Method and system for monitoring compliance of solid waste discharge information in a discharge permit
By analyzing and structuring discharge permits using large-scale models, and combining multimodal models and rule bases, the problem of unstructured data processing in solid waste discharge permit supervision was solved, achieving efficient and accurate automated supervision and reducing time and labor costs.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA ENERGY LONGYUAN ENVIRONMENTAL PROTECTION CO LTD
- Filing Date
- 2026-03-03
- Publication Date
- 2026-06-12
Smart Images
Figure CN122197852A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of environmental protection supervision, and in particular to a method, intelligent agent, and system for monitoring the compliance of solid waste discharge information in a discharge permit. Background Technology
[0002] With the continuous improvement of the ecological and environmental protection legal system, the compliant management of solid and hazardous waste (general industrial solid waste and hazardous waste) has become a core aspect of ecological and environmental governance. The "Law of the People's Republic of China on the Prevention and Control of Environmental Pollution by Solid Waste" clearly requires industrial solid waste generators to establish and improve pollution prevention and control responsibility systems. Relevant policy documents from the Ministry of Ecology and Environment have further strengthened the requirements for standardized management of hazardous waste throughout the entire process. According to the "Implementation Plan for the Comprehensive Implementation of the Discharge Permit System" (Environmental Assessment
[2024] No. 79), the discharge permit management of industrial solid waste must be fully completed by 2025, marking that the compliance supervision of solid waste discharge information in discharge permits has become the key to the comprehensive supervision of solid and hazardous waste.
[0003] The information related to solid and hazardous waste in the discharge permit is mainly recorded in the "Basic Information Table of Solid Waste" and the "Information Table of Self-Storage and Self-Utilization / Disposal Facilities". The core supervision content includes the accuracy of the information recorded and the consistency of data between the two tables. Currently, several technological approaches have been developed in the field of solid and hazardous waste information compliance supervision: First, traditional manual supervision, where inspectors directly review enterprises' paper or electronic discharge permits, verifying basic information such as solid waste names, codes, and hazardous characteristics, as well as the matching relationship between storage and disposal facility types and solid waste destinations. Second, a multi-departmental joint review digital platform, which establishes a multi-departmental collaborative mechanism, allowing professional departments such as environmental protection and solid waste management to conduct online synchronous reviews, with the system automatically summarizing opinions and pushing a list of issues. Third, a comprehensive full-element verification system, relying on a verification list and issue database covering environmental procedures, multi-pollutant emissions, and post-permit management, using machine learning algorithms to identify potential risk points in solid and hazardous waste. Fourth, an AI-assisted review system, utilizing natural language processing, computer vision, and other technologies, combined with an environmental regulations knowledge base and review rule base, to intelligently analyze and verify discharge permits and implementation reports, achieving full coverage of all pollution factors.
[0004] However, existing technologies still have significant shortcomings and limitations in application, making it difficult to meet the needs of refined and efficient supervision of solid and hazardous waste discharge permits, as follows:
[0005] First, the manual inspection model suffers from inherent efficiency bottlenecks. This model requires inspectors to review unstructured paper documents or PDF images one by one, either on-site or online. This not only consumes significant manpower and time but also makes it difficult to achieve high-frequency, broad-coverage supervision in scenarios involving a wide distribution of waste-related enterprises and a diverse range of solid waste types. The high proportion of repetitive work leads to overall low efficiency. Furthermore, on-site inspections are constrained by work schedules and time / location, making real-time monitoring impossible and severely lacking in timeliness. The accuracy of inspections also heavily relies on personnel's mastery of complex standards and regulations, demanding extremely high professional skills and making them susceptible to errors due to human negligence.
[0006] Secondly, the inter-departmental joint review digital platform has failed to overcome its shortcomings in data integration and intelligent verification. Although such platforms have optimized inter-departmental collaboration processes, they have not solved the core problem of unstructured solid and hazardous waste information in pollution discharge permits. Manual intervention is still required to complete information extraction and comparison, making data integration and analysis difficult. Automatic verification cannot be achieved through the retrieval of structured data, making it difficult to fundamentally reduce reliance on manpower and time costs.
[0007] Furthermore, the comprehensive verification system suffers from functional redundancy and insufficient targeting. Its verification checklist covers multiple environmental elements, but its focus on solid and hazardous waste discharge permit information is insufficient, and it lacks a dedicated verification mechanism for the data association logic of the two core tables.
[0008] Finally, existing AI-assisted review systems have not yet achieved in-depth and accurate analysis of solid and hazardous waste information. Although these systems incorporate cutting-edge technologies, their adaptability to the solid and hazardous waste professional field needs improvement. Some systems suffer from insufficient accuracy in solid waste identification and incomplete data association and verification logic for specific forms in permits. Furthermore, they do not fully integrate the requirements for the entire life cycle management of solid and hazardous waste, making it difficult to achieve intelligent management across the entire chain from information analysis to risk warning, and thus unable to completely replace manual compliance assessments. Summary of the Invention
[0009] Based on the above problems, this invention solves the technical problems of "difficulty in processing unstructured data, low efficiency in data verification, and weak real-time monitoring capability" in the supervision of solid and hazardous waste discharge permit information in existing technologies. By adopting the supervision method for the compliance of solid waste discharge information in discharge permits proposed in this invention, non-compliant items can be accurately identified, and the time cost of supervision can be significantly reduced.
