A safe production service quality improvement response system oriented to government-enterprise linkage

Abstract

The application discloses a safety production service quality improvement response system for government-enterprise linkage, and belongs to the technical field of state monitoring of hydroelectric generating equipment.The application arranges multiple types of sensors on a stator core, synchronously collects vibration, temperature, air gap and partial discharge signals, and synchronously collects operation condition parameters;preprocesses the collected multiple source signals;constructs a multi-physical field coupling model of an electromagnetic field, a temperature field and a structure field, extracts multi-physical field coupling characteristic parameters;calculates a comprehensive health index HI of the stator core;inputs the multiple source monitoring data and the comprehensive health index into a trained graph attention network and a long short-term memory network mixed deep learning model, predicts a future health state of the stator core, and outputs a fault early warning signal and a fault type identification result.The application realizes multi-source information fusion and physical mechanism driven online monitoring and trend prediction of the stator core, and has the advantages of comprehensive monitoring, accurate diagnosis and timely early warning.

Description

Technical Field

[0001] This invention relates to the technical field of safety production supervision, specifically to a safety production service quality improvement response system for government-enterprise collaboration. Background Technology

[0002] Work safety is a matter of vital importance to the safety of people's lives and property. Traditional work safety supervision models mainly rely on government departments conducting regular inspections, special supervisions, and post-incident accountability, which have the following prominent problems: First, there is an information asymmetry between the government and enterprises. Safety data is isolated between government regulatory departments and enterprises, forming "data silos." The government finds it difficult to grasp the real-time status of enterprises' safe production, while enterprises cannot obtain policy regulations and risk warning information in a timely manner.

[0003] Second, regulatory response is lagging. Traditional regulation relies mainly on periodic inspections, which cannot achieve real-time monitoring and dynamic early warning of safety risks. Emergency response is often only initiated after an accident occurs, lacking the ability to prevent accidents before they happen and intervene during the accident.

[0004] Third, coordination and collaboration are inefficient. There is a lack of efficient information sharing and collaborative response mechanisms among government departments, enterprises, professional rescue forces, and other stakeholders. Cross-departmental collaboration in emergency plans is weak, and proactive response is slow.

[0005] Fourth, there is insufficient service supply. Existing safety production management systems are mostly focused on regulatory enforcement, lacking the ability to proactively serve enterprises and provide personalized assistance, thus failing to meet the differentiated safety service needs of enterprises of different industries and sizes.

[0006] In response to the aforementioned problems, several safety production information systems have emerged in recent years. For example, Chinese patent CN122134529A discloses a multi-source heterogeneous data fusion safety monitoring and emergency command system for power generation enterprises, which realizes the real-time generation of emergency response plans through a dynamic intelligent simulation engine; Chinese patent CN121599472A discloses a real-time early warning system for construction safety risks based on multi-source data fusion, which performs data fusion and dynamic risk calculation through a machine learning model; and Chinese patent CN121190022A discloses a high-risk industry safety production intelligent management and control system based on dynamic verification, which ensures that the verification data is tamper-proof through a blockchain evidence storage module.

[0007] However, the above-mentioned existing technologies still have the following shortcomings: (1) They lack a systematic architecture design for government-enterprise collaboration. The functional modules of the government and enterprises are relatively independent and fail to achieve true two-way interaction; (2) Data fusion is mostly limited to a single industry or a single type of data and fails to achieve deep fusion of multi-source heterogeneous data across departments and levels; (3) They lack intelligent risk assessment and personalized service capabilities based on large language models; (4) They fail to establish a closed-loop service quality improvement response mechanism for government-enterprise collaboration.

