Gas user intelligent safety inspection system based on Internet of Things

Abstract

The invention discloses a gas user intelligent safety inspection system based on the Internet of Things. According to the invention, through a multi-stage intelligent decision-making mechanism, the active protection capability of gas safety management is significantly improved. A traditional alarm system depends on single sensor threshold triggering, is susceptible to environmental interference and lags in response, while the system can accurately distinguish scenes such as real leakage and cooking interference by fusing multi-dimensional data such as gas concentration, temperature and flame and combining a user behavior mode to dynamically evaluate a risk level. For example, when it is detected that the gas concentration is abnormal but not accompanied by sudden temperature rise or a flame signal, the system automatically lowers the alarm level and only pushes prompt information; and when high-risk leakage is confirmed, the valve is immediately closed in a linkage mode, exhaust equipment is started, synchronous early warning is conducted through multiple channels such as telephones, short messages and a community platform, and it is ensured that the user and rescue strength respond in the golden time.

Description

Technical Field

[0001] The present invention belongs to the technical field of gas safety inspection, and specifically relates to an intelligent safety inspection system for gas users based on the Internet of Things. Background Technique

[0002] The intelligent safety inspection system for gas users is a high-tech product integrating detection, alarm, and control functions, mainly used to monitor potential safety hazards such as leakage and excessive concentration in the gas transmission and distribution system. This system usually includes parts such as gas sensors, data collectors, alarm devices, and remote monitoring platforms, which can monitor the gas concentration in real time. Once an abnormality is detected, it will immediately give an audible and visual alarm and start emergency measures. Through precise data analysis and processing, the gas safety inspection system can effectively prevent gas accidents, ensure the safety of people's lives and property, provide strong support for the safe production of gas enterprises, and is an important part of the modern urban gas safety management system.

[0003] However, the existing technologies mainly rely on a single sensor threshold to trigger alarms, making it difficult to distinguish real gas leakage from daily activity interference, resulting in a high false alarm rate and a rigid response method. At the same time, there is a lack of intelligent integration of user characteristics and surrounding rescue resources, and the emergency measures stay at the level of one-way notification, unable to achieve risk classification and disposal and multi-party collaborative rescue, resulting in a lag in hidden danger discovery and low disposal efficiency. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent safety inspection system for gas users based on the Internet of Things to solve the problems mentioned above.

[0005] The technical solution adopted by the present invention is as follows: An intelligent safety inspection system for gas users based on the Internet of Things, the system includes: a data acquisition module, a data transmission module, a data processing and analysis module, a warning and alarm module, a user interaction module, and a remote control module;

[0006] Inside the warning and alarm module, there are a multi-source data fusion sub-module, an intelligent risk classification sub-module, a dynamic response strategy sub-module, a feedback learning sub-module, and a community emergency linkage sub-module;

[0007] The sensor output end of the data acquisition module is connected to the input end of the data transmission module through a physical or digital interface, and the original monitoring data such as gas concentration and temperature is transmitted in real time;

[0008] The output end of the data transmission module uploads the data in a packaged form to the cloud receiving port of the data processing and analysis module by means of the Internet of Things communication protocol to complete the cross-network transmission of the original information.

[0009] The standardized result output terminals of the data processing and analysis module are respectively connected to the risk assessment interface of the warning and alarm module and the data visualization port of the user interaction module;

[0010] The command issuing end of the warning and alarm module is connected to the actuator interface of the remote control module through an encrypted channel to directly drive the valve motor or the exhaust equipment;

[0011] The feedback data return port of the user interaction module is reversely connected to the machine learning training unit of the data processing and analysis module to form a closed-loop input of user behavior annotation data.

[0012] The device status feedback end of the remote control module is directly connected to the valve detector of the data acquisition module at the hardware level to feedback the valve opening and closing status to the sensor network in real time for operation status verification.