[0010] This invention proposes a method for monitoring the compliance of solid waste discharge information in a discharge permit, comprising: Obtain the discharge permit document, use a large model to parse the discharge permit and extract key information from the solid waste discharge information. The key information includes at least the basic information of solid waste, information on self-storage facilities, and information on self-utilization or self-disposal facilities. At the same time, extract the company name and waste code. The hazardous waste management ledger is obtained based on the company name and waste code. The large model performs standard structured processing on the data in the hazardous waste management ledger and outputs standard structured data. The compliance of key information in solid waste discharge information is reviewed based on a pre-defined rule base, and the consistency between key information in solid waste discharge information and information in standard structured data of hazardous waste management ledger is reviewed. Output the review results and rectification suggestions.
[0011] In addition, when using a large model to parse the discharge permit, the discharge permit is first pre-parsed, and the key page numbers in the discharge permit are located based on the keywords in the preset knowledge base, and the key page numbers are converted into images. Then, a multimodal model is used to parse the key page numbers that have been converted into images; The analysis results of the multimodal model are calibrated based on a preset rule base.
[0012] In addition, when using a large model to parse the discharge permit, the discharge permit is first pre-parsed, and the key page numbers in the discharge permit are located based on the keywords in the preset knowledge base, and the key page numbers are converted into images. The key page numbers are parsed and converted into images in two ways: one way uses a table OCR engine to recognize the tables and text in the key page numbers, and the other way uses a multimodal model to parse the tables and text. Compare the fields parsed from the two sources. If the fields are the same, they are adopted directly. If the fields are different, a conflict resolution process is used. The fields parsed from the two paths are integrated as the parsing output. The fields contain traceability information, including the source path of the field and the reason for conflict resolution.
[0013] In addition, the conflict resolution process includes: detecting the similarity between the fields parsed from both sources and the predefined rules, detecting the proximity between the fields parsed from both sources and historical data, and selecting one of the parsed fields based on the similarity and proximity. If it is not possible to select one of the parsed fields based on similarity and proximity, then the images of the fields and key page numbers will be marked as requiring manual review.
[0014] In addition, obtaining hazardous waste management records includes: obtaining records of hazardous waste generation, storage, outbound, self-utilization or disposal, and outsourced utilization or disposal within a preset time period.
[0015] In addition, before outputting the review results, the review results are distinguished as compliant and non-compliant. If compliant, the review results are output directly. If non-compliant, output the non-compliant item.
[0016] In addition, the review results from the expert collaboration platform will be used to train a large model.
[0017] The present invention also proposes an intelligent agent for monitoring the compliance of solid waste discharge information in a discharge permit, employing the monitoring method for monitoring the compliance of solid waste discharge information in a discharge permit as described in any of the preceding claims.
[0018] This invention also proposes a monitoring system for the compliance of solid waste discharge information in discharge permits, comprising: The knowledge graph construction module is used to obtain regulations in the field of solid waste discharge and construct a domain knowledge graph containing entity relationships and matching relationships; The rules engine module is used to transform compliance verification logic into executable rule scripts. The rule scripts include at least waste name and code matching rules and facility type and destination matching rules. The data parsing and matching module is used to parse the key information table in the pollution discharge permit and transform the data in the key information table into structured fields. It then calls the rule engine module, which executes rule scripts based on the domain knowledge graph to perform compliance checks on the structured fields.
[0019] This invention also proposes a monitoring system for the compliance of solid waste discharge information in discharge permits, comprising: Edge devices and cloud-based monitoring platforms; Edge devices acquire discharge permit documents, use large models to parse the discharge permits and extract key information from solid waste discharge information. Key information includes at least basic solid waste information, self-storage facility information, and self-utilization or self-disposal facility information. At the same time, the company name and waste code are extracted. The cloud-based monitoring platform retrieves the hazardous waste management ledger based on the company name and waste code. The large model performs standard structured processing on the data in the hazardous waste management ledger and outputs standard structured data. The compliance of key information in solid waste discharge information is reviewed based on a pre-defined rule base, and the consistency between key information in solid waste discharge information and information in standard structured data of hazardous waste management ledger is reviewed. Output the review results and rectification suggestions.