[0008] To address these shortcomings, this invention proposes a safety production service quality improvement response system for government-enterprise collaboration. Summary of the Invention

[0009] This invention addresses the technical problems of existing safety production management systems, such as information asymmetry between government and enterprises, delayed regulatory response, low efficiency of collaborative linkage, and insufficient service supply. It provides a safety production service quality improvement and response system for government-enterprise collaboration, employing the following technical solution: This invention provides a safety production service quality improvement response system for government-enterprise collaboration, comprising: The data acquisition layer is used to collect safety production-related data from multiple data sources, including IoT sensors deployed at the enterprise end, video surveillance equipment, enterprise safety production management systems, government regulatory databases, insurance institution data systems, and meteorological and environmental data platforms. The data fusion and governance layer, connected to the data acquisition layer, is used to perform data cleaning, format standardization, time-series alignment, and correlation fusion processing on the acquired multi-source heterogeneous data to generate a structured and secure dataset in a unified format. The intelligent risk assessment engine connects to the data fusion and governance layer and is used to perform dynamic risk calculation and graded early warning based on the structured security dataset and a pre-trained risk assessment model to generate risk assessment results. The government-enterprise collaborative service module is connected to the intelligent risk assessment engine and is used to push differentiated security service information and collaborative handling tasks to the government and enterprises respectively based on the risk assessment results. The response quality improvement closed-loop management module is connected to the government-enterprise collaborative service module. It is used to track, record, evaluate, and optimize the entire process of security service information and collaborative handling tasks, forming a closed-loop management of "discovery-push-handling-evaluation-optimization". The blockchain evidence storage module connects the data acquisition layer, the intelligent risk assessment engine, and the response quality improvement closed-loop management module. It is used to store key security data and operation records on the blockchain to ensure the immutability and full traceability of the data.

[0010] Optionally, the data fusion and governance layer includes: The data cleaning submodule identifies and processes missing, outlier, and duplicate values. The format standardization submodule converts different data formats into a unified standard data format. Timing alignment submodule: performs timestamp alignment and interpolation compensation on data collected at different frequencies; Data association and fusion submodule: Based on spatiotemporal correlation, it fuses multi-source data into a structured dataset.

[0011] Optionally, the intelligent risk assessment engine includes: Risk Feature Extraction Submodule: Extracts security risk feature vectors from structured datasets; Dynamic risk assessment submodule: Based on the aforementioned feature vector, calculate the real-time security risk index R. The calculation model for the real-time security risk index R is as follows: ; in, The equipment safety risk sub-index is obtained by comparing the collected parameters such as equipment vibration frequency, operating temperature, current, and voltage with the normal operating range, calculating the deviation, and then weighting and normalizing them. The environmental safety risk sub-index is obtained by comparing parameters such as ambient temperature, humidity, concentration of harmful gases, and dust concentration with safety thresholds, calculating the degree of exceedance, and then weighting and normalizing the results. The sub-index of personnel behavior safety risk is obtained by normalizing the safety equipment wearing rate and the number of violations through video surveillance analysis using a behavior recognition model. The safety risk sub-index is obtained by comprehensively evaluating management data such as the completion rate of hazard rectification and the coverage rate of safety training. Let be the weighting coefficient, satisfying The determination was made through a combination of the analytic hierarchy process and expert scoring. The graded early warning submodule divides risks into low risk (0≤R<30), medium risk (30≤R<60), high risk (60≤R<80), and major risk (80≤R≤100) based on the R value range, and generates four levels of early warning: blue, yellow, orange, and red. Risk Trend Prediction Submodule: Based on historical risk data, it uses an LSTM time series prediction model to predict risk trends for the next 7 days.

[0012] Optionally, the government-enterprise collaboration service module includes: Government-side service sub-module: Provides an overview of enterprise security status, risk classification statistics, law enforcement inspection suggestions, and emergency response command; Enterprise-side service sub-module: Provides real-time risk alerts, hazard identification guidance, safety training push notifications, and rectification task management; Intelligent Question Answering Submodule: Based on large language models and Retrieval Enhanced Generation (RAG) technology, it provides enterprises with intelligent question answering services such as regulatory interpretation, hazard guidance, and emergency consultation. Collaborative task scheduling submodule: Automatically generates collaborative handling tasks based on risk level and event type and pushes them accurately.