[0013] In a preferred embodiment, a gas leakage detection sensor is arranged inside the data acquisition module. The sensor uses a highly sensitive semiconductor element to detect the change of the concentration of combustible gas in the air in real time. When the concentration exceeds the preset safety threshold, the internal circuit will immediately generate an electrical signal and trigger the alarm mechanism.

[0014] In a preferred embodiment, the data transmission module uses NB-IoT technology as the main transmission means. Its narrowband characteristics can penetrate wall obstacles in complex urban environments and directly upload the collected data to the cloud server. This technology has a built-in sleep and wake-up mechanism, automatically enters the low-power state when there is no data transmission, and only starts full-power communication when there is an alarm signal or regular reporting, so that the device battery life can reach several years. To cope with extreme situations, the module also integrates a wired communication interface as a redundant backup, and immediately switches to the optical fiber or Ethernet transmission channel when the wireless network fails.

[0015] In a preferred embodiment, the data processing and analysis module deeply mines the historical gas usage data through a supervised learning model to construct a correlation model between user behavior portraits and device states.

[0016] In a preferred embodiment, the multi-source data fusion sub-module integrates multi-dimensional sensor data such as gas leakage concentration, temperature, flame, pressure, and valve status, and uses time series correlation analysis and probability models to eliminate single-point data errors;

[0017] The calculation formula of the Bayesian network joint probability model is:

[0018]

[0019] Where P(Leak∣S 1 ,S 2 ,...,Sn ) represents the posterior probability of gas leakage occurring under the conditions of sensor data S1, S2,..., Sn.

[0020] P(S i ∣Leak) represents the conditional probability that sensor S detects an anomaly when there is a gas leak. i

[0021] P(Leak) represents the prior probability of gas leakage.

[0022] In a preferred embodiment, the intelligent risk grading sub-module calculates the comprehensive risk score in real time based on the multi-source data fusion result, combined with the user behavior pattern and device status, through a lightweight random forest model. The model inputs include:

[0023] Sensor data: normalized gas concentration, temperature, flame intensity, etc.

[0024] User profile: age, living alone status.

[0025] Environmental parameters: ventilation status.

[0026] Historical records: number of recent false alarms.

[0027] The model output is divided into three levels of risk:

[0028] Low risk: only APP push reminder.

[0029] Medium risk: sound and light alarm + SMS notification.

[0030] High risk: automatic valve closing + phone alarm + community linkage;

[0031] Among them, the calculation formula of the risk score weighting model is:

[0032]

[0033] Among them:

[0034] R represents the comprehensive risk score.

[0035] Si represents the normalized value of the i-th sensor data.

[0036] wi represents the sensor weight.

[0037] U represents the user profile coefficient.

[0038] H represents the historical false alarm attenuation factor.

[0039] α, β, γ represent adjustment coefficients.

[0040] In a preferred embodiment, the dynamic response strategy submodule dynamically selects the optimal response action according to the risk score and user profile, specifically including:

[0041] Elderly families: priority will be given to telephone notifications. If no one answers the call, the community will be contacted to visit the elderly.

[0042] Valve abnormality: Start the spare solenoid valve and push the maintenance work order.

[0043] Offline scenario: local sound and light alarm + valve forced closure.

[0044] At the same time, the response strategy is linked with the smart home, such as automatically turning on the exhaust fan and turning off the main power switch when the alarm is triggered;

[0045] The response priority decision function calculation formula is:

[0046]

[0047] Where R represents the risk score. The valve status indicates normal or abnormal, which is provided by the valve detector. User confirmation of safety means that the user clicks the "Confirm Safety" button after receiving the warning.

[0048] In a preferred embodiment, the feedback learning submodule converts user active feedback into a driving force for system self-optimization. When a user marks an alarm as a false alarm through a mobile phone APP, the system automatically captures multi-dimensional environmental data within five minutes before and after the alarm is triggered, including gas concentration fluctuation curves, indoor temperature and humidity changes, equipment operation records, etc., to construct a false alarm case data set. After desensitization, these data are input into the cloud training platform, and the sensor weight distribution in the risk scoring model is dynamically adjusted by comparing the feature differences between real leaks and false alarm scenarios.