[0020] This invention solves the technical problems of "difficulty in processing unstructured data, low efficiency in data verification, and weak real-time monitoring capability" in the supervision of solid and hazardous waste discharge permit information, which have failed to be effectively addressed by existing technologies. By adopting the supervision method for the compliance of solid waste discharge information in discharge permits proposed in this invention, non-compliant items can be accurately identified, and the time cost of supervision can be significantly reduced. Attached Figure Description
[0021] Figure 1 A flowchart illustrating a method for monitoring the compliance of solid waste discharge information in a discharge permit according to an embodiment of the present invention; Figure 2 for Figure 1 The Zhongda model performs standard structured processing on the data in the hazardous waste management ledger and outputs a schematic diagram of the standard structured data. Detailed Implementation
[0022] The present invention will be further described in detail below with reference to specific embodiments and accompanying drawings. This description is intended only to illustrate specific embodiments of the invention and does not constitute any limitation on the invention. The scope of protection of the invention is defined by the claims.
[0023] Reference Figure 1 This invention proposes a method for monitoring the compliance of solid waste discharge information in a discharge permit, comprising: Step S001: Obtain the discharge permit document, use a large model to parse the discharge permit and extract key information from the solid waste discharge information. The key information includes at least the basic information of solid waste, information on self-storage facilities, and information on self-utilization or self-disposal facilities. At the same time, extract the company name and waste code. Step S002: Obtain the hazardous waste management ledger based on the enterprise name and waste code. The large model performs standard structured processing on the data in the hazardous waste management ledger and outputs standard structured data. Step S003: Review the compliance of key information in solid waste discharge information according to the preset rule base, and review the consistency between key information in solid waste discharge information and information in standard structured data of hazardous waste management ledger. Step S004: Output the review results and rectification suggestions.
[0024] The default rule base is a pre-organized rule base, which stores and organizes rules such as the "National Hazardous Waste List (2025 Edition)" and "Solid Waste Classification Codes".
[0025] The basic information on solid waste corresponds to the "Basic Information Table for Solid Waste", while the information on self-storage facilities and self-utilization or self-disposal facilities corresponds to the "Information Table for Self-Storage and Self-Utilization / Disposal Facilities".
[0026] Optionally, the following items can be validated against the rule base using a large model: 1. Check the "Solid Waste Category" item in the "Solid Waste Basic Information Table" to see if the "Waste Name" recorded in the discharge permit is consistent with the corresponding waste name in the "National Hazardous Waste List (2025 Edition)" for the item with "Hazardous Waste Category". If they are inconsistent, check if the "Waste Name" recorded in the discharge permit is consistent with the waste name corresponding to its "Waste Code" recorded in the hazardous waste management ledger (hazard generation record table, warehousing record table, warehousing record table, self-utilization / disposal record table, and outsourced utilization / disposal record table) for the past year. 2. Check the "Solid Waste Category" item in the "Solid Waste Basic Information Table" to see if the "Hazardous Characteristics" recorded in the discharge permit are consistent with the corresponding hazardous characteristics of the "Waste Code" in the "National Hazardous Waste List (2025 Edition)" (Note: If the "National Hazardous Waste List (2025 Edition)" indicates that the "Waste Code" has one or more hazardous characteristics represented by the listed code, then the "Hazardous Characteristics" recorded in the discharge permit must include the hazardous characteristics of the "Waste Code" in the list; otherwise, the "Hazardous Characteristics" recorded in the discharge permit must be the same as the hazardous characteristics of the "Waste Code" in the list). 3. Check the "Solid Waste Category" item in the "Solid Waste Basic Information Table" to see if the "Waste Name" recorded in the discharge permit is consistent with the corresponding waste type name in the "Solid Waste Classification Code" (Note: Currently, the discharge permit only records the waste category for general industrial solid waste, not the waste code, so check if the waste name in the discharge permit is consistent with the waste type name in the "Solid Waste Classification Code"). 4. If the destination of hazardous waste in the "Solid Waste Basic Information Table" is "self-storage", check whether the "self-storage and self-utilization / disposal facility information table" has "self-storage facility" in the "Solid Waste Category" item for "Hazardous Waste"; 5. If the destination of general industrial solid waste in the "Basic Information Table of Solid Waste" is "self-storage", check whether the "self-storage and self-utilization / disposal facility information table" has "self-storage facility" in the "Facility Type" corresponding to the item with "General Industrial Solid Waste" as the "Solid Waste Category". 6. If the destination of hazardous waste in the "Solid Waste Basic Information Table" is listed as "self-utilization" or "self-disposal", check the "Solid Waste Category" item in the "Self-Storage and Self-Utilization / Disposal Facility Information Table" to see if there is a "Self-Utilization / Disposal Facility" in the corresponding "Facility Type". 7. If the destination of general industrial solid waste in the "Solid Waste Basic Information Table" is listed as "self-utilization" or "self-disposal", check the "Solid Waste Category" item in the "Self-Storage and Self-Utilization / Disposal Facility Information Table" to see if there is a "Self-utilization / Disposal Facility" in the corresponding "Facility Type".
