Non-resident user safety supervision system based on smart gas

Abstract

The invention discloses a non-resident user safety supervision system based on smart gas, and relates to the technical field of smart gas safety supervision, and the system comprises a smart gas non-resident user service platform which is used for providing a multi-terminal convenient operation entrance for non-resident users through a user interaction module, precise contact of self-inspection reminding and hidden danger early warning is achieved by means of a message pushing module, storage, query and visual analysis of user archives, self-inspection and rectification data are completed by means of a data management module, and a user side full-process service closed loop is formed; the intelligent fuel gas non-resident user supervision platform is used for carrying out intelligent identification and model optimization on hidden danger data through an AI analysis module, generating a grading inspection plan by means of a supervision decision module, supporting cross-department cooperative supervision, and realizing data security interaction with a service platform, intelligent equipment and a third-party system by using a data interface module. And an intelligent and collaborative supervision system is constructed. According to the invention, full-process management and control of gas consumption safety of non-resident users can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent gas safety supervision, and specifically relates to a non-residential user safety supervision system based on intelligent gas. Background Art

[0002] The following problems exist in the current gas safety supervision of non-residential users:

[0003] 1. At the level of user self-inspection, non-residential users (such as catering places, industrial and commercial enterprises) generally lack unified hidden danger investigation guidelines and digital tools, and mainly rely on manual inspection records or paper ledgers, resulting in the following defects: (i) Vague self-inspection standards: Different users have different understandings of gas facility safety hidden dangers (such as the judgment of risk points such as hose aging and valve leakage), and some users do not even conduct self-inspections regularly, and only judge the safety status based on experience. (ii) Inefficient reporting channels: After discovering hidden dangers, users need to report them to the supervision department or gas supply enterprise through traditional methods such as phone calls, WeChat, or filling out forms offline. The information transmission is lagged and key details (such as the location of hidden dangers, severity, on-site photos, etc.) are easily omitted. (iii) Difficult to track rectification: The supervision department cannot grasp the progress of hidden danger rectification in real time, and needs to conduct multiple manual follow-up visits to confirm. There is a lack of effective tracking means for users who refuse to rectify or whose rectification is not in place, resulting in a long hidden danger closed-loop management cycle, and even forming a loophole of "checking but not rectifying". Typical scenario: A certain catering merchant found that the gas pipeline in the kitchen was slightly rusty, but because he didn't know whether it was a major hidden danger, he didn't report it in time; the problem was only discovered by the supervision department during a on-site inspection two weeks later, and there was a risk of gas leakage during this period.

[0004] 2. The supervision department relies on manual grading, the risk assessment is highly subjective, and there is a lack of a data-driven quantitative model, resulting in the following defects: (i) Single evaluation dimension: Only classify based on basic information such as user type (such as catering, industry), gas consumption, etc., without integrating dynamic indicators such as historical hidden danger records, equipment operation data (such as the alarm frequency of alarms), and user self-inspection cooperation, resulting in the risk level not being able to truly reflect the actual safety status. (ii) Significant subjective influence: Different supervision personnel may have different risk judgments for the same user. For example, for the indicator of "the service life of the hose exceeds 18 months", some personnel judge it as a high risk, while some personnel regard it as a general hidden danger, lacking a unified quantitative standard. (iii) The value of data is not exploited: The accumulated massive inspection records and hidden danger data are not analyzed and modeled through big data, and the potential risk rules (such as the correlation between the valve leakage incidence rate of a certain type of venue and environmental humidity) cannot be identified, making it difficult to early warn high-risk areas or user groups. Typical scenario: Two catering enterprises of similar scale, enterprise A had 3 gas leakage alarms due to improper operation in the past year, and enterprise B had no alarm records. However, due to the lack of a data model by the supervision department, the two were still classified into the same risk level, and key supervision was not implemented for enterprise A.

[0005] 3. The inspection plan is configured in a rough manner, failing to implement differential supervision for users with different risk levels, resulting in drawbacks such as "one-size-fits-all" or "focusing on the big and neglecting the small" in inspection work: (i) Insufficient supervision of high-risk users: For users with long-term major hidden dangers and inactive rectification, the inspection frequency is not increased or the supervision measures are not upgraded, resulting in the continuous accumulation of their risks. (ii) Over-inspection of low-risk users: For users with good compliance records and low risk levels, they are still inspected at the same unified frequency, wasting the manpower and material resources of supervision and possibly interfering with the normal operation of users; (iii) Lack of dynamic adjustment mechanism: Once the inspection plan is formulated, it is difficult to adjust it in real time. For example, if a user newly has major hidden dangers during a certain inspection, the supervision department cannot timely adjust it from "low risk" to "high risk" and increase the inspection frequency. Typical scenario: In an industrial park, a high-risk enterprise using a gas boiler (which has had an explosion accident) and a low-risk convenience store using only small gas cookers are both set to be inspected once a quarter. The former fails to detect new hidden dangers in a timely manner due to insufficient inspection frequency, while the latter is affected by frequent inspections and its operation efficiency is reduced. Summary of the Invention

[0006] In view of the technical problems that the current supervision method for non-resident users leads to low inspection efficiency, rectification efficiency, and work efficiency, the present invention provides a non-resident user safety supervision system based on intelligent gas.