[0049] The community emergency linkage submodule opens up a collaborative channel between home safety data and public rescue resources. When the system determines that it is a high-risk alarm, it automatically sends a standardized alarm work order to the community management platform, which includes the user's address, contact information, gas leak concentration curve, executed safety operation records and indoor floor plan. The community platform uses the preset emergency response protocol to initiate different rescue processes according to the alarm level: the first-level alarm is directly connected to the scene by the fire department, and the second-level alarm dispatches the property safety officer with portable detection equipment to conduct priority home inspections.

[0050] In a preferred embodiment, the mobile phone APP of the user interaction module adopts a hierarchical visual design. The main interface uses color coding to intuitively display the overall status of the gas system. Green represents safety, yellow indicates warning, and red indicates emergency. When an abnormality is detected, the interface automatically pops up a three-dimensional pipeline diagram, accurately calibrates the location of the problem, and superimposes a processing suggestion button. For the elderly user group, the APP has specially developed a voice guidance function, which continuously plays concise handling instructions after the alarm is triggered to avoid operational errors under nervous emotions.

[0051] In a preferred embodiment, the electric ball valve of the remote control module adopts stepper motor drive technology, which can complete 90-degree rotation within 0.5 seconds after receiving the valve closing command to achieve full locking. The response speed in emergency situations is more than 3 times faster than that of traditional solenoid valves. The control protocol is designed with a double verification mechanism, and cloud commands must be verified by the dynamic key on the device side before they can be executed, effectively preventing the risk of misoperation caused by network attacks. The system also has a valve status self-check function, which performs a 1-degree micro-motion test on a daily basis, detects the wear of mechanical parts through torque sensors, and warns of possible jamming failures in advance.

[0052] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0053] 1. In the present invention, the active protection capability of gas safety management is significantly improved through a multi-level intelligent decision-making mechanism. The traditional alarm system relies on a single sensor threshold trigger, which is susceptible to environmental interference and has a delayed response. The system integrates multi-dimensional data such as gas concentration, temperature, flame, etc., and dynamically evaluates the risk level in combination with user behavior patterns. It can accurately distinguish between real leaks and cooking interference and other scenarios. For example, when an abnormal gas concentration is detected but is not accompanied by a sudden temperature rise or flame signal, the system automatically lowers the alarm level and only pushes a prompt message; when a high-risk leak is confirmed, the valve is immediately closed and the exhaust equipment is started, and warnings are issued simultaneously through multiple channels such as telephone, text messages, and community platforms to ensure that users and rescue forces respond within golden time.

[0054] 2. In the present invention, a closed-loop management network for family safety and public emergency services is constructed through intelligent early warning and community collaboration mechanisms. For special groups such as elderly people living alone and families with children, the system automatically optimizes the alarm strategy based on user portraits, such as giving priority to notifying relatives by phone or community grid members to come to the door for confirmation, to make up for the shortcomings of some users who are not sensitive to the operation of electronic devices. At the same time, the community linkage submodule pushes standardized alarm information to responsible units such as property and fire protection, clarifies the handling process and time limit requirements, and avoids delays caused by information gaps in traditional emergency responses. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 is the overall system block diagram of the present invention;

[0056] Figure 2 This is the system block diagram of the warning and alarm module in the present invention. Detailed implementation manners

[0057] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0058] Embodiment:

[0059] Refer to Figure 1-2 ,

[0060] An intelligent safety inspection system for gas users based on the Internet of Things, the system includes: a data acquisition module, a data transmission module, a data processing and analysis module, a warning and alarm module, a user interaction module, and a remote control module;

[0061] Inside the warning and alarm module, there are a multi-source data fusion sub-module, an intelligent risk grading sub-module, a dynamic response strategy sub-module, a feedback learning sub-module, and a community emergency linkage sub-module;

[0062] The sensor output end of the data acquisition module is connected to the input end of the data transmission module through a physical or digital interface, and the original monitoring data such as gas concentration and temperature is transmitted in real time;

[0063] The output end of the data transmission module packs and uploads the data to the cloud receiving port of the data processing and analysis module by means of the Internet of Things communication protocol, completing the cross-network transmission of the original information.