[0027] Based on the compliance verification, the review results are output item by item in sequence, and rectification suggestions are given for non-compliant items: 1. Check if the discharge permit is valid at the current time. If it is no longer valid, prompt "Please upload a valid discharge permit" and stop outputting the review results for subsequent items; otherwise, the item is compliant and the review result for item 2 will be output.
[0028] 2. Check if the discharge permit contains a "Solid Waste Discharge Information" module. If not, display the message "Solid waste discharge information is not filed in the discharge permit" and stop outputting the review results for subsequent items; otherwise, this item is compliant, and output the review results for items 3 to 8.
[0029] 3. Whether the names of hazardous wastes in the "Basic Information Table for Solid Waste" of the discharge permit are consistent with the "National Hazardous Waste List (2025 Edition)" or the hazardous waste management ledger. If they are inconsistent, this item is non-compliant, and the notice will state: "According to Article 13 of the 'Regulations on the Administration of Discharge Permits' (State Council Decree No. 736), the discharge permit shall record the following information: '...location and number of pollutant discharge outlets, pollutant discharge methods and discharge destinations, etc.' and Article 15, 'an increase in the number of pollutant discharge outlets or the type, quantity, and concentration of pollutants discharged. Enterprises should accurately and truthfully declare the hazardous wastes generated during actual production and operation." Otherwise, this item is compliant.
[0030] 4. Check whether the name of general industrial solid waste in the "Basic Information Table of Solid Waste" of the discharge permit is consistent with the "Solid Waste Classification Code". If they are inconsistent, this item is non-compliant, and the notice will state: "According to Article 13 of the 'Regulations on the Administration of Discharge Permits' (State Council Decree No. 736), the discharge permit shall record the following information: '...location and number of pollutant discharge outlets, pollutant discharge methods and discharge destinations, etc.' and Article 15, 'an increase in the number of pollutant discharge outlets or the type, quantity, and concentration of pollutants discharged.' Enterprises should accurately and truthfully declare the general industrial solid waste generated during actual production and operation." Otherwise, this item is compliant.
[0031] 5. Whether the hazardous characteristics of hazardous waste in the "Solid Waste Basic Information Table" of the discharge permit are consistent with the "National Hazardous Waste List (2025 Edition)". If they are inconsistent, this item is non-compliant, and the notice states: "According to Article 13 of the 'Regulations on the Administration of Discharge Permits' (State Council Decree No. 736), the discharge permit shall record the following information: '...location and number of pollutant discharge outlets, pollutant discharge methods and discharge destinations, etc.' and Article 15, 'an increase in the number of pollutant discharge outlets or the type, amount, and concentration of pollutants discharged. Enterprises should accurately and truthfully declare the hazardous waste generated in the actual production and operation process." Otherwise, this item is compliant.
[0032] 6. If the "Solid Waste Basic Information Form" in the Discharge Permit lists "self-storage" as the destination for general industrial solid waste or hazardous waste, does the "Self-Storage and Self-Utilization / Disposal Facility Information Form" include "self-storage facilities"? If not, this item is non-compliant, and the notice states: "According to Article 13 of the 'Regulations on Discharge Permit Management' (State Council Decree No. 736), the Discharge Permit shall record the following information: '...location and number of pollutant discharge outlets, pollutant discharge methods and destinations, etc.' and Article 15, 'an increase in the number of pollutant discharge outlets or the type, quantity, and concentration of pollutants discharged.' Enterprises should accurately and truthfully declare the destination of industrial solid waste generated during actual production and operation." Otherwise, this item is compliant.
[0033] 7. In the "Basic Information Table for Solid Waste" of the discharge permit, the destination of general industrial solid waste or hazardous waste is listed as "self-utilization" or "self-disposal." Does the "Information Table for Self-Storage and Self-Utilization / Disposal Facilities" include "self-utilization / disposal facilities"? If not, this item is non-compliant, and the notice states: "According to Article 13 of the 'Regulations on the Administration of Discharge Permits' (State Council Decree No. 736), the discharge permit shall record the following information: '...location and number of pollutant discharge outlets, pollutant discharge methods and destinations, etc.' and Article 15, 'an increase in the number of pollutant discharge outlets or the type, quantity, and concentration of pollutants discharged.' Enterprises should accurately and truthfully declare the destination of industrial solid waste generated during actual production and operation." Otherwise, this item is compliant.
[0034] This invention employs a large model, such as the multimodal visual recognition capability of a large model, to complete the parsing of solid waste discharge information and automatic extraction of related information from the discharge permit (which can be an electronic document or a copy). It also uses a large model algorithm for compliance verification, accurately identifying non-compliant items with an accuracy rate of no less than 85%, thus avoiding inaccurate monitoring results due to human judgment errors or omissions.
[0035] By leveraging the visual recognition capabilities of a large-scale model, the solid waste information table data in the discharge permit (copy) document is extracted by row and column and parsed into structured data, solving the common problems of poor table data parsing and easy cross-referencing in existing models. Through large-scale model training, the parsing effect and accuracy when the same row of data spans multiple pages are improved.