[0013] Optionally, the response quality improvement closed-loop management module includes: Task tracking submodule: Records the receipt, processing, and completion status of tasks, as well as timestamps for each stage, in real time; The effectiveness evaluation submodule calculates the service quality improvement effectiveness evaluation value E after the task is completed. The calculation model for the service quality improvement effectiveness evaluation value E is as follows: ;in, This is a real-time security risk index before the task is executed. This is the real-time security risk index after the task is completed and a preset evaluation period (preferably 7 days); when E>0, it indicates a decrease in risk, and the larger the E value, the more significant the quality improvement effect; when E≤0, it indicates ineffectiveness and triggers feedback optimization. Feedback Optimization Submodule: Automatically adjusts the weight parameters of the intelligent risk assessment engine and the collaborative service push strategy based on the E-value.

[0014] Optionally, the blockchain evidence storage module stores enterprise hidden danger investigation records, rectification completion records, government law enforcement inspection records, risk warning and disposal records, and service quality improvement effect evaluation records on the blockchain. Each piece of data generates a unique evidence storage ID (SHA-256 hash), which supports regulatory authorities to verify the authenticity with one click.

[0015] Optionally, the system adopts a "cloud-edge-device" collaborative architecture: the cloud deploys a data fusion and governance layer, an intelligent risk assessment engine, and a blockchain evidence storage module; the edge is deployed on the enterprise side, running a lightweight data preprocessing and real-time risk assessment model to achieve millisecond-level local response; the terminals include government regulatory terminals and enterprise terminals, used for data collection and information display.

[0016] In summary, the present invention has at least one of the following beneficial technical effects: Deep integration of government and enterprise data: Connecting data from multiple sources such as government, enterprises, insurance, and meteorology, breaking down data silos, and providing a comprehensive data foundation for accurate analysis.

[0017] Intelligent risk dynamic assessment: Through multi-factor weighted models and deep learning, real-time risk index calculation and trend prediction are achieved, improving the accuracy and timeliness of early warning.

[0018] Government-enterprise collaborative closed-loop service: Construct a closed loop of "discovery-push-handling-evaluation-optimization" to achieve a positive interaction between government supervision and enterprise self-governance.

[0019] Data trustworthiness and accountability: Blockchain-based evidence storage ensures that critical data is tamper-proof and supports traceability and verification.

[0020] Personalized intelligent services: Intelligent question answering based on a large language model provides customized security guidance for enterprises, making up for the lack of service supply. Attached Figure Description

[0021] Figure 1 This is an overall system diagram of Embodiment 1 of this application. Detailed Implementation

[0022] The following combination Figure 1 The present invention will be described in further detail below.

[0023] Example 1: Refer to Figure 1 This application provides a safety production service quality improvement response system for government-enterprise collaboration, comprising: The data acquisition layer is used to collect safety-related data from multiple data sources, including IoT sensors deployed at the enterprise level, video surveillance equipment, enterprise safety production management systems, government regulatory databases, insurance institution data systems, and meteorological and environmental data platforms. (For example, multiple data sources are connected through standardized interfaces. At the enterprise level, IoT sensors are deployed to collect parameters such as vibration frequency (Hz), operating temperature (°C), current (A), and voltage (V), with a sampling frequency of 1 time / second; video surveillance equipment collects video streams from the work site at a frame rate of 25fps, transmitted via the RTSP protocol; the enterprise safety production management system provides management data such as hazard investigation records, rectification records, and training records through a RESTful API. At the government level, basic enterprise information, historical law enforcement records, and administrative penalty information are obtained through a data exchange platform. In addition, data from insurance institution data systems (safety production liability insurance underwriting and claims data) and meteorological and environmental data platforms (regional temperature, humidity, wind force, rainstorm warnings, etc.) are accessed via APIs.)