[0064] The standardized result output end of the data processing and analysis module is respectively connected to the risk assessment interface of the warning and alarm module and the data visualization port of the user interaction module;

[0065] The command issuing end of the warning and alarm module is connected to the actuator interface of the remote control module through an encrypted channel, directly driving the valve motor or the exhaust equipment;

[0066] The feedback data return port of the user interaction module is reversely connected to the machine learning training unit of the data processing and analysis module, forming a closed-loop input of the user behavior annotation data.

[0067] The device status feedback end of the remote control module is directly connected to the valve detector of the data acquisition module at the hardware level, and the valve opening and closing status is fed back to the sensor network in real time for operation status verification.

[0068] The data acquisition module is internally equipped with a gas leakage detection sensor. This sensor uses highly sensitive semiconductor components to detect the concentration change of combustible gas in the air in real time. When the concentration exceeds the preset safety threshold, the internal circuit will immediately generate an electrical signal and trigger the alarm mechanism. Different from the traditional single-threshold alarm method, this sensor adopts dynamic calibration technology, which can automatically adjust the detection sensitivity according to external factors such as environmental humidity and air pressure, avoiding false alarms caused by environmental interference. For example, when the air humidity is high during the rainy season, the sensor will appropriately enhance its anti-interference ability to ensure that the detection results truly reflect the gas leakage situation. This intelligent calibration function greatly improves the accuracy and reliability of data acquisition, providing high-quality raw data for subsequent analysis.

[0069] The data transmission module uses NB-IoT technology as the main transmission means. Its narrowband characteristics can penetrate wall barriers in complex urban environments and directly upload the collected data to the cloud server. This technology has a built-in sleep-wake mechanism that automatically enters the low-power state when there is no data transmission and only starts full-power communication when there is an alarm signal or regular reporting. This enables the device to have a battery life of several years. To cope with extreme situations, the module also integrates a wired communication interface as a redundant backup, which immediately switches to the fiber or Ethernet transmission channel when the wireless network fails. This dual-mode communication design effectively balances energy consumption and reliability, ensuring that critical data can be uploaded in real time in any environment.

[0070] The data processing and analysis module deeply mines historical gas usage data through a supervised learning model to build a correlation model between user behavior portraits and device states. For example, for the gas usage habits of the elderly living alone, the algorithm will identify abnormal patterns such as the stove being left on for a long time and comprehensively evaluate the risk level in combination with real-time sensor data. During the training process, transfer learning technology is used to transfer the characteristics of accident cases in the public safety database to the local model, improving the prediction accuracy in small-sample scenarios. This intelligent analysis not only enables real-time monitoring but also can predict potential risks 24 hours in advance, truly achieving the safety goal of preventing problems before they occur.

[0071] The multi-source data fusion sub-module integrates multi-dimensional sensor data such as gas leakage concentration, temperature, flame, pressure, and valve status, and uses time series correlation analysis and probability models to eliminate single-point data errors. For example, when the gas leakage sensor detects an increase in concentration, it combines whether the temperature sensor synchronously increases (possibly caused by leaked gas encountering a fire source) and the status of the flame sensor to determine whether it is a real leakage or kitchen cooking interference. The Kalman filter is used to dynamically correct sensor noise, and the joint probability of different sensor data is calculated through a Bayesian network to identify false alarm scenarios (such as short-term flame triggering but no gas leakage). In addition, environmental parameters (such as indoor ventilation status) are introduced to adaptively calibrate the sensor threshold. For example, in a well-ventilated scenario, the gas concentration threshold can be appropriately relaxed to avoid frequent false alarms;

[0072] The calculation formula for the joint probability model of the Bayesian network is:

[0073]

[0074] Where P(Leak∣S 1 ,S 2 ,...,S n ) represents the posterior probability of gas leakage occurring under the conditions of sensor data S1, S2,..., Sn.