[0036] Using a large model, the pollution discharge permit (copy) documents stored in image form are intelligently parsed, and the data in the image tables is transformed into structured data. This enables the retrieval of structured data and direct compliance verification, solving the data integration and analysis problem of solid waste discharge information in pollution discharge permits.
[0037] Cloud-based monitoring eliminates the need for inspectors to physically visit companies to review discharge permits, saving time and reducing inspection time by approximately 95%, significantly lowering the time cost of inspections. It also allows for simultaneous intelligent compliance checks on discharge permits from multiple companies (e.g., 5 companies), eliminating the need for on-site verification by inspectors. Inspectors only need to manually review individual issues, further reducing labor costs.
[0038] The large-scale model is integrated into the solid and hazardous waste management system, allowing inspection reports and rectification suggestions to be directly pushed to enterprises through the system, urging them to complete closed-loop rectification within the system. The large-scale model is used to deeply analyze inspection results, generating inspection maps to easily identify enterprises requiring key monitoring and automatically adjust subsequent inspection plans.
[0039] This invention solves the technical problems of "difficulty in processing unstructured data, low efficiency in data verification, and weak real-time monitoring capability" in the supervision of solid and hazardous waste discharge permit information, which have failed to be effectively addressed by existing technologies. By adopting the supervision method for the compliance of solid waste discharge information in discharge permits proposed in this invention, non-compliant items can be accurately identified, and the time cost of supervision can be significantly reduced.
[0040] In one embodiment, when using a large model to parse the discharge permit, the discharge permit is first pre-parsed, and the key page number in the discharge permit is located according to the keywords in the preset knowledge base, and the key page number is converted into an image. Then, a multimodal model is used to parse the key page numbers that have been converted into images; The analysis results of the multimodal model are calibrated based on a preset rule base.
[0041] Optionally, semantic relevance analysis is used to locate page numbers containing key information (such as company name, expiration date, and hazardous waste code); the located key page numbers are converted from PDF documents to high-resolution image format, and image quality enhancement processing is performed to optimize subsequent recognition results; The converted image is input into a multimodal large model, and structured information is extracted by combining targeted parsing prompts, including but not limited to: license validity period (start date and end date), company name, hazardous waste code, and other relevant information; The purpose of pre-parsing is not to extract specific field values, but to prepare for detailed parsing: to find the pages and approximate locations containing important information, thereby avoiding detailed parsing of the entire PDF, reducing the processing burden of large models, focusing on key areas, and avoiding noise interference.
[0042] Optionally, after parsing the discharge permit, it is determined whether the discharge permit is currently valid. If the discharge permit has expired, the determination is terminated. It is then determined whether the discharge permit contains "solid waste discharge information." If not, the determination is terminated. If the discharge permit contains "solid waste discharge information," then based on the located "solid waste discharge information" section, the "solid waste category," "solid waste name," "code," "hazardous characteristics," and "destination" of the solid waste in the "Solid Waste Basic Information Table" are parsed, and the "solid waste category" and "facility type" in the "Self-Storage and Self-Utilization / Disposal Facility Information Table" are parsed.
[0043] In one embodiment, when using a large model to parse the discharge permit, the discharge permit is first pre-parsed, and the key page number in the discharge permit is located according to the keywords in the preset knowledge base, and the key page number is converted into an image. The key page numbers are parsed and converted into images in two ways: one way uses a table OCR engine to recognize the tables and text in the key page numbers, and the other way uses a multimodal model to parse the tables and text. Compare the fields parsed from the two sources. If the fields are the same, they are adopted directly. If the fields are different, a conflict resolution process is used. The fields parsed from the two paths are integrated as the parsing output. The fields contain traceability information, including the source path of the field and the reason for conflict resolution.
[0044] This embodiment employs a two-pronged parsing approach. The table OCR engine, used to recognize tables and text within key page numbers, offers higher accuracy and faster parsing speed for fixed tables. The multimodal model for table and text parsing demonstrates strong generalization ability and can handle a wider range of layouts. Furthermore, the strategy of comparing the parsing results enhances accuracy.
[0045] In one embodiment, the conflict resolution process includes: detecting the similarity between the two parsed fields and predefined rules, detecting the proximity between the two parsed fields and historical data, and selecting one of the parsed fields based on the similarity and proximity. If it is not possible to select one of the parsed fields based on similarity and proximity, then the images of the fields and key page numbers will be marked as requiring manual review.
[0046] By handling the conflict between the two detection paths, the detection results are made more accurate.
[0047] In one embodiment, obtaining the hazardous waste management ledger includes: obtaining records of hazardous waste generation, storage, outbound, self-utilization or disposal, and outsourced utilization or disposal within a preset time period.
[0048] The preset time period is, for example, one year. The above record table provides comprehensive information for comparison.