[0024] The data fusion and governance layer, connected to the data acquisition layer, is used to perform data cleaning, format standardization, time-series alignment, and correlation fusion processing on the acquired multi-source heterogeneous data to generate a structured and secure dataset in a unified format. The data fusion and governance layer includes: The data cleaning submodule identifies and processes missing, outlier, and duplicate values. (For missing values, linear interpolation is used (missing rate <5%), imputation using the mean of the same type (5%~20%), or discarding and logging (>20%). For outliers, the 3σ principle is used to identify them (exceeding the mean ± 3 times the standard deviation) and they are marked for review. For duplicate values, deduplication is performed based on timestamps and data source identifiers.) The format standardization submodule converts different data formats into a unified standard data format (converting all data into JSON format, including fields: source_id (data source identifier), timestamp (ISO 8601 timestamp), data_type (such as "device_vibration" or "video_helmet_rate"), content (specific numerical or textual data), and version (data version)). The time-series alignment submodule performs timestamp alignment and interpolation compensation on data collected at different frequencies. (Using a 1-minute time window, data from each data source is mapped to the corresponding window, and linear interpolation is used to compensate for missing points within the window. For example, if the sensor collects data every 1 second, the average of 60 data points within that minute is taken as the representative value for that minute; if the video analysis outputs the helmet wearing rate every 10 seconds, the average of 6 data points within that minute is taken.) Data association and fusion submodule: Based on spatiotemporal correlation, multi-source data is fused into a structured dataset (using "enterprise ID + time window (minutes)" as the key, all cleaned, standardized and aligned data are integrated into a single structured secure data record, which is stored in a cloud time-series database (such as InfluxDB)).

[0025] The intelligent risk assessment engine connects to the data fusion and governance layer and is used to perform dynamic risk calculation and graded early warning based on the structured security dataset and a pre-trained risk assessment model to generate risk assessment results. The intelligent risk assessment engine includes: Risk Feature Extraction Submodule: Extracts security risk feature vectors from structured datasets, as follows: Equipment characteristics: Calculate the root mean square value of the vibration frequency. Peak Temperature change rate Current harmonic distortion rate (THD), etc.

[0026] Environmental characteristics: extraction temperature Humidity (H) and concentration of harmful gases Dust concentration The real-time value and its deviation from the safety threshold.

[0027] Behavioral characteristics: Video analysis using the YOLOv8 object detection model to obtain helmet wearing rate. Seatbelt wearing rate The number of violations was counted using an attitude estimation model. Number of times entering dangerous areas (Unit: times / hour)

[0028] Management characteristics: Obtain the completion rate of hazard rectification from the enterprise management system. Safety training coverage Safety inspection completion rate Emergency response plan drill completion rate (All figures are expressed as percentages).

[0029] Dynamic risk assessment submodule: Based on the aforementioned feature vector, calculate the real-time security risk index R. The calculation model for the real-time security risk index R is as follows: ; in, The equipment safety risk sub-index is obtained by comparing the collected parameters such as equipment vibration frequency, operating temperature, current, and voltage with the normal operating range, calculating the deviation, and then weighting and normalizing them. The environmental safety risk sub-index is obtained by comparing parameters such as ambient temperature, humidity, concentration of harmful gases, and dust concentration with safety thresholds, calculating the degree of exceedance, and then weighting and normalizing the results. The sub-index of personnel behavior safety risk is obtained by normalizing the safety equipment wearing rate and the number of violations through video surveillance analysis using a behavior recognition model. The safety risk sub-index is obtained by comprehensively evaluating management data such as the completion rate of hazard rectification and the coverage rate of safety training.

[0030] Among them: Equipment safety risk sub-index The calculation model is as follows: Assume the normal operating range of the equipment is the vibration frequency. ,temperature Current ,Voltage Define the deviation of each parameter: Similarly, we can obtain ;but: ; where weight The values ​​are set by experts based on the device type, and the sum is 1.

[0031] Environmental safety risk sub-index The calculation model is as follows: Set a safe temperature threshold Suitable humidity range Hazardous gas threshold Dust concentration Various exceedances: ; ; ; ; .

[0032] Management of safety risk sub-index The calculation model is as follows: The coefficient here is set according to the degree of harm caused by the behavior.