[0075] P(S i ∣Leak) represents the conditional probability that sensor S i detects an anomaly when there is a gas leakage (obtained through training with historical data).

[0076] P(Leak) represents the prior probability of gas leakage (set according to regional accident statistics, for example, 0.1%).

[0077] Example: If the gas concentration sensor S1 is triggered (P(S1∣Leak) = 0.95), but the flame sensor S2 is not triggered (P(S2∣Leak) = 0.05), the posterior probability is significantly reduced and it is determined to be a low risk.

[0078] Based on the multi-source data fusion result, the intelligent risk grading sub-module combines the user behavior pattern (such as no one at home during the working day when going out) and the device status (whether the valve is normal), and calculates the comprehensive risk score (0 - 100 points) in real time through a lightweight random forest model. The model inputs include:

[0079] Sensor data: Normalized gas concentration, temperature, flame intensity, etc.

[0080] User profile: Age (risk weight for the elderly + 20%), living alone status (risk + 30%).

[0081] Environmental parameter: ventilation status (risk +15% when ventilation is poor).

[0082] Historical record: recent false alarm count (frequent false alarms reduce confidence).

[0083] The model output is divided into three levels of risk:

[0084] Low risk (60 - 70 points): only APP push reminder.

[0085] Medium risk (70 - 90 points): audible and visual alarm + SMS notification.

[0086] High risk (>90 points): automatic valve closing + phone alarm + community linkage;

[0087] Among them, the calculation formula of the risk scoring weighted model is:

[0088]

[0089] Among them

[0090] R represents the comprehensive risk score (0 - 100 points).

[0091] Si represents the normalized value of the data of the i-th sensor (such as gas concentration / threshold).

[0092] wi represents the sensor weight (gas leakage weight 0.4, flame 0.3, pressure 0.2, valve 0.1).

[0093] U represents the user portrait coefficient (for single - living users U = 1.3, for non - single - living users U = 1.0).

[0094] H represents the historical false alarm attenuation factor (for each recent false alarm, H = 0.95).

[0095] α, β, γ represent adjustment coefficients (by default, \(\alpha = 0.7\), \(\beta = 0.2\), \(\gamma = 0.1\)).

[0096] Example: for gas leakage, S1 = 1.5 (50% above the threshold); for single - living users, U = 1.3,

[0097] Then R = 0.7·(0.4×1.5)+0.2×1.3 = 0.42 + 0.26 = 68 → medium risk.

[0098] The dynamic response strategy sub - module dynamically selects the optimal response action according to the risk score and user portrait, specifically including:

[0099] For elderly families: give priority to phone notification (due to decreased hearing sensitivity), if not answered, then link the community to visit.

[0100] Valve anomaly (unable to close automatically): Start the standby solenoid valve and push a maintenance work order.

[0101] Offline scenario: Local audible and visual alarm + forced valve closure (hardware redundant design).

[0102] At the same time, the response strategy is linked with smart home, such as automatically turning on the exhaust fan and shutting down the main power switch when the alarm is triggered;

[0103] The calculation formula of the response priority decision function is:

[0104]

[0105] Where R represents the risk score.

[0106] The valve status indicates normal (1) or abnormal (0), which is provided by the valve detector.

[0107] User confirmation of safety means that the user clicks the "Confirm Safety" button after receiving the early warning.

[0108] Example: When R = 95 and the valve is normal, perform the combined actions of "valve closing + phone call + community", and the response delay is less than 2 seconds.

[0109] The feedback learning sub-module converts the user's active feedback into the driving force for the system's self-optimization. When the user marks a certain alarm as a false alarm through the mobile APP, the system automatically captures multi-dimensional environmental data within five minutes before and after the alarm is triggered, including the gas concentration fluctuation curve, indoor temperature and humidity changes, equipment operation records, etc., to construct a false alarm case data set. After these data are desensitized, they are input into the cloud training platform. By comparing the characteristic differences between real leakage and false alarm scenarios, the sensor weight distribution in the risk scoring model is dynamically adjusted. For example, if the system finds that the user frequently triggers false alarms due to short-term flames during cooking, it automatically reduces the decision weight of the flame sensor and simultaneously increases the correlation coefficient of the gas concentration and pressure sensors. This closed-loop learning mechanism enables the model to adapt to the living habits of different families, and the false alarm rate can continuously decrease as the usage time increases, ultimately achieving more accurate personalized safety protection with increasing use;

[0110] The community emergency linkage submodule opens up a collaborative channel between home safety data and public rescue resources. When the system determines that it is a high-risk alarm, it automatically sends a standardized alarm work order to the community management platform, which covers the user's address, contact information, gas leak concentration curve, executed safety operation records and indoor floor plan. The community platform uses a preset emergency response protocol to initiate different rescue processes according to the alarm level: the first-level alarm is directly connected to the scene by the fire department, and the second-level alarm dispatches property security personnel with portable detection equipment to conduct priority home inspections. The system will also use geographic fence technology to push risk warnings to other gas users within 100 meters of the alarm point, reminding them to temporarily close valves and open windows for ventilation. This layered response mechanism not only avoids excessive consumption of public resources, but also ensures that major dangerous situations can obtain precise support from multi-level rescue forces within the golden disposal time.

[0111] The mobile phone APP of the user interaction module adopts a hierarchical visual design. The main interface uses color coding to intuitively display the overall status of the gas system. Green represents safety, yellow indicates warning, and red indicates emergency. When an abnormality is detected, the interface automatically pops up a three-dimensional pipeline diagram, accurately calibrates the location of the problem, and superimposes a processing suggestion button. For the elderly user group, the APP has specially developed a voice guidance function, which continuously plays concise handling instructions after the alarm is triggered to avoid operational errors under nervous emotions. This humanized interactive design fully considers the differences in needs of different users to ensure that safety information can be quickly understood and responded to.

[0112] The electric ball valve of the remote control module adopts stepper motor drive technology, which can complete a 90-degree rotation within 0.5 seconds to achieve full locking after receiving the valve closing command. The response speed in emergency situations is more than 3 times faster than that of traditional solenoid valves. The control protocol is designed with a double verification mechanism. Cloud commands must be verified by the dynamic key on the device side before they can be executed, effectively preventing the risk of misoperation caused by network attacks. The system also has a valve status self-check function, which performs a 1-degree micro-motion test on a daily basis, detects the wear of mechanical parts through torque sensors, and warns of possible jamming failures in advance. This closed-loop control system ensures both the timeliness of emergency response and the stability of long-term operation of the equipment.

[0113] From the above we can know:

[0114] In the present invention, through a multi-level intelligent decision-making mechanism, the active protection ability of gas safety management is significantly improved. The traditional alarm system relies on the triggering of a single sensor threshold, which is vulnerable to environmental interference and has a lagging response. In contrast, this system can accurately distinguish scenarios such as real gas leakage and cooking interference by integrating multi-dimensional data such as gas concentration, temperature, and flame, and dynamically evaluating the risk level in combination with user behavior patterns. For example, when abnormal gas concentration is detected without a sudden rise in temperature or a flame signal, the system automatically reduces the alarm level and only pushes a reminder message; when a high-risk gas leakage is confirmed, it immediately triggers the closing of the valve and the activation of the exhaust equipment, and synchronously issues early warnings through multiple channels such as phone calls, text messages, and community platforms to ensure that users and rescue forces can respond within the golden time.

[0115] In the present invention, through an intelligent early warning and community collaboration mechanism, a closed-loop management network for home safety and public emergency services is constructed. For special groups such as the families of the elderly and children living alone, the system automatically optimizes the alarm strategy based on the user profile. For example, it gives priority to notifying relatives or community grid administrators by phone to visit and confirm, making up for the shortcoming that some users are not sensitive to the operation of electronic devices. At the same time, the community linkage sub-module standardizes the push of alarm information to responsible units such as property management and fire departments, clarifies the disposal process and time limit requirements, and avoids delays caused by information gaps in traditional emergency responses.

[0116] It should be noted that in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements not only includes those elements but also includes other elements not expressly listed, or elements inherent to such process, method, article, or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the existence of additional identical elements in the process, method, article, or device comprising the element.

[0117] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent safety inspection system for gas users based on the Internet of Things, characterized by: The system includes: a data acquisition module, a data transmission module, a data processing and analysis module, an early warning and alarm module, a user interaction module and a remote control module; The early warning and alarm module is internally provided with a multi-source data fusion submodule, an intelligent risk classification submodule, a dynamic response strategy submodule, a feedback learning submodule and a community emergency linkage submodule; The sensor output end of the data acquisition module is connected to the input end of the data transmission module through a physical or digital interface to transmit raw monitoring data such as gas concentration and temperature in real time; The output end of the data transmission module uses the Internet of Things communication protocol to package the data and upload it to the cloud receiving port of the data processing and analysis module to complete the cross-network transmission of the original information; The standardized result output terminal of the data processing and analysis module is respectively connected to the risk assessment interface of the early warning and alarm module and the data visualization port of the user interaction module; The sending end under the instruction of the early warning and alarm module is connected to the actuator interface of the remote control module through an encrypted channel to directly drive the valve motor or exhaust equipment; The feedback data return port of the user interaction module is reversely connected to the machine learning training unit of the data processing and analysis module to form a closed-loop input of the user behavior annotation data; The device status feedback end of the remote control module forms a hardware-level direct connection with the valve detector of the data acquisition module, and feeds back the valve opening and closing status to the sensor network in real time for operation status verification.

2. The gas user intelligent safety inspection system based on the Internet of Things as claimed in claim 1, characterized in that: The data acquisition module is internally provided with a gas leak detection sensor, which detects the concentration changes of combustible gas in the air in real time through highly sensitive semiconductor elements. When the concentration exceeds a preset safety threshold, the internal circuit will immediately generate an electrical signal and trigger an alarm mechanism.

3. The gas user intelligent safety inspection system based on the Internet of Things as claimed in claim 1, characterized in that: The data transmission module adopts NB-IoT technology as the main transmission means. Its narrowband characteristics can penetrate wall obstacles in complex urban environments and directly upload the collected data to the cloud server. The technology has a built-in sleep and wake-up mechanism, which automatically enters a low-power state when there is no data transmission, and only starts full-power communication when there is an alarm signal or a scheduled report, so that the device can last for several years. To cope with extreme situations, the module also integrates a wired communication interface as a redundant backup, and immediately switches to the optical fiber or Ethernet transmission channel when the wireless network fails.

4. The gas user intelligent safety inspection system based on the Internet of Things as claimed in claim 1, characterized in that: The data processing and analysis module conducts in-depth mining of historical gas usage data through a supervised learning model, and constructs a user behavior profile and equipment status association model.

5. The gas user intelligent safety inspection system based on the Internet of Things as claimed in claim 1, characterized in that: The multi-source data fusion submodule integrates multi-dimensional sensor data such as gas leakage concentration, temperature, flame, pressure, valve status, etc., and uses time series correlation analysis and probability models to eliminate single-point data errors; The calculation formula of the Bayesian network joint probability model is: Where P(Leak|S1,S2,...,S n ) represents the posterior probability of gas leakage under the conditions of sensor data S1, S2, ..., Sn; P(S i |Leak) indicates that the sensor S i The conditional probability of detecting an anomaly; P(Leak) represents the prior probability of gas leakage.