[0049] In one embodiment, the checks performed after the large model resolves the discharge permit include: 1. Check if the discharge permit is valid; 2. Check whether the discharge permit contains the "Solid Waste Discharge Information" section, including the "Solid Waste Basic Information Table" and the "Self-Storage and Self-Utilization / Disposal Facility Information Table"; 3. Check whether the "Solid Waste Name" and "Hazardous Characteristics" of hazardous waste in the "Solid Waste Basic Information Table" are consistent with the information recorded in the "National Hazardous Waste List"; 4. Check whether the "Solid Waste Name" for hazardous waste in the "Solid Waste Basic Information Form" is consistent with the waste name recorded in the enterprise's hazardous waste generation ledger for the most recent year; 5. Check whether the "Solid Waste Name" for general industrial solid waste in the "Solid Waste Basic Information Table" is consistent with the information recorded in the "Solid Waste Classification Code"; 6. If the destination of general industrial solid waste or hazardous waste in the "Solid Waste Basic Information Table" is "self-storage", then check whether there is a corresponding "self-storage facility" for general industrial solid waste or hazardous waste in the "Self-storage and Self-utilization / Disposal Facility Information Table". 7. If the destination of general industrial solid waste or hazardous waste in the "Solid Waste Basic Information Table" is listed as "self-utilization" or "self-disposal", check whether there is a corresponding "self-utilization / disposal facility" for general industrial solid waste or hazardous waste in the "Self-storage and Self-utilization / Disposal Facility Information Table". 8. Based on the compliance review results, generate a review report and rectification suggestions.
[0050] In one embodiment, before outputting the review results, the review results are distinguished as compliant or non-compliant. If compliant, the review results are output directly. If non-compliant, output the non-compliant item.
[0051] By differentiating the review results, different output formats and styles can be used for output. Additionally, non-compliant items can be directly sent to the enterprise's system.
[0052] Optionally, if the issue is non-compliant, direct rectification is sufficient. If it is compliant but the intelligent agent makes a recognition error, an appeal opportunity is provided, requiring verification by an expert platform. Only appealed items require expert platform verification.
[0053] In one embodiment, the review results are graded before being output; if the compliance is level one, the review results are output directly. If the compliance level is Level 2, then candidate matching items will be generated for manual selection of non-compliant review items; If the compliance level is three, then the non-compliant review items are marked as high priority items awaiting review; The non-compliant review items, discharge permit documents, key information from solid waste discharge information, standard structured data from waste management ledgers, and review results are sent to the expert collaboration platform. The expert collaboration platform verifies all the information and data received, marks the errors, and provides the correct review results.
[0054] Based on intelligent verification, a manual review process is introduced. By adopting the "large model initial screening + expert review" model, the verification challenges in complex scenarios (such as fuzzy text recognition and solid waste classification in special industries) are solved, balancing review efficiency and accuracy.
[0055] Intelligent initial screening and grading: The system automatically verifies the information on the analyzed discharge permits and classifies the results into three levels: Level 1 (Explicit Compliance): Directly outputs conclusions without human intervention; Level 2 (Minor Questionable): For example, if "the waste name is similar to but not completely consistent with the list", the system will generate candidate matching items for manual selection; "the waste name is similar to but not completely consistent with the list" is considered non-compliant.
[0056] Level 3 (Seriously questionable): Such as "code missing" or "facility type and destination completely mismatched", automatically marked as high-priority items to be reviewed.
[0057] Expert Collaboration Platform: Develop a web-based review interface that displays the original scanned copy of the questionable item, the parsed data, the legal basis, and the system's preliminary judgment. Experts can directly verify the erroneous parsing results or adjust the verification rules (such as adding industry-specific name mapping relationships).
[0058] The beneficial effects of this embodiment include: It can adapt to complex scenarios: it solves the shortcomings of large models in recognizing fuzzy text and industry-specific terms, and improves the accuracy of complex case verification to 95%.
[0059] System self-optimization: Through continuous optimization of parsing and verification logic based on expert feedback, the accuracy of automatic recognition can be further improved after six months of use.
[0060] In one embodiment, the review results from the expert collaboration platform are used to train a large model.
[0061] Enhanced knowledge accumulation mechanism: Expert review results are automatically fed back to the system training library. For example, the correspondence between "common waste names" and standard names in a certain industry is included in the mapping table, which can improve the automatic identification accuracy of similar cases by 30%.
[0062] The present invention also proposes an intelligent agent for monitoring the compliance of solid waste discharge information in a discharge permit, employing the monitoring method for monitoring the compliance of solid waste discharge information in a discharge permit as described in any of the preceding claims.