[0033] Management of safety risk sub-index The calculation model is as follows: ; The weighting coefficients are determined using the AHP method, satisfying the following conditions: The value was determined by combining the analytic hierarchy process (AHP) with expert scoring, and was set to [value]. ; The graded early warning submodule divides risks into low risk (0≤R<30), medium risk (30≤R<60), high risk (60≤R<80), and major risk (80≤R≤100) based on the R value range, and generates four levels of early warning: blue, yellow, orange, and red. Risk Trend Prediction Submodule: Using an LSTM model, input the daily average R-value sequence of the past 30 days and output the predicted value for the next 7 days to help enterprises deploy preventive measures in advance.

[0034] The government-enterprise collaborative service module is connected to the intelligent risk assessment engine and is used to push differentiated security service information and collaborative handling tasks to the government and enterprises respectively based on the risk assessment results. The government-enterprise collaborative service module includes: Government-side service sub-module: Provides an overview of enterprise security status, risk classification statistics, law enforcement inspection suggestions, and emergency response command; The enterprise-side service submodule provides real-time risk alerts, hazard identification guidance, safety training updates, and rectification task management. (For example, when an enterprise's R value reaches medium risk (e.g., R=45), the collaborative task scheduling submodule automatically generates an "enterprise rectification task" (requiring completion within 48 hours) and a "government verification task" (requiring on-site verification within 24 hours), and pushes these to the responsible parties via the app and SMS. Simultaneously, the enterprise-side service submodule pushes targeted hazard identification guidelines and training courses; the government-side service submodule highlights the enterprise's status on the monitoring dashboard.)

[0035] The intelligent question-answering submodule is built upon open-source large language models (such as ChatGLM) and a local knowledge base (including laws, regulations, standards, accident cases, etc.). When an enterprise user asks "How to handle abnormal vibration in a reactor," the system retrieves relevant documents using RAG and generates an answer that includes cause analysis, inspection steps, and remedial measures. Collaborative task scheduling submodule: automatically generates collaborative disposal tasks and accurately pushes them according to risk levels and event types.

[0036] Response quality improvement closed-loop management module, connected to said government-enterprise collaborative service module, configured to perform tracking and recording, effect evaluation and feedback optimization on the whole execution process of security service information and collaborative disposal tasks, and form a closed-loop management of "detection-push-disposal-evaluation-optimization".

[0037] Said response quality improvement closed-loop management module comprises: Task tracking submodule: records the receiving, processing and completion status of tasks and timestamps of each link in real time; Effect evaluation submodule: after the task is completed, calculates the service quality improvement effect evaluation value E, and the calculation model of the service quality improvement effect evaluation value E is as follows: ; wherein, is the real-time security risk index before task execution, is the real-time security risk index after the task is executed and completed and passes a preset evaluation cycle (preferably 7 days); when E>0, it indicates that the risk has decreased, and the larger the E value, the more significant the quality improvement effect; when E≤0, it indicates invalidity and triggers feedback optimization; (for example: the task tracking submodule of the response quality improvement closed-loop management module records the full life cycle status from task generation to task closing. The effect evaluation submodule recalculates the R value of the enterprise on the 7th day after the task is completed, and substitutes it into the calculation formula of the service quality improvement effect evaluation value E to obtain E. If E>20%, it is valid; if 0<E≤20%, the effect is average; if E≤0, it is invalid. The feedback optimization submodule analyzes the cause and automatically adjusts the weight coefficient for invalid or average-effect cases (for example, if the effect of equipment risk-related tasks is poor, the α value is appropriately increased to enhance the equipment risk sensitivity) or shortens the rectification period. All adjustments are recorded on the blockchain).

[0038] Feedback optimization submodule: automatically adjusts the weight parameters of the intelligent risk research and judgment engine and the collaborative service push strategy according to the E value.

[0039] Blockchain evidence storage module, connected to said data acquisition layer, said intelligent risk research and judgment engine and said response quality improvement closed-loop management module, configured to store key security data and operation records on the blockchain for evidence storage, so as to ensure the immutability and full traceability of data; Said blockchain evidence storage module stores enterprise hidden danger investigation records, rectification completion records, government law enforcement inspection records, risk early warning and disposal records, and service quality improvement effect evaluation records on the blockchain for evidence storage, and each piece of data generates a unique evidence storage ID (SHA‑256 hash), which supports one-click verification by regulatory authorities.

[0040] The system adopts a collaborative architecture of "cloud-edge-device": the cloud deploys a data fusion and governance layer, an intelligent risk assessment engine, and a blockchain evidence storage module; the edge is deployed on the enterprise side, running a lightweight data preprocessing and real-time risk assessment model to achieve millisecond-level local response; the terminals include government regulatory terminals and enterprise terminals, used for data collection and information display.

[0041] Example 2: This application provides a specific application example: On July 6, 2026, at 9:00 AM, a chemical company A's system collected data showing a reactor vibration frequency f=68 (normal range 45±5Hz) and a temperature T=85℃ (normal ≤80°C). Video analysis was used to determine the helmet wearing rate. =80% (100% required), number of violations = 2 times / hour, hazard rectification completion rate =70%. Calculated... =72, , =45, =35, and based on the weights, R = 0.25×72 + 0.25×15 + 0.30×45 + 0.20×35 = 42.25, which is a medium risk (yellow warning). The system pushed a rectification task, and the company completed the bearing replacement and personnel training on the same day. Seven days later, R dropped to 18.6, and E = (42.25-18.6) / 42.25×100% = 56.0%, showing significant results. The data was stored on the blockchain for evidence.

[0042] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A safety production service quality improvement response system for government-enterprise collaboration, characterized in that, include: The data acquisition layer is used to collect safety production-related data from multiple data sources, including IoT sensors deployed at the enterprise end, video surveillance equipment, enterprise safety production management systems, government regulatory databases, insurance institution data systems, and meteorological and environmental data platforms. The data fusion and governance layer, connected to the data acquisition layer, is used to perform data cleaning, format standardization, time-series alignment, and correlation fusion processing on the acquired multi-source heterogeneous data to generate a structured and secure dataset in a unified format. The intelligent risk assessment engine connects to the data fusion and governance layer and is used to perform dynamic risk calculation and graded early warning based on the structured security dataset and a pre-trained risk assessment model to generate risk assessment results. The government-enterprise collaborative service module is connected to the intelligent risk assessment engine and is used to push differentiated security service information and collaborative handling tasks to the government and enterprises respectively based on the risk assessment results. The response quality improvement closed-loop management module is connected to the government-enterprise collaborative service module. It is used to track, record, evaluate, and optimize the entire process of security service information and collaborative handling tasks, forming a closed-loop management of "discovery-push-handling-evaluation-optimization". The blockchain evidence storage module connects the data acquisition layer, the intelligent risk assessment engine, and the response quality improvement closed-loop management module, and is used to store key security data and operation records on the blockchain for evidence storage.

2. The safety production service quality improvement response system for government-enterprise collaboration according to claim 1, characterized in that, The data fusion and governance layer includes: The data cleaning submodule is used to identify and process missing values, outliers, and duplicate values ​​in multi-source heterogeneous data; The format standardization submodule is used to convert different data formats from different data sources into a unified standard data format. The timing alignment submodule is used to perform timestamp alignment and interpolation compensation processing on data from different acquisition frequencies. The data association and fusion submodule is used to fuse cleaned, standardized, and aligned multi-source data based on the spatiotemporal relationships between data to generate a structured and secure dataset.

