Indoor gas leakage monitoring and intelligent joint control system

Abstract

The invention discloses an indoor gas leakage monitoring and intelligent joint control system, and belongs to the technical field of intelligent safety monitoring systems. Comprising a sensing layer, a network layer, a control layer and a platform layer, the sensing layer comprises a multi-sensor array and an edge computing node, the multi-sensor array is combined with a catalytic combustion sensor, an infrared sensor, a semiconductor sensor and a temperature and humidity sensor, and sensor data are fused through a weighting algorithm so as to reduce the false alarm rate; a temperature and humidity sensor dynamically adjusts a detection threshold value to adapt to different climate conditions, and an edge computing node is deployed with a lightweight machine learning model. Through multi-sensor data fusion and a weighting algorithm, the false alarm rate is reduced to be smaller than 0.1% per year, environment self-adaptive calibration and AI auxiliary decision making are performed, interference of alcohol, perfume and the like is effectively eliminated, different climate conditions are adapted, leakage judgment is completed at an equipment end through edge calculation, cloud delay is avoided, a multi-stage emergency strategy is provided, and intelligent safe ecology driven by the Internet of Things is constructed.

Description

Technical Field

[0001] This invention relates to the field of intelligent safety monitoring system technology, and more specifically, to an indoor gas leak monitoring and intelligent control system. Background Technology

[0002] With the advancement of urbanization, natural gas has gradually become a widely used energy source in residential and commercial environments. Due to the flammability and toxicity of natural gas, gas leaks are frequent, seriously threatening the lives and property of users. Existing natural gas leak detection equipment typically relies on chemical or electrochemical sensors to monitor indoor gas concentrations. While these sensors can detect leaks, they are easily affected by environmental changes, leading to false alarms or missed alarms.

[0003] Gas leaks pose a significant safety hazard to homes, industries, and businesses, potentially leading to serious accidents such as explosions, fires, or carbon monoxide poisoning. Traditional gas alarms only provide audible and visual alarms and cannot take proactive measures (such as shutting off valves or ventilating), and they suffer from problems such as high false alarm rates, weak networking capabilities, and delayed response times.

[0004] In view of this, the present invention is proposed to solve the above-mentioned technical problems. Summary of the Invention

[0005] The purpose of this invention is to provide an indoor gas leak monitoring and intelligent control system to solve the technical problems mentioned above.

[0006] To achieve the above objectives, the present invention provides the following technical solution: An indoor gas leak monitoring and intelligent control system adopts a three-layer intelligent architecture of "perception-decision-execution" and combines it with an Internet of Things cloud platform to build a complete gas safety closed-loop control system, including a perception layer, a network layer, a control layer and a platform layer. The perception layer includes a multi-sensor array and edge computing nodes. The multi-sensor array combines catalytic combustion sensors, infrared sensors, semiconductor sensors, and temperature and humidity sensors. The sensor data is fused through a weighted algorithm to reduce the false alarm rate. The temperature and humidity sensors dynamically adjust the detection threshold to adapt to different climatic conditions. The edge computing nodes are equipped with lightweight machine learning models. The network layer adopts dual-mode communication, which includes local short-range communication and wide-area Internet of Things communication (NB-IoT). The control layer includes an intelligent linkage engine and multi-level actuators. The intelligent linkage engine has a multi-level emergency strategy based on concentration thresholds, and the multi-level actuators include high-speed solenoid valves, intelligent ventilation systems, and intelligent window actuators. The platform layer is a cloud-based data analysis and remote management platform, equipped with digital twin modeling and intelligent operation and maintenance.

[0007] Furthermore, the weighted algorithm is as follows: the fused concentration value C = W1×C1 + W2×C2 + W3×C3; Where C1 is catalytic combustion sensor data, C2 is infrared sensor data, and C3 is semiconductor sensor data, with weights W1=0.4, W2=0.4, and W3=0.2.