[0063] The intelligent agent application development comprises several sub-modules, such as knowledge acquisition, knowledge processing, database querying, prompt word template assembly, and rule comparison. The knowledge acquisition module is responsible for acquiring data from external sources and processing it through the intelligent agent for monitoring solid and hazardous waste information from discharge permits, as well as rule comparison. The knowledge processing module includes database querying, prompt word template assembly, and rule comparison; these steps are used for further data processing and analysis. The intelligent agent module is responsible for coordinating and managing the entire intelligent agent application process. A structured database stores various types of data, including OSS ID data, rectification opinion data, hazardous waste management ledger data, discharge permit data, and enterprise organizational structure data.
[0064] This invention solves the technical problems of "difficulty in processing unstructured data, low efficiency in data verification, and weak real-time monitoring capability" in the supervision of solid and hazardous waste discharge permit information, which have failed to be effectively addressed by existing technologies. By adopting the supervision method for the compliance of solid waste discharge information in discharge permits proposed in this invention, non-compliant items can be accurately identified, and the time cost of supervision can be significantly reduced.
[0065] This invention also proposes a monitoring system for the compliance of solid waste discharge information in discharge permits, comprising: The knowledge graph construction module is used to obtain regulations in the field of solid waste discharge and construct a domain knowledge graph containing entity relationships and matching relationships; The rules engine module is used to transform compliance verification logic into executable rule scripts. The rule scripts include at least waste name and code matching rules and facility type and destination matching rules. The data parsing and matching module is used to parse the key information table in the pollution discharge permit and transform the data in the key information table into structured fields. It then calls the rule engine module, which executes rule scripts based on the domain knowledge graph to perform compliance checks on the structured fields.
[0066] By constructing a standardized rule engine and a domain knowledge graph in the field of solid and hazardous waste, the autonomous reasoning ability of large models can be replaced, realizing the structured compliance verification of solid waste information in the discharge permit, which is suitable for regulatory needs with clear rules and fixed scenarios.
[0067] Optionally, domain knowledge graph construction: integrate regulations and standards such as the National Hazardous Waste List (2025 Edition), Solid Waste Classification Code, and Discharge Permit Management Regulations to construct a knowledge graph containing relationships such as "waste code-name-hazardous characteristics" and "facility type-destination matching", forming a visualized entity association network.
[0068] The rules engine module design transforms compliance verification logic (such as matching waste names with codes and matching facility types with destinations) into executable rule scripts (such as "if the 'destination' is 'self-storage', the facility type must include 'self-storage facility'"). Optionally, a total of 32 core rules are preset.
[0069] The data parsing and matching module uses OCR technology to parse the "Basic Information Table of Solid Waste" and the "Information Table of Self-Storage and Self-Utilization / Disposal Facilities" in the discharge permit, converting the data into structured fields (such as "Waste Code", "Destination", "Facility Type"). The rule engine module calls the domain knowledge graph to perform field matching, such as automatically comparing the consistency of waste names with the directory, or verifying the correspondence between facility types and destinations.
[0070] Output: Items that do not meet the rules are automatically marked and associated with the relevant legal basis, generating a standardization review report.
[0071] The effective effects of this embodiment include: High rule transparency: The verification logic is presented in a visual rule script, which is easy for regulators to understand and adjust. It is suitable for scenarios with frequent regulatory updates, and the rule iteration efficiency is improved by 40%.
[0072] Lightweight system: No need for large model computing power support, deployment costs are reduced by 60%, and it can run on local servers in grassroots regulatory departments.
[0073] Stable accuracy: Within the scope of the rules, the accuracy of compliance verification reaches 90%, avoiding misjudgments caused by data bias in large models.
[0074] This invention also proposes a monitoring system for the compliance of solid waste discharge information in discharge permits, comprising: Edge devices and cloud-based monitoring platforms; Edge devices acquire discharge permit documents, use large models to parse the discharge permits and extract key information from solid waste discharge information. Key information includes at least basic solid waste information, self-storage facility information, and self-utilization or self-disposal facility information. At the same time, the company name and waste code are extracted. The cloud-based monitoring platform retrieves the hazardous waste management ledger based on the company name and waste code. The large model performs standard structured processing on the data in the hazardous waste management ledger and outputs standard structured data. The compliance of key information in solid waste discharge information is reviewed based on a pre-defined rule base, and the consistency between key information in solid waste discharge information and information in standard structured data of hazardous waste management ledger is reviewed. Output the review results and rectification suggestions.
[0075] Optionally, the multimodal recognition model is a large visual language model.
[0076] The large model performs standard structured processing on the data in the hazardous waste management ledger, outputting standard structured data, for example... Figure 2 As shown.
[0077] Definitions: Storage: The activity of temporarily placing solid waste in a specific facility or place.
[0078] Utilization: The activity of extracting substances from solid waste as raw materials or fuels.
[0079] Disposal: Activities involving the incineration of solid waste and other methods that alter the physical, chemical, or biological properties of solid waste to reduce the amount of solid waste generated, reduce its volume, reduce or eliminate its hazardous components, or the final disposal of solid waste in landfills that comply with environmental protection regulations.