[0007] The technical solution adopted by the non-resident user safety supervision system based on intelligent gas of the present invention to solve the above technical problems is as follows:

[0008] A non-resident user safety supervision system based on intelligent gas, which includes:

[0009] An intelligent gas non-resident user service platform, which is used to provide a convenient operation entry for non-resident users through a user interaction module, collect in real time the basic information of registered non-resident users, gas consumption data of intelligent gas meters, equipment operation status, and hidden danger self-inspection data through a data collection module to form a standardized data set, achieve the accurate reach of self-inspection reminders and hidden danger warnings through a message push module, and complete the storage, query, and visual analysis of user files, self-inspection, and rectification data through a data management module to form a full-process service closed-loop at the user side;

[0010] An intelligent gas non-resident user supervision platform, which is used to receive the standardized data set of the intelligent gas non-resident user service platform, perform intelligent identification and model optimization on hidden danger data through an AI analysis module, generate a hierarchical inspection plan and support cross-departmental collaborative supervision relying on a supervision decision module, and realize data security interaction with the service platform, intelligent devices, and third-party systems through a data interface module to build an intelligent and collaborative supervision system.

[0011] Optionally, the user interaction module involved provides functions of QR code registration, self-inspection photo uploading, and rectification feedback, where:

[0012] Non-resident users complete registration by scanning the exclusive QR code on the gas pipeline / equipment or entering enterprise information on the Web / App / miniprogram side. The system automatically generates a unique user identifier and associates the geographical coordinates with the gas-using equipment ledger to form a digital user file.

[0013] Non-resident users take photos or videos on-site through the App / miniprogram according to the self-inspection list pushed by the service platform. The system automatically adds watermarks containing user identifiers, timestamps, and GPS locations to the multimedia files to ensure the integrity of the data. Subsequently, after the non-resident users confirm that it is correct, they can upload it to the service platform with one click.

[0014] For the potential hazards warned by the supervision platform, after the non-resident users complete the rectification, they take photos after rectification and attach written explanations, and upload them to the service platform with one click to apply for a review.

[0015] Optionally, based on the non-resident user types and potential hazard history records, the message push module presets reminder strategies for self-inspection reminders and potential hazard warnings through a rule engine.

[0016] The reminder strategy for self-inspection reminders is: N days before the start of the self-inspection cycle, push the inspection list and deadline to non-resident users through at least one of SMS, in-app messages, and automatic outbound voice calls.

[0017] The reminder strategy for potential hazard warnings is: when the AI analysis module identifies a potential hazard, it triggers a warning message in real time, pushes the details of the potential hazard to non-resident users, and simultaneously notifies the relevant person in charge of non-resident users.

[0018] Further optionally, the data management module involved provides data storage and query functions, as well as statistical analysis and visualization functions, where:

[0019] The data management module encrypts and stores non-resident user registration information, self-inspection photos / videos, and rectification records through the data storage function, and supports quick retrieval by time, non-resident user name, and potential hazard type through the data query function.

[0020] The data management module automatically generates reports covering non-resident user self-inspection completion rates, potential hazard type distributions, and rectification timeliness rates through statistical analysis and visualization functions, and visually displays them on the Web side through line charts and pie charts.

[0021] Further optionally, the AI analysis module involved integrates a potential hazard identification model and an evaluation and analysis model, where:

[0022] Train a hidden danger identification model based on the YOLO object detection algorithm. The hidden danger identification model automatically identifies the hidden dangers in the self-inspection photos and outputs the types of hidden dangers. When the supervisor manually annotates and corrects the output results, the annotated data is regularly returned to the training set to optimize the hidden danger identification model.

[0023] Combined with the standardized data set collected in real time by the service platform, based on the analytic hierarchy process, invite gas safety experts to score the weights of four indicators: the credit rating of industrial and commercial households, the gas use stability, the equipment qualification rate, and the timeliness of hidden danger rectification, and establish a user safety evaluation and analysis model. The evaluation and analysis model is used to calculate the safety scores of non-resident users based on the input data, and implement a three-level red, yellow, and green coding for non-resident users based on the calculated safety scores, providing a quantitative basis for supervision decision-making.

[0024] Further optionally, the involved supervision decision-making module automatically generates a differentiated inspection plan according to the three-level red, yellow, and green coding implemented by the evaluation and analysis model integrated in the AI analysis module for non-resident users.

[0025] The supervision decision-making module automatically identifies the hidden dangers according to the hidden danger identification model integrated in the AI analysis module, and dynamically increases the inspection frequency of this area in combination with the number of occurrences of this type of hidden danger in a certain area. Subsequently, it shares the hidden danger information with the emergency management department and the fire department through the data interface module, and jointly carries out special rectification actions.