[0008] Furthermore, in the dynamic temperature and humidity adjustment detection threshold, the corrected threshold = baseline threshold × [1 + 0.005 × (T - T0) + 0.003 × (H - H0)]; Where T0 = 25℃, H0 = 50%, T0 and H0 are the reference temperature and humidity, T and H are the actual temperature and humidity, and the reference threshold is 10% LEL.

[0009] Furthermore, a lightweight machine learning model, such as a random forest or LSTM, is used on the edge device to analyze data patterns in real time, distinguish between real leaks and false alarms, and output the leak probability or classification result. When the probability is ≥0.8, it is determined to be a real leak.

[0010] Furthermore, the edge computing node is an STM32U5 MCU, used for local leak detection, with a response latency of <1 second.

[0011] Furthermore, the multi-level emergency response strategy is as follows: Primary leakage: triggers the intelligent audible and visual alarm, closes the solenoid valve, and activates the intelligent ventilation system; Intermediate leakage: triggers the intelligent sound and light alarm, closes the solenoid valve, activates the intelligent ventilation system, and controls the intelligent window actuator to open; Severe leakage: The solenoid valve is shut off, the audible and visual alarm is activated, and an emergency notification is sent to the user and safety inspector.

[0012] Furthermore, the platform layer supports integration with HomeKit, Mi Home, and Huawei HarmonyOS platform, enabling collaborative control with multi-level actuators; It enables collaborative device control through open APIs and features dual-mode redundancy between cloud and local operation. Real-time alarm information is pushed via APP / SMS, and historical data review and leakage trend analysis are supported.

[0013] Furthermore, the perception layer, network layer, and control layer achieve low-latency response through a timing coordination mechanism: the sensor array of the perception layer outputs sampling data at a frequency of 100ms / time, the edge computing node takes less than 200ms for fusion processing of the sampling data and AI inference, and the actuator of the control layer completes the action response in less than 500ms after receiving the trigger command. At the same time, it reports data to the platform layer synchronously through the network layer, forming a closed-loop control timing chain.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) Improved the accuracy and reliability of gas leak detection: Through multi-sensor data fusion, weighted algorithm to reduce false alarm rate to <0.1% / year, environmental adaptive calibration and AI-assisted decision-making, interference from alcohol, perfume and other substances can be effectively eliminated and adapted to different climate conditions.

[0015] (2) Millisecond-level leak response and automated handling: Leakage determination is completed at the device end through edge computing, avoiding cloud delay, ensuring triggering linkage within <1 second, and having multi-level emergency strategies, which can automatically take measures such as ventilation, valve closure, and window opening according to the degree of leakage (10%-20%LEL, 20%-50%LEL, >50%LEL).

[0016] (3) It has built an IoT-driven smart security ecosystem: It supports seamless access to smart home platforms such as HomeKit and Mijia to achieve collaborative control of devices; it has cloud-local dual-mode redundancy and supports remote monitoring and early warning (APP / SMS real-time push) to meet the needs of smart home and industrial security.

[0017] (4) Wide range of applications: Through modular design and multiple product lines (home / commercial / industrial version), it covers residential buildings, commercial places and industrial facilities, with an applicable area of ​​100-1000㎡, which is more than 300% larger than traditional alarms. Detailed Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] An indoor gas leak monitoring and intelligent control system is applied to the kitchen of a residential building. It adopts a three-layer intelligent architecture of "perception-decision-execution" and combines it with an Internet of Things cloud platform to build a complete gas safety closed-loop control system, including a perception layer, a network layer, a control layer and a platform layer. The multi-sensor array of the perception layer is installed in the center of the kitchen ceiling. The perception layer includes a multi-sensor array and edge computing nodes. The multi-sensor array combines catalytic combustion sensors, infrared sensors (NDIR), various semiconductor sensors, and temperature and humidity sensors. Sensor data is fused through a weighted algorithm to reduce the false alