and raw and auxiliary materials, and production process and intermediate products. Based on the traversal retrieval statistics of the frequency of co-occurrence between two entities, the probability of each connection condition is determined by its proportion in the total sample; the experience of industry experts can be further introduced to correct the statistical conditional probability. For example, through traversal retrieval statistics, it is concluded that the conditional probability of the product "nut" and the production process "galvanizing" is 81%, and the conditional probability of the production process "nickel plating" is 16%, as well as other production processes "chrome plating" and "spray painting".

[0067] Finally, a visual graph database is established based on the filled triple structure data table (including the above-calculated relationship conditional probability attributes). The graph database neo4j is generally used for storage to facilitate query and quality assessment. For example, the product "nuts" is queried in the graph database, and the relationship indexes all the production processes, raw materials, intermediates, toxic and hazardous substances, volatile and easily migrating pollutants and conditional probabilities involved in its production; this can be used to verify whether the queried relationship conforms to the defined ontology structure or whether there is a relationship missing. If so, return to modify the original triple structure data table. At the same time, when the representative enterprise sample is further expanded, the entities, entity attributes, relationships between entities, and the corresponding entity relationship conditional probabilities in the triple structure data table can be updated respectively.

[0068] Next, based on the conditional probability between entity relationships in the graph database, the probability of the enterprise to be evaluated involving volatile and mobile pollutants is inferred. Specifically, a traversal search is performed in the graph database to find all paths (assuming there are n paths) whose starting point is the product entity of the enterprise to be evaluated and whose end point is the volatile and mobile pollutant entity, and the weights of the adjacent node edges on each path are retrieved (i is the path number, j is the number of the edge on path i), and the comprehensive probability of volatile and mobile pollutants generated by the product is obtained by the following formula (m is the number of edges on path i):

[0069]

[0070] If you have information on the products of the company to be evaluated and the production process or raw materials used in the product, when calculating the above-mentioned comprehensive probability, you only need to set the input edge weight of the production process or raw materials to 1, and set the input edge weights of the remaining production processes or raw materials to 0.

[0071] When an enterprise involves multiple products, the maximum value of the probability of volatile and migratory pollutants involved in each product is taken; when information on non-main product attributes is involved, the probability is converted by 20%, as shown in Table 3.

[0072] Table 3 Probability results of enterprises involving volatile and migratory pollutants

[0073]

[0074]

[0075] Finally, based on the above reasoning rules, the probability of volatile and migratory pollutants involved in the batch of enterprises to be evaluated summarized in the second aspect is calculated. If the product-related data of the enterprise to be evaluated is semi-structured, the Python script is used to call the large language model to batch identify its products, raw materials, and production process entities, and batch match them with the corresponding entities in the existing knowledge graph library.

[0076] In a specific embodiment, the large language model prompt is set as follows:

[0077] You are an assistant who extracts the main products and business scope of an enterprise. Now you need to extract the information of "product", "production process" and "raw and auxiliary materials" from the original text: (1) "Product" refers to the objects produced by the enterprise for sale, such as "leather gloves" and "metal products"; (2) "Production process" refers to the process required to produce the product, such as "galvanizing"; (3) "Raw and auxiliary materials" refer to the raw materials and auxiliary materials needed to produce the product.

[0078] Please extract the above information from the following information. If not, return "None" and return it in JSON format:

[0079] <Original text information to be entered>.

[0080] <Original text information to be input> is replaced with the specific text information to be extracted, such as metal parts and galvanizing, and the returned structured results are as follows:

[0081] {"product":"metal parts",

[0082] "Production process": "Galvanizing",

[0083] "Raw materials": "None"}

[0084] Semantically match the identified entities with the entities in the existing knowledge graph library. For example, the above results are matched with the product entity "metal products" and the production process "galvanizing" in the knowledge graph library, and then the probability of the enterprise being involved in volatile and migrating pollutants is obtained according to the inference formula.

[0085] In the embodiment, Python scripts are used to batch import semi-structured data such as the main products and business scope of the enterprises to be evaluated summarized in the second aspect. After structural processing and entity matching by the large language model, batch reasoning is used to calculate the probability of each enterprise involving easily migrating pollutants. Enterprises with a probability of not less than 0.9 are selected as enterprises involving easily volatile and easily migrating pollutants, and the city involves more than 2,000 enterprises.