[0026] Further optionally, the involved data interface module supports internal data interaction, specifically manifested as: real-time synchronization of user registration, self-inspection, and rectification data with the service platform to ensure the consistency of information between the service platform and the supervision platform; docking with intelligent gas meter data to warn of potential leakage risks through abnormal flow.

[0027] Further optionally, the involved data interface module supports external data docking, specifically manifested as: encrypted data transmission with a third-party supervision system through an API interface, supporting cross-departmental data sharing, and simultaneously obtaining external policy documents and industry standard data to update the supervision rules of the supervision platform.

[0028] Further optionally, the specific implementation process of the involved safety supervision system includes the following steps:

[0029] S1. Non-resident users complete registration through the Web / App / miniprogram terminal of the intelligent gas non-resident user service platform by scanning the exclusive two-dimensional code of the gas equipment or manually entering enterprise information. After completing the registration, a unique user identifier is automatically generated, and the geographical coordinates and the gas use equipment ledger are associated to create a digital user file and store it in the data management module.

[0030] S2. Through the service platform, basic information of registered non-residential users, smart gas meter gas usage data, equipment operating status, and hidden danger self-inspection data are collected in real time to form a standardized data set and encrypted and synchronized to the smart gas non-residential user supervision platform;

[0031] S3. Based on the Analytic Hierarchy Process (AHP), gas safety experts were invited to assign weighted scores to four indicators: business credit rating, gas usage stability, equipment qualification rate, and timely rectification of hidden dangers, to establish a user safety evaluation and analysis model.

[0032] S4. Input the standardized data set into the evaluation and analysis model, which automatically calculates the user safety score and assigns a red, yellow, or green code to non-resident users based on the calculated safety score, thereby achieving user risk level classification;

[0033] S5. The supervision decision module of the supervision platform automatically generates differentiated inspection plans based on the red, yellow, and green three-level codes implemented for non-resident users based on the evaluation and analysis model integrated by the AI analysis module;

[0034] S6. The message push module uses a rule engine to push self-check reminders based on the user's risk level and the contact information reserved during registration. It supports multiple channels such as SMS, App station messages, and voice calls, and comes with a personalized checklist.

[0035] S7. Non-resident users can use the service platform's user interaction module to take photos or videos of the site according to the checklist. The system automatically adds a watermark containing the user's ID, timestamp, and GPS location to the multimedia file to prevent data tampering. After uploading, the supervision platform's AI analysis module calls the hidden danger identification model to perform intelligent analysis on the image and output a report on the hidden danger type and location.

[0036] S8. For non-resident users who have not conducted self-inspections, the message push module uses the self-inspection reminder strategy preset by the rule engine to push the checklist and deadline to non-resident users through at least one of SMS, App station message, and automatic voice call N days before the start of the self-inspection period; For users who have discovered hidden dangers during self-inspection but have not rectified them, the message push module uses the self-inspection hidden danger warning reminder strategy preset by the rule engine to push the hidden danger details to non-resident users when the AI analysis module identifies the hidden danger and triggers the warning message in real time, and simultaneously notifies the relevant person in charge of the non-resident user. After the user completes the rectification, he / she uploads before-and-after comparison photos and explanations. After the supervisory personnel review the hidden danger on site, the hidden danger status is updated and archived in the data management module;

[0037] S9. The supervision decision-making module regularly analyzes the full amount of hidden danger data and organizes experts to optimize the indicator weights of the evaluation and analysis model; shares high-risk user data with the fire protection and market supervision departments through the data interface module, jointly carries out special rectification actions on gas safety, and realizes cross-departmental supervision coordination.

[0038] A non-resident user safety supervision system based on intelligent gas of the present invention has the following beneficial effects compared with the prior art:

[0039] 1. Through the QR code registration for industrial and commercial households and the technology of uploading self-inspection data with watermarks, the present invention ensures the authenticity and traceability of self-inspection data. Combining with the AI hidden danger analysis and early warning mechanism, it realizes the full-process closed-loop management of hidden danger investigation and rectification, and improves the timely rate of hidden danger rectification by more than 40%, forming an efficient closed-loop of "self-inspection - early warning - rectification - review".

[0040] 2. By collecting four types of core data, namely basic information, gas consumption data, equipment information, and hidden danger data, the present invention constructs an evaluation and analysis model based on the analytic hierarchy process to objectively quantify the user risk level, solves the problems of strong subjectivity and inconsistent standards in traditional manual grading, and provides data support for precise supervision.

[0041] 3. Based on the classification and code assignment results of user safety scores, the present invention implements differential inspection plans for users with green codes (low risks), yellow codes (medium risks), and red codes (high risks), so that the inspection coverage rate of high-risk users reaches 100%, while reducing the supervision cost of low-risk users by 30%, and significantly improving the resource utilization efficiency.