6. The gas user intelligent safety inspection system based on the Internet of Things as claimed in claim 1, characterized in that: The intelligent risk grading submodule calculates the comprehensive risk score in real time through a lightweight random forest model based on the multi-source data fusion results, combined with user behavior patterns and device status; Model inputs include: Sensor data: normalized gas concentration, temperature, flame intensity, etc.; User profile: age, living alone; Environmental parameters: ventilation status; History: recent false alarm count; The model output is divided into three levels of risk: Low risk: Only APP push reminders; Medium risk: sound and light alarm + SMS notification; High risk: automatic valve closing + telephone alarm + community linkage; The calculation formula of the risk score weighted model is: in: R represents the comprehensive risk score; Si represents the normalized value of the i-th sensor data; wi represents the sensor weight; U represents the user portrait coefficient; H represents the historical false alarm attenuation factor; α, β, γ represent adjustment coefficients.

7. The gas user intelligent safety inspection system based on the Internet of Things as claimed in claim 1, characterized in that: The dynamic response strategy submodule dynamically selects the optimal response action based on the risk score and user profile, specifically including: Elderly families: Priority will be given to telephone notifications. If no one answers, the community will visit the elderly; Valve abnormality: start the standby solenoid valve and push the maintenance work order; Offline scenario: local sound and light alarm + valve forced closure; At the same time, the response strategy is linked with the smart home, such as automatically turning on the exhaust fan and turning off the main power switch when the alarm is triggered; The response priority decision function calculation formula is: Where R represents the risk score; valve status represents normal or abnormal, which is provided by the valve detector; user confirmation of safety means that the user clicks the "Confirm Safety" button after receiving the warning.

8. The gas user intelligent safety inspection system based on the Internet of Things as claimed in claim 1, characterized in that: The feedback learning submodule converts user active feedback into a driving force for system self-optimization. When a user marks an alarm as a false alarm through a mobile phone APP, the system automatically captures multi-dimensional environmental data within five minutes before and after the alarm is triggered, including gas concentration fluctuation curves, indoor temperature and humidity changes, equipment operation records, etc., to build a false alarm case data set. After desensitization, these data are input into the cloud training platform, and by comparing the feature differences between real leaks and false alarm scenarios, the sensor weight distribution in the risk scoring model is dynamically adjusted. The community emergency linkage submodule opens up a collaborative channel between home safety data and public rescue resources; when the system determines that it is a high-risk alarm, it automatically sends a standardized alarm work order to the community management platform, which covers the user's address, contact information, gas leakage concentration curve, executed safety operation records and indoor floor plan; the community platform uses a preset emergency response protocol to initiate different rescue processes according to the alarm level: the first-level alarm is directly connected to the scene by the fire department, and the second-level alarm dispatches property safety personnel with portable detection equipment to prioritize home inspections.

9. The gas user intelligent safety inspection system based on the Internet of Things as claimed in claim 1, characterized in that: The mobile phone APP of the user interaction module adopts a hierarchical visual design. The main interface uses color coding to intuitively display the overall status of the gas system. Green represents safety, yellow indicates warning, and red indicates emergency. When an abnormality is detected, the interface automatically pops up a three-dimensional pipeline diagram, accurately calibrates the location of the problem, and superimposes a processing suggestion button. For the elderly user group, the APP has specially developed a voice guidance function, which continuously plays concise handling instructions after the alarm is triggered to avoid operational errors caused by tension.

10. The gas user intelligent safety inspection system based on the Internet of Things as claimed in claim 1, characterized in that: The electric ball valve of the remote control module adopts stepper motor drive technology, which can complete a 90-degree rotation within 0.5 seconds to achieve complete locking after receiving the valve closing command. The response speed in emergency situations is more than 3 times faster than that of traditional solenoid valves. The control protocol is designed with a double verification mechanism, and cloud commands must be verified by the dynamic key on the device side before they can be executed, effectively preventing the risk of misoperation caused by network attacks. The system also has a valve status self-check function, and performs a 1-degree micro-motion test on a daily basis. The wear of mechanical parts is detected by a torque sensor, and possible jamming failures are warned in advance.

Images