[0086] Fourthly, statistical analysis is conducted on the pollution index characteristics of regulated soil polluting enterprises, and the pollution index data of the enterprises to be evaluated are screened accordingly, so that the value of each indicator is not lower than the lower quartile value of the corresponding indicator of the regulated enterprises, and enterprises that are prone to soil pollution are comprehensively screened out.

[0087] In the embodiment, 150 samples of regulated and soil polluting enterprises are selected from more than 2,000 enterprises involved in volatile and migratory pollutants, and the lower quartile values ​​of the index data of their production and operation activity time, production scale, "three wastes" emission management level, and ground anti-seepage level are counted respectively. Based on this, data screening is performed on the indicators of the enterprises to be evaluated, so that each indicator value is not lower than the lower quartile value of the corresponding indicator of the regulated enterprises, namely, the production and operation activity time is not less than 17 years, the production scale classification is not less than 10, and the "three wastes" emission management level is not less than 0.1. In this way, more than 100 enterprises to be included in the supervision that are prone to soil pollution are screened (excluding those already included in the supervision).

[0088] All enterprises that have been brought under supervision in the city are verified according to the data limit of this indicator, and the enterprises screened out are all soil polluting enterprises. This method is based on the screening of pollution characteristics of existing enterprises in the city. The pollution characteristics of enterprises in different cities are different, and their data screening values ​​can be different; different screening quantile thresholds can be selected in combination with the supervision intensity of different cities, such as decile, median, etc.

[0089] Fifthly, the conditions for the diffusion of soil pollution in enterprises and the possibility of its impact on the surrounding areas are analyzed, and indicator thresholds are set in layers according to the expert prior knowledge to screen out enterprises with good pollution migration and diffusion conditions and high surrounding sensitivity.

[0090] In the embodiment, among the enterprises that are prone to soil pollution, based on the pollution migration and diffusion conditions, the enterprises (more than 20) whose underground soil permeability grade is not less than 11 and whose groundwater depth is less than 6 are screened and marked as enterprises of concern for pollution diffusion; based on the surrounding sensitivity, the enterprises (more than 10) whose relative density of surrounding sensitive points is higher than 0.6 are screened and marked as surrounding sensitive enterprises; the enterprises (several) that meet both screening conditions are marked as key risk investigation enterprises.

[0091] Based on this rule, enterprises with specific risk types can be quickly screened. Different indicator thresholds can be set according to the hydrogeological conditions and regulatory strength of different cities, such as groundwater depth less than 2.

[0092] The sixth aspect is to build an automatic recommendation computing device, which includes a data loading and storage module, an attribute data processing module, a spatial data processing module, a knowledge graph reasoning module, a pollution feature statistics module, a classification screening and list recommendation module, and a spatial display module.

[0093] In the embodiment, the data loading and storage module imports the batch-collected full-caliber enterprise list, the supervised enterprise list, the enterprise spatial coordinates, the soil layer property grid data, the groundwater depth point data, the sensitive point POI data and other source data, classifies and marks the storage, and can import and replace multiple times according to the data update situation;

[0094] In the attribute data processing module, click on list integration, and automatically calculate according to the integration rules described in the first aspect to unify the enterprise list, and manually correct the duplicate enterprises that cannot be automatically identified. Click on attribute data cleaning, and according to the attribute data cleaning rules described in the second aspect, identify the corresponding field data in the data loading and storage module, carry out automatic standard structured processing, and store it in the enterprise list field table; manual batch correction can be performed on the data reported in the automatic processing errors.

[0095] In the spatial data processing module, according to the spatial data processing rules described in the second aspect, the corresponding spatial layer data in the data loading and storage module is identified, the ArcGIS mapping analysis software is set to be called, the analysis rules are embedded in the software interface, the corresponding spatial feature data of the corresponding enterprise location is automatically extracted, and the matching is written into the enterprise list field table.

[0096] In the knowledge graph reasoning module, the filled triple structure data table described in the third aspect is loaded to establish a visual graph database; support the loading and updating of new enterprise triple data, and the entities, entity attributes, entity relationships and corresponding entity relationship conditional probabilities in the relevant triple structure data table are updated immediately. Automatically identify the main product fields of enterprises in the data loading and storage module, support calling the large language model for data structuring and entity matching; embed the knowledge reasoning formula described in the third aspect, batch calculate and derive the probability of each enterprise involving volatile and migrating pollutants, and filter and derive enterprises involving volatile and migrating pollutants after setting the probability threshold.