[0042] 4. The present invention uses a hidden danger recognition model trained based on the YOLO object detection algorithm to realize intelligent hidden danger recognition, combines with a dynamically updated evaluation model to predict the risk trend, supports the transformation of supervision decision-making from "experience-driven" to "data-driven", improves the safety supervision efficiency by more than 50%, and effectively reduces the risks of manual misjudgment and missed inspection. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Attached Figure 1 is a schematic diagram of the system architecture of Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0044] In order to make the technical solutions, technical problems to be solved, and technical effects of the present invention clearer and more understandable, the following combines specific embodiments to clearly and completely describe the technical solutions of the present invention.

[0045] Embodiment 1:

[0046] Referring to Attached Figure 1 , this embodiment proposes a non-resident user safety supervision system based on intelligent gas, which includes:

[0047] The intelligent gas non-residential user service platform is used to provide non-residential users with convenient operation entrances for multiple terminals through the user interaction module, collect the basic information of registered non-residential users, the gas consumption data of intelligent gas meters, the equipment operation status and the hidden danger self-inspection data in real time through the data collection module, form a standardized data set, achieve the accurate delivery of self-inspection reminders and hidden danger warnings through the message push module, and use the data management module to complete the storage, query and visual analysis of user files, self-inspection and rectification data, forming a full-process service closed-loop at the user end;

[0048] The intelligent gas non-residential user supervision platform is used to receive the standardized data set of the intelligent gas non-residential user service platform, conduct intelligent identification and model optimization of the hidden danger data through the AI analysis module, generate a hierarchical inspection plan based on the supervision decision-making module and support cross-departmental collaborative supervision, and use the data interface module to achieve secure data interaction with the service platform, intelligent devices and third-party systems, constructing an intelligent and collaborative supervision system.

[0049] This embodiment specifically describes the modules included in the intelligent gas non-residential user service platform:

[0050] The involved user interaction module provides functions such as QR code registration, self-inspection photo upload and rectification feedback. Among them:

[0051] Non-residential users complete registration by scanning the exclusive QR code on the gas pipeline / equipment or entering enterprise information on the Web / App / miniprogram side. The system automatically generates a unique user identifier and associates the geographical coordinates with the gas consumption equipment ledger, forming a digital user file;

[0052] Non-residential users take photos or videos on-site through the App / miniprogram according to the self-inspection list pushed by the service platform. The system automatically adds a watermark containing the user identifier, timestamp and GPS positioning to the multimedia file to ensure the data cannot be tampered with. Subsequently, after the non-residential user confirms it is correct, it is uploaded to the service platform with one key;

[0053] For the hidden dangers warned by the supervision platform, after the non-residential users complete the rectification, they take photos of the rectification and attach a written description, and upload it to the service platform with one key to apply for a review.

[0054] Based on the non-residential user type and the hidden danger history record, the message push module presets the reminder strategies for self-inspection reminders and hidden danger warnings through the rule engine;

[0055] The reminder strategy for self-inspection reminders is: N days before the start of the self-inspection cycle, push the inspection list and the deadline to non-residential users through at least one of SMS, in-app messages on the App, and automatic outbound voice calls;

[0056] The reminder strategy for potential hazard warning is as follows: When the AI analysis module identifies a potential hazard, it triggers a warning message in real time, pushes the details of the potential hazard to non-resident users, and simultaneously notifies the relevant person in charge of non-resident users.

[0057] The involved data management module provides data storage and query functions, as well as statistical analysis and visualization functions, where:

[0058] The data management module encrypts and stores the registration information of non-resident users, self-inspection photos / videos, and rectification records through the data storage function, and supports quick retrieval by time, non-resident user name, and potential hazard type through the data query function;

[0059] The data management module automatically generates reports covering the self-inspection completion rate of non-resident users, the distribution of potential hazard types, and the rectification timeliness rate through the statistical analysis and visualization functions, and visually displays them on the Web side through line charts and pie charts.

[0060] This embodiment specifically describes the modules included in the intelligent gas non-resident user supervision platform:

[0061] The involved AI analysis module integrates a potential hazard identification model F(X) and an evaluation and analysis model M, where:

[0062] The potential hazard identification model F(X) is trained based on the YOLO object detection algorithm. The potential hazard identification model automatically identifies potential hazards in self-inspection photos and outputs the types of potential hazards; when the supervisor manually annotates and corrects the output results, the annotated data is periodically returned to the training set to optimize the potential hazard identification model F(X);

[0063] Combined with the standardized data set collected in real time by the service platform, based on the analytic hierarchy process, gas safety experts are invited to score the weights of four indicators: the credit rating of industrial and commercial households, the gas use stability, the equipment qualification rate, and the potential hazard rectification timeliness, to establish a user safety evaluation and analysis model M; the evaluation and analysis model M is used to calculate the safety score of non-resident users based on the input data, and implement a three-level red-yellow-green coding for non-resident users based on the calculated safety score, providing a quantitative basis for supervision decisions.