3. The safety production service quality improvement response system for government-enterprise collaboration according to claim 1, characterized in that, The intelligent risk assessment engine includes: Risk Feature Extraction Submodule: Extracts security risk feature vectors from structured datasets; Dynamic risk assessment submodule: Based on the aforementioned feature vector, calculate the real-time security risk index R. The calculation model for the real-time security risk index R is as follows: ； in, The equipment safety risk sub-index is obtained by comparing the collected parameters such as equipment vibration frequency, operating temperature, current, and voltage with the normal operating range, calculating the deviation, and then weighting and normalizing them. The environmental safety risk sub-index is obtained by comparing parameters such as ambient temperature, humidity, concentration of harmful gases, and dust concentration with safety thresholds, calculating the degree of exceedance, and then weighting and normalizing the results. The sub-index of personnel behavior safety risk is obtained by normalizing the safety equipment wearing rate and the number of violations through video surveillance analysis using a behavior recognition model. The safety risk sub-index is obtained by comprehensively evaluating management data such as the completion rate of hazard rectification and the coverage rate of safety training. Let be the weighting coefficient, satisfying The determination was made through a combination of the analytic hierarchy process and expert scoring. The graded early warning submodule is used to classify the risk level into four levels: low risk, medium risk, high risk, and major risk, based on the numerical range of the real-time security risk index R, and generate corresponding graded early warning information. The risk trend prediction submodule is used to predict the trend of safety risks within a preset time period based on historical risk data and through a time-series prediction model.

4. A safety production service quality improvement response system for government-enterprise collaboration as described in claim 3, characterized in that, The graded early warning submodule classifies risk levels based on the numerical range of the real-time security risk index R: When 0 ≤ R < 30, it is judged as low risk, and a blue warning message is generated; When 30≤R<60, it is judged as medium risk and a yellow warning message is generated; When 60≤R<80, it is judged as high risk and an orange warning message is generated; When 80≤R≤100, it is judged as a major risk and a red warning message is generated.

5. A safety production service quality improvement response system for government-enterprise collaboration as described in claim 4, characterized in that, The government-enterprise collaborative service module includes: The government-side service submodule is used to provide government regulatory departments with an overview of enterprise security status, risk classification and distribution statistics, law enforcement inspection suggestions, and emergency response command functions. The enterprise-side service submodule is used to provide enterprises with real-time risk alerts, guidance on hazard identification and management, safety training course delivery, and rectification task management functions. The intelligent question-and-answer submodule, based on a large language model and retrieval enhancement generation technology, provides enterprises with intelligent question-and-answer services, including interpretation of safety production regulations, guidance on hazard identification, and consultation on emergency response plans. The collaborative task scheduling submodule is used to automatically generate collaborative handling tasks based on risk level and event type, and push the tasks to relevant responsible persons on the government and enterprise sides.

6. A safety production service quality improvement response system for government-enterprise collaboration as described in claim 1, characterized in that, The response quality improvement closed-loop management module includes: Task tracking submodule: Records the receipt, processing, and completion status of tasks, as well as timestamps for each stage, in real time; The effectiveness evaluation submodule calculates the service quality improvement effectiveness evaluation value E after the task is completed. The calculation model for the service quality improvement effectiveness evaluation value E is as follows: ;in, This is a real-time security risk index before the task is executed. This is the real-time security risk index after the task is completed and a preset evaluation period (preferably 7 days); when E>0, it indicates a decrease in risk, and the larger the E value, the more significant the quality improvement effect; when E≤0, it indicates ineffectiveness and triggers feedback optimization. Feedback Optimization Submodule: Automatically adjusts the weight parameters of the intelligent risk assessment engine and the collaborative service push strategy based on the E-value.

7. A safety production service quality improvement response system for government-enterprise collaboration as described in claim 1, characterized in that, The blockchain evidence storage module stores the following key data on the blockchain: enterprise hidden danger investigation records, rectification completion records, government law enforcement inspection records, risk warning and disposal records, and service quality improvement effect evaluation records. Each piece of data on the blockchain generates a unique evidence storage ID, which supports regulatory authorities to verify the data through the evidence storage ID.

8. A safety production service quality improvement response system for government-enterprise collaboration as described in claim 1, characterized in that, The system is deployed using a "cloud-edge-device" collaborative architecture. The data fusion and governance layer, the intelligent risk assessment engine, and the blockchain evidence storage module are deployed in the cloud. The edge is deployed on the enterprise side, with lightweight data preprocessing and real-time risk assessment models, achieving millisecond-level local response; The terminals include government regulatory terminals and enterprise terminals, which are used for data collection and information display.

Images