[0097] In the pollution characteristic statistics module, according to the soil pollution enterprise characteristic statistics rules described in the fourth aspect, the enterprise list fields in the data loading and storage module are automatically identified, and the corresponding pollution index screening thresholds are calculated. At the same time, manual changes to the pollution index screening thresholds are supported in the interface; after confirming the thresholds, the screened-out enterprises that are prone to soil pollution and are to be included in the supervision are selected.

[0098] In the classification screening and list recommendation module, according to the description in the fifth aspect, the threshold for manually filling in the indicator screening is set in the interface, and the enterprise list fields in the data loading and storage module are automatically identified. Enterprises of concern for pollution spread, enterprises of concern for surrounding sensitive areas, and enterprises for key risk investigation are exported based on the threshold screening.

[0099] In the spatial display module, set up the ArcGIS mapping analysis software to automatically access the coordinates of the aforementioned multi-level screening recommended enterprises, distinguish them by color and legend marks, and display the spatial distribution of enterprises in each recommended list. Figure 6 , supporting visual identification of key focus areas.

[0100] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A recommended method for screening and classifying soil pollution risk enterprises. Features: The following steps are included: Step 1: Starting from the full-caliber enterprise list, screen and integrate the list of key industry enterprises that need attention according to specific rules as the base number for investigation; Step 2: Based on the availability and convenience of data, multi-source data on enterprise production characteristics and environmental distribution characteristics are gathered to screen enterprise basic attribute indicators with high correlation indicating the risk of soil and groundwater pollution in plots. The enterprise basic attribute indicators include enterprise production scale, production and operation activity time, three wastes emission management level, enterprise main product information, ground anti-seepage level of enterprise location, soil permeability, groundwater depth and surrounding sensitive point density; Step 3: Risk screening first focuses on the production characteristics of the enterprise. By establishing a knowledge graph, it can be inferred from known information whether the enterprise is involved in volatile and migratory pollutants. In the reasoning process, a large language model is combined to achieve efficient data identification and batch matching, and enterprises involved in volatile and migratory pollutants are calculated and screened out; Step 4: Determine the possibility of pollutants from enterprises entering the soil. Based on the pollution characteristic statistics of regulated enterprises and the characteristic thresholds of various pollution indicators, select enterprises that are prone to soil pollution from those involving volatile and migratory pollutants; Step 5: Distinguish the degree of impact of enterprise pollution migration and diffusion on the surrounding environment again, set characteristic thresholds based on the hydrogeological conditions of pollution diffusion and the distribution characteristics of the surrounding environment, and combine prior knowledge to select enterprises that are prone to soil pollution, enterprises that are highly sensitive to pollution, and enterprises that are subject to key risk screening, from enterprises that are prone to soil pollution, to form a classified supervision list; and identify key areas of concern based on the spatial distribution characteristics of various types of enterprises; In step 4, based on knowledge graph reasoning, the probability of the relationship between the company's products and the volatile and migratory pollutants is calculated. When the product association data of the company to be evaluated is input, the probability of the company's involvement in volatile and migratory pollutants is immediately output; The specific method for calculating the probability of the relationship between enterprise products and volatile and migratory pollutants is: based on the purpose of knowledge reasoning, combining the expert industry knowledge of enterprise production, pollution production and emission characteristics and pollutant attributes to define the knowledge graph ontology, including entities, entity attributes and relationships between entities; designing a triple structure data table to store the above entities, entity attributes and relationships between entities; Corresponding to entities, entity attributes, and relationship fields between entities, using the pollution discharge permit registration data of representative sample enterprises, extract entity data and its attributes and relationships one by one, store them in a triple structure data table, and carry out data cleaning and integration to achieve mapping to the knowledge graph ontology; Analyze the relationship probability between entities to assign weights to key node relationships, determine the connection condition probability of each edge in the knowledge graph, and fill this conditional probability into the relationship attribute of the relationship table in the triple table; A visual graph database is established based on the populated triple structure data table, and the graph database is used for storage; when new enterprise triple data is loaded and updated, the entities, entity attributes, entity relationships, and corresponding entity relationship conditional probabilities in the relevant triple structure data table are also updated; Based on the conditional probability between entity relationships in the graph database, the probability of the enterprise to be evaluated involving volatile and migrating pollutants is calculated according to the inference formula; if the product-related data of the enterprise to be evaluated is semi-structured, the large language model is called to perform batch structured recognition of product entities, and batch match and reasoning calculation are performed with the corresponding entities in the existing knowledge graph library; Set probability thresholds for the inference results and screen out enterprises involved in volatile and migratory pollutants from key industry enterprises; The comprehensive probability of volatile and migratory pollutants generated by the product is calculated using the following formula: Where i is the path number, j is the number of the edge on the i path, and m is the number of edges on path i. If the product of the enterprise to be evaluated, the production process or raw and auxiliary materials information used by the product are available at the same time, when calculating the above comprehensive probability, only the input edge weight of the production process or raw and auxiliary materials needs to be set to 1, and the input edge weights of the remaining production processes or raw and auxiliary materials need to be set to 0.