[0080] As needed, the above technical solutions can be combined to achieve the best technical effect.
[0081] The above description is merely the principle and preferred embodiment of the present invention. It should be noted that, for those skilled in the art, several other modifications can be made based on the principle of the present invention, and these modifications should also be considered within the scope of protection of the present invention.
Claims
1. A method for monitoring the compliance of solid waste discharge information in a discharge permit, characterized in that, include: Obtain the discharge permit document, use a large model to parse the discharge permit and extract key information from the solid waste discharge information. The key information includes at least the basic information of solid waste, information on self-storage facilities, and information on self-utilization or self-disposal facilities. At the same time, extract the company name and waste code. The hazardous waste management ledger is obtained based on the company name and waste code. The large model performs standard structured processing on the data in the hazardous waste management ledger and outputs standard structured data. The compliance of key information in solid waste discharge information is reviewed based on a pre-defined rule base, and the consistency between key information in solid waste discharge information and information in standard structured data of hazardous waste management ledger is reviewed. Output the review results and rectification suggestions.
2. The method for monitoring the compliance of solid waste discharge information in a discharge permit according to claim 1, characterized in that, When using a large model to parse the discharge permit, the discharge permit is first pre-parsed. Based on the keywords in the preset knowledge base, the key page numbers in the discharge permit are located and converted into images. Then, a multimodal model is used to parse the key page numbers that have been converted into images; The analysis results of the multimodal model are calibrated based on a preset rule base.
3. The method for monitoring the compliance of solid waste discharge information in a discharge permit according to claim 1, characterized in that, When using a large model to parse the discharge permit, the discharge permit is first pre-parsed. Based on the keywords in the preset knowledge base, the key page numbers in the discharge permit are located and converted into images. The key page numbers are parsed and converted into images in two ways: one way uses a table OCR engine to recognize the tables and text in the key page numbers, and the other way uses a multimodal model to parse the tables and text. Compare the fields parsed from the two sources. If the fields are the same, they are adopted directly. If the fields are different, a conflict resolution process is used. The fields parsed from the two paths are integrated as the parsing output. The fields contain traceability information, including the source path of the field and the reason for conflict resolution.
4. The method for monitoring the compliance of solid waste discharge information in a discharge permit according to claim 3, characterized in that, The conflict resolution process includes: detecting the similarity between the fields parsed from both sources and the predefined rules, detecting the proximity between the fields parsed from both sources and historical data, and selecting one of the parsed fields based on the similarity and proximity. If it is not possible to select one of the parsed fields based on similarity and proximity, then the images of the fields and key page numbers will be marked as requiring manual review.
5. The method for monitoring the compliance of solid waste discharge information in a discharge permit according to claim 1, characterized in that, Obtaining hazardous waste management records includes: obtaining records of hazardous waste generation, warehousing, outbound, self-utilization or disposal, and outsourced utilization or disposal for a preset time period.
6. The method for monitoring the compliance of solid waste discharge information in a discharge permit according to claim 1, characterized in that, Before outputting the review results, distinguish between compliant and non-compliant results. If compliant, output the review results directly. If non-compliant, output the non-compliant item.
7. The method for monitoring the compliance of solid waste discharge information in a discharge permit according to claim 6, characterized in that, The review results from the expert collaboration platform were used to train a large model.
8. A monitoring intelligent agent for the compliance of solid waste discharge information in a discharge permit, characterized in that, The method for monitoring the compliance of solid waste discharge information in the discharge permit as described in any one of claims 1-7 is adopted.
9. A monitoring system for the compliance of solid waste discharge information in a discharge permit, characterized in that, include: The knowledge graph construction module is used to obtain regulations in the field of solid waste discharge and construct a domain knowledge graph containing entity relationships and matching relationships; The rules engine module is used to transform compliance verification logic into executable rule scripts. The rule scripts include at least waste name and code matching rules and facility type and destination matching rules. The data parsing and matching module is used to parse the key information table in the pollution discharge permit and transform the data in the key information table into structured fields. It then calls the rule engine module, which executes rule scripts based on the domain knowledge graph to perform compliance checks on the structured fields.
10. A monitoring system for the compliance of solid waste discharge information in a discharge permit, characterized in that, include: Edge devices and cloud-based monitoring platforms; Edge devices acquire discharge permit documents, use large models to parse the discharge permits and extract key information from solid waste discharge information. Key information includes at least basic solid waste information, self-storage facility information, and self-utilization or self-disposal facility information. At the same time, the company name and waste code are extracted. The cloud-based monitoring platform retrieves the hazardous waste management ledger based on the company name and waste code. The large model performs standard structured processing on the data in the hazardous waste management ledger and outputs standard structured data. The compliance of key information in solid waste discharge information is reviewed based on a pre-defined rule base, and the consistency between key information in solid waste discharge information and information in standard structured data of hazardous waste management ledger is reviewed. Output the review results and rectification suggestions.