[0064] The involved supervision decision module automatically generates a differentiated inspection plan according to the three-level red-yellow-green coding implemented by the evaluation and analysis model M integrated in the AI analysis module for non-resident users; for example, a) for non-resident users with a red code (i.e., high risk), set on-site inspections once every quarter, covering all self-inspection items, b) for non-resident users with a yellow code (i.e., medium risk), set on-site inspections once every six months, randomly checking 50% of the self-inspection items, c) for non-resident users with a green code (i.e., low risk), set on-site inspections once a year, randomly checking 20% of the self-inspection items;

[0065] Based on the potential hazards automatically identified by the hazard identification model F(X) integrated in the AI analysis module, the supervision decision-making module dynamically increases the inspection frequency of a certain area in combination with the occurrence times of this type of hazard in that area. Subsequently, it shares the hazard information with the emergency management department and the fire department through the data interface module, and jointly carries out special rectification actions.

[0066] The involved data interface module supports internal data interaction, specifically manifested as: real-time synchronization of user registration, self-inspection, and rectification data with the service platform to ensure the consistency of information between the service platform and the supervision platform; docking with smart gas meter data to warn of potential leakage risks through abnormal flow. The involved data interface module supports external data docking, specifically manifested as: encrypted data transmission with a third-party supervision system through an API interface, supporting cross-departmental data sharing, and simultaneously obtaining external policy documents and industry standard data to update the supervision rules of the supervision platform.

[0067] Embodiment 2:

[0068] Based on the safety supervision system of Embodiment 1, this embodiment proposes a safety supervision method, and the specific implementation process includes the following steps:

[0069] S1. Non-resident users complete registration by scanning the exclusive QR code of the gas equipment or manually entering enterprise information through the Web / App / miniprogram terminal of the intelligent gas non-resident user service platform. After completing the registration, a unique user identifier is automatically generated, and the geographical coordinates and gas equipment ledger are associated to create a digital user file and store it in the data management module.

[0070] When specifically executing step S1, non-resident users complete registration by scanning the exclusive QR code of the gas equipment or manually entering enterprise information through the Web / App / miniprogram terminal of the intelligent gas non-resident user service platform. The registration information of industrial and commercial households is not limited to the basic information of industrial and commercial households (such as business license number, contact information of the person in charge), gas usage type (industrial / commercial / catering), gas supply unit, brand model and installation time of the gas stove, etc. After the system verifies the integrity of the information, a unique user ID is generated and an electronic file is created, associating the geographical location with the gas equipment ledger.

[0071] S2. Through the service platform, the basic information, gas usage data of smart gas meters, equipment operation status, and hazard self-inspection data of registered non-resident users are collected in real time, forming a standardized data set and encrypted and synchronized to the intelligent gas non-resident user supervision platform.

[0072] When specifically executing step S2, assume that the standardized data set is represented as D = {D1, D2, D3, D4}. Among them, D1 represents the basic information of industrial and commercial households (including the registration information of industrial and commercial households, the term of the gas use contract, credit rating, etc.). Specifically, the average monthly gas consumption, instantaneous flow rate, and peak and valley periods of gas use can be obtained in real time by docking with the data of the gas supply unit or the API of the intelligent gas meter; D2 represents the gas use data of industrial and commercial households (including the average monthly gas consumption, peak load period, seasonal fluctuation coefficient, etc.). Specifically, the average monthly gas consumption, instantaneous flow rate, and peak and valley periods of gas use can be obtained in real time by docking with the data of the gas supply unit or the API of the intelligent gas meter; D3 represents the equipment information (including the model of the cooking stove / alarm, installation time, regular inspection report, etc.). Specifically, users can be required to regularly upload the inspection report of the cooking stove / alarm, and OCR automatic recognition of key information is supported; D4 represents the hidden danger data (including historical hidden danger types, rectification timeliness rate, re-inspection qualification rate, etc.). Specifically, it can be docked with the household safety inspection data of the gas supply unit to integrate historical inspection and rectification information.

[0073] S3. Based on the analytic hierarchy process, invite gas safety experts to score the weights of four indicators: the credit rating of industrial and commercial households, gas use stability, equipment qualification rate, and hidden danger rectification timeliness, and establish a user safety evaluation analysis model M.

[0074] When specifically executing step S3, assume that the expression of the evaluation analysis model M established by the user is as follows:

[0075]

[0076] In the formula, w i represents a preset parameter, and the sum of all ws is equal to 1. For example, specifically set w1 = 0.1, w2 = 0.3, w3 = 0.2, w4 = 0.4.

[0077] Use the evaluation analysis model M to standardize each indicator. For example:

[0078] Gas use stability: f(D i ) = 100 - 20 × coefficient of variation (coefficient of variation = standard deviation / mean);

[0079] Hidden danger rectification timeliness rate:

[0080] S4. Input the standardized data set into the evaluation analysis model M. The evaluation analysis model M automatically calculates the user safety score, and assigns red, yellow, and green three-level codes to non-resident users based on the calculated safety score, so as to realize the classification of user risk levels.

[0081] When executing step S4, assume that the standardized data set D is input into the evaluation analysis model M to calculate the user safety score: The scoring range is 0 - 100, and the higher the score, the higher the safety level.