2. According to claim 1, a recommended method for screening and classifying and supervising soil pollution risk enterprises, Features: The specific method of step 1 is: obtain the full-caliber yellow page enterprise and industrial and commercial enterprise data in batches, form a list of key industry enterprises according to the screening rules of key soil pollution industries, obtain the spatial coordinates of enterprises based on the map service API interface, and merge duplicate and redundant enterprises; Compare the list of enterprises that have been included in supervision by the management department, mark the supervision status of each enterprise, and distinguish between those that have been included in supervision and those that have not.

3. According to claim 1, a recommended method for screening and classifying and supervising soil pollution risk enterprises, Features: In step 2, the basic attribute indicators of enterprises are standardized using the following methods: for production scale, numerical grading is performed according to the number of employees and annual turnover distribution of the enterprise; for production and operation activity time, the annual interval from enterprise registration to closure is calculated; For the three wastes emission management level, calculate the distribution density of nearby regulated enterprises; for the main product information of enterprises, perform batch structured recognition through large language models; The enterprise spatial data indicators are standardized using the following method: the enterprise distribution layer is superimposed with the surface impermeable surface layer, soil layer property layer, groundwater depth layer, and sensitive point distribution layer after spatial information coordinate conversion, and the spatial characteristic data corresponding to the enterprise location are extracted, including the ground anti-seepage ratio, soil permeability, groundwater depth, and surrounding sensitive point density.

4. According to claim 1, a recommended method for screening and classifying and supervising soil pollution risk enterprises, Features: In step 5, the pollution index characteristics of the regulated soil pollution enterprises are statistically analyzed, and the pollution index data of the enterprises to be evaluated are screened accordingly, so that each index value is not lower than the lower quartile value of the corresponding index of the regulated enterprises, and the enterprises that are prone to soil pollution are comprehensively screened out.

5. According to claim 1, a recommended method for screening and classifying and supervising soil pollution risk enterprises, Features: In step 5, the indicator thresholds are set hierarchically according to the expert prior knowledge. Based on the pollution diffusion conditions, enterprises with underground soil permeability higher than a specific value and groundwater depth lower than a specific value are screened. Based on the impact of pollution on the surrounding environment, enterprises with a density of sensitive points around their locations higher than a specific value are screened to achieve rapid screening of specific types of enterprises.

6. A recommended device for screening and classified supervision of soil pollution risk enterprises, Features: It includes an integrated data loading and storage module, an attribute data processing module, a spatial data processing module, a knowledge graph reasoning module, a pollution feature statistics module, a classification screening and inventory recommendation module and a spatial display module, which are used to implement the method described in any one of claims 1 to 5 when executing a computing program; The data loading and storage module is used to classify, mark and store the acquired multi-source data, and supports data import and update; The attribute data processing module is used to integrate and clean inventory attribute data, perform standardized processing according to rules, and support batch export; The spatial data processing module is used to use ArcGIS mapping analysis software to perform standardized processing according to rules and support batch export; The knowledge graph reasoning module is used to load the populated triple structure data table, build a graph database and support data updates; The large language model is used to perform batch structural processing on the main product information of the enterprise to be evaluated, and the probability of its involvement in volatile and migratory pollutants is calculated according to the rules. After setting the probability threshold, the enterprises involved in volatile and migratory pollutants are screened out; The pollution characteristic statistics module is used to automatically identify and count the relevant pollution index data of regulated enterprises and provide pollution index screening thresholds; The hierarchical screening and list recommendation module is used to automatically export the recommended list according to the default indicator screening thresholds, and supports manual update of each indicator screening threshold; The spatial display module is used to classify and display the spatial distribution of various recommended enterprises, distinguishing them by color and legend markings, and supporting the identification of key areas of focus.

7. A recommendation device for screening and classified supervision of soil pollution risk enterprises according to claim 6, Features: The integrated data loading and storage module, attribute data processing module, spatial data processing module, knowledge graph reasoning module, pollution characteristic statistics module, classification screening and inventory recommendation module and spatial display module are implemented based on a computer server.

Images