[0082] For a specific example calculation, assuming a commercial user: D1 (credit rating B, score 80), D2 (coefficient of variation 0.2, score 90), D3 (equipment inspection passed, score 95), D4 (correction timeliness rate 85%, score 85), then SCORE = 0.1×80+0.3×90+0.2×95+0.4×85=88, corresponding to a green code (low risk).

[0083] S5. The supervision decision-making module of the supervision platform automatically generates a differentiated inspection plan based on the red, yellow and green three-level coding implemented for non-resident users according to the evaluation and analysis model M integrated by the AI analysis module.

[0084] Automatically assign codes based on the scoring results. The rules are as follows:

[0085] Green Code (low risk, SCORE ≥ 80): indicates that the user's safety management standards are met and priority is given to inclusion in the low-risk user list, with reduced on-site inspection frequency and increased remote monitoring data push. For example, for non-residential users with a green code (i.e. low risk), an annual on-site inspection is set, with 20% of self-inspection items randomly checked.

[0086] Yellow code (medium risk, 60≤SCORE<80): indicates historical risks or management loopholes, marking users on the medium-risk list. During inspections, portable testing equipment must be brought for in-depth testing. For example, for non-residential users with yellow codes (i.e., medium risk), on-site inspections are scheduled every six months, with 50% of self-inspection items randomly checked.

[0087] Red code (high risk, SCORE < 60): indicates that there are currently unrectified hidden dangers or high-frequency hidden dangers, and the user is marked as a high-risk user. The inspection plan is synchronized with the emergency management department, and joint law enforcement is implemented when necessary; for example, for non-residential users with a red code (i.e. high risk), an on-site inspection is set once every quarter to cover all self-inspection items.

[0088] S6. The message push module pushes self-inspection reminders based on the user's risk level and the contact information of the person in charge reserved during registration through the rule engine. It supports multiple channels such as SMS, App station messages, and voice calls, and comes with a personalized inspection checklist; the self-inspection reminder content can include a gas environment inspection checklist (such as ventilation conditions, fire-fighting facilities), equipment self-inspection items (stove air tightness, hose aging degree, alarm effectiveness) and rectification deadline.

[0089] S7. Non-resident users take photos or videos of the scene according to the checklist through the user interaction module of the service platform. The system automatically adds a watermark containing user identification, timestamp, and GPS location to the multimedia file to prevent data tampering, forming a self-check data set P = {P1, P2, ..., P m}, where P iMultimedia data corresponding to the i-th self-inspection item.

[0090] After the upload is completed, the AI analysis module of the supervision platform first preprocesses the image (grayscale conversion, noise filtering), and then uses the hidden danger recognition model F(X) trained based on the YOLO object detection algorithm to perform image analysis. The hidden danger recognition model F(X) outputs the hidden danger type and location report, and then classifies and stores them according to the hidden danger type to form a time series of hidden danger investigation files.

[0091] Assume that the hidden danger recognition model F(X) outputs the hidden danger type Y = {y1, y2,..., y k}, where y1 is the overuse of the hose, y2 is the alarm not being powered on, and y3 is the insufficient ventilation in the gas-using environment.

[0092] It should be added that the hidden danger recognition model F(X) is pre-trained based on the COCO dataset and fine-tuned for the gas hidden danger scenario, and supports the detection of 12 types of hidden dangers such as hose cracks, alarm indicator light status, and gas leakage at the stove interface (through the bubble detection algorithm).

[0093] S8. For non-resident users who have not conducted self-inspection (no data has been uploaded 3 days after the rectification deadline), the message push module, through the self-inspection reminder strategy preset by the rule engine, pushes the inspection list and deadline to non-resident users at least one of the ways of SMS, in-app message of the App, and automatic outbound voice calls N days before the start of the self-inspection period; for users who have hidden dangers found in self-inspection but have not been rectified (the output Y of the hidden danger recognition model F(X) is not empty, and no feedback has been received 7 days after the hidden danger is marked), the message push module, through the self-inspection hidden danger early warning reminder strategy preset by the rule engine, pushes the details of the hidden danger to non-resident users when the AI analysis module identifies the hidden danger and triggers an early warning message in real time, and simultaneously notifies the relevant person in charge of non-resident users. After the user completes the rectification, upload the comparison photos and descriptions before and after (such as "The overdue hose has been replaced, attached with the purchase certificate of the new hose"). After the supervisor on-site review is correct (such as the hose replacement time and the alarm detection report are correct), update the hidden danger status (such as marked as "rectified") and file it in the data management module.

[0094] In actual use, for users who have hidden dangers found in self-inspection but have not been rectified (the output Y of the hidden danger recognition model F(X) is not empty, and no feedback has been received 7 days after the hidden danger is marked), a three-level reminder strategy can be specifically set: ① The first-level reminder (triggered immediately): Send an SMS to the contact person of the industrial and commercial household, including the hidden danger picture, rectification requirements, and a 3-day rectification deadline; ② The second-level reminder (1 day overdue): Push the details of the hidden danger to the street safety management system and generate a grid-based supervision task; ③ The third-level reminder (3 days overdue): Report to the district and county supervision platform and include it in the government work safety supervision list.

[0095] The reminder content should include details of potential hazards, rectification requirements, and the responsibility traceability number. All reminder records can be traced on the supervision platform to form a responsibility chain, ensuring that the entire process of "discovery - notification - rectification" of potential hazards is traceable and a supervision closed-loop is formed.

[0096] S9. The supervision decision-making module regularly analyzes the full amount of potential hazard data, organizes experts to optimize the index weights of the evaluation and analysis model M; shares high-risk user data with the fire department and the market supervision department through the data interface module, and jointly conducts special rectification actions for gas safety to achieve cross-departmental supervision collaboration.

[0097] In summary, by using the non-resident user safety supervision system based on intelligent gas of the present invention, through the merchant QR code registration and the self-check data upload technology with watermark, the authenticity and traceability of the self-check data are ensured; by constructing an evaluation and analysis model based on the analytic hierarchy process, the objective quantification of the user risk level is realized, and the problems of strong subjectivity and inconsistent standards in traditional manual grading are solved; based on the grading and code assignment results of the user safety score, a differentiated inspection plan is implemented for users, significantly improving the resource utilization efficiency.

[0098] The above specific application examples have elaborated in detail the principle and implementation manner of the present invention. These examples are only used to help understand the core technical content of the present invention. Based on the above specific embodiments of the present invention, any improvements and modifications made by those skilled in the art of this technical field without departing from the principle of the present invention shall fall within the scope of the patent protection of the present invention.

Claims

1. A non-resident user safety supervision system based on intelligent gas, characterized in that It includes: The intelligent gas non-residential user service platform is used to provide non-residential users with convenient operation entrances for multiple terminals through the user interaction module, collect the basic information of registered non-residential users, the gas consumption data of intelligent gas meters, the equipment operation status and the hidden danger self-inspection data in real time through the data collection module, form a standardized data set, achieve the accurate reach of self-inspection reminders and hidden danger warnings with the help of the message push module, and use the data management module to complete the storage, query and visual analysis of user files, self-inspection and rectification data, forming a full-process service closed loop for the user side; The intelligent gas non-residential user supervision platform is used to receive the standardized data set of the intelligent gas non-residential user service platform, intelligently identify and optimize the hidden danger data through the AI analysis module, generate a hierarchical inspection plan based on the supervision decision-making module and support cross-departmental collaborative supervision, and use the data interface module to realize the secure data interaction with the service platform, intelligent devices and third-party systems, constructing an intelligent and collaborative supervision system.

2. The non-resident user safety supervision system based on intelligent gas according to claim 1, characterized in that, The user interaction module provides functions of QR code registration, self-inspection photo uploading and rectification feedback, where: Non-residential users complete registration by scanning the exclusive QR code on the gas pipeline / equipment or entering enterprise information on the Web / App / miniprogram side. The system automatically generates a unique user identifier and associates the geographical coordinates with the gas consumption equipment ledger, forming a digital user file; According to the self-inspection list pushed by the service platform, non-residential users take photos or videos on site through the App / miniprogram. The system automatically adds a watermark containing the user identifier, timestamp and GPS positioning to the multimedia file to ensure the data cannot be tampered with. Then, after the non-residential user confirms it is correct, it is uploaded to the service platform with one key; For the hidden dangers warned by the supervision platform, after the non-residential user completes the rectification, take a photo of the rectification and attach a written description, and upload it to the service platform with one key to apply for a review.

3. The non-resident user safety supervision system based on intelligent gas according to claim 2, characterized in that, Based on the non-residential user type and the hidden danger history record, the message push module presets the reminder strategies for self-inspection reminders and hidden danger warnings through the rule engine; The reminder strategy for the self-inspection reminder is: N days before the start of the self-inspection cycle, push the inspection list and the deadline to non-residential users through at least one of SMS, in-app messages on the App, and automatic outbound voice calls; The reminder strategy for the hidden danger warning is: when the AI analysis module identifies a hidden danger, it triggers a warning message in real time, pushes the hidden danger details to non-residential users, and simultaneously notifies the relevant person in charge of non-residential users.

4. The non-resident user safety supervision system based on intelligent gas according to claim 3, characterized in that, The data management module provides data storage and query functions, as well as statistical analysis and visualization functions, where: The data management module encrypts and stores the registration information of non-residential users, self-inspection photos / videos and rectification records through the data storage function, and supports quick retrieval by time, non-residential user name and hidden danger type through the data query function; The data management module automatically generates reports covering the self-inspection completion rate of non-residential users, the distribution of hidden danger types and the rectification timeliness rate through the statistical analysis and visualization function, and visually displays them on the Web side through line charts and pie charts.

5. A non-resident user safety supervision system based on intelligent gas according to any one of claims 1-4, characterized in that, The AI analysis module integrates a hidden danger identification model and an evaluation analysis model, where: Based on the YOLO object detection algorithm, a hidden danger recognition model is trained. The hidden danger recognition model automatically identifies the hidden dangers in the self-inspection photos and outputs the types of hidden dangers. When the supervisors manually annotate and correct the output results, the annotated data is regularly fed back to the training set to optimize the hidden danger recognition model. Combined with the standardized data set collected in real time by the service platform, based on the analytic hierarchy process, gas safety experts are invited to score the weights of four indicators: the credit rating of industrial and commercial households, the gas use stability, the equipment qualification rate, and the timeliness of hidden danger rectification, and a user safety evaluation and analysis model is established. Based on the input data, the evaluation and analysis model calculates the safety scores of non-resident users, and assigns red, yellow, and green codes to non-resident users based on the calculated safety scores, providing a quantitative basis for supervision decisions.

6. The non-resident user safety supervision system based on intelligent gas according to claim 5, characterized in that, According to the red, yellow, and green codes assigned to non-resident users by the evaluation and analysis model integrated in the AI analysis module, the supervision decision-making module automatically generates a differentiated inspection plan. Based on the hidden dangers automatically identified by the hidden danger recognition model integrated in the AI analysis module, the supervision decision-making module dynamically increases the inspection frequency of a certain area in combination with the occurrence times of the hidden danger type in that area, and then shares the hidden danger information with the emergency management department and the fire department through the data interface module to jointly carry out special rectification actions.

7. The non-resident user safety supervision system based on intelligent gas according to claim 6, characterized in that, The data interface module supports internal data interaction, specifically manifested as: real-time synchronization of user registration, self-inspection, and rectification data with the service platform to ensure the consistency of information between the service platform and the supervision platform; docking with intelligent gas meter data to warn of potential leakage risks through abnormal flow.

8. The non-resident user safety supervision system based on intelligent gas according to claim 7, characterized in that, The data interface module supports external data docking, specifically manifested as: encrypted data transmission with a third-party supervision system through an API interface, supporting cross-departmental data sharing, and at the same time obtaining external policy documents and industry standard data to update the supervision rules of the supervision platform.

9. The non-resident user safety supervision system based on intelligent gas according to claim 7, characterized in that, The specific implementation process of the safety supervision system includes the following steps: S1. Non-resident users complete registration by scanning the exclusive QR code of the gas equipment or manually entering enterprise information through the Web / App / miniprogram terminal of the intelligent gas non-resident user service platform. After registration, a unique user identifier is automatically generated, and the geographical coordinates and the gas use equipment ledger are associated to create a digital user profile and store it in the data management module. S2. Through the service platform, the basic information, intelligent gas meter gas use data, equipment operation status, and hidden danger self-inspection data of the registered non-resident users are collected in real time, forming a standardized data set and encrypted synchronization to the intelligent gas non-resident user supervision platform. S3. Based on the analytic hierarchy process, gas safety experts are invited to score the weights of four indicators: the credit rating of industrial and commercial households, the gas use stability, the equipment qualification rate, and the timeliness of hidden danger rectification, and a user safety evaluation and analysis model is established. S4. The standardized data set is input into the evaluation and analysis model, and the evaluation and analysis model automatically calculates the safety scores of users, and assigns red, yellow, and green codes to non-resident users based on the calculated safety scores, thereby realizing the classification of user risk levels. S5. According to the red, yellow, and green codes assigned to non-resident users by the evaluation and analysis model integrated in the AI analysis module, the supervision decision-making module of the supervision platform automatically generates a differentiated inspection plan. S6. The message push module, based on the user risk level and the contact information reserved during registration, uses the rule engine to push self-inspection reminders in a targeted manner, supporting multiple channels such as SMS, in-app messages, and voice calls, and attaching a personalized inspection checklist. S7. Non-resident users take on-site photos or videos according to the inspection checklist through the user interaction module of the service platform. The system automatically adds watermarks containing user identification, timestamp, and GPS location to the multimedia files to prevent data tampering. After uploading, the AI analysis module of the supervision platform calls the hidden danger identification model to perform intelligent analysis on the images and outputs a report on the type and location of the hidden dangers. S8. For non-resident users who have not conducted self-inspection, the message push module, through the self-inspection reminder strategy preset by the rule engine, pushes the inspection checklist and the deadline to non-resident users at least one way among automatic outbound calls via SMS, in-app messages, and voice calls N days before the start of the self-inspection cycle. For users who have found hidden dangers during self-inspection but have not rectified them, the message push module, through the self-inspection hidden danger early warning reminder strategy preset by the rule engine, pushes the details of the hidden dangers to non-resident users when the AI analysis module identifies the hidden dangers and triggers an early warning message in real time, and simultaneously notifies the relevant person in charge of the non-resident users. After the users complete the rectification, they upload the comparison photos and descriptions before and after. After the supervision personnel conduct on-site verification and find no errors, they update the status of the hidden dangers and file them in the data management module. S9. The supervision decision-making module regularly analyzes the full amount of hidden danger data, organizes experts to optimize the index weights of the evaluation analysis model, shares high-risk user data with the fire and market supervision departments through the data interface module, and jointly conducts special rectification actions for gas safety to achieve cross-departmental supervision coordination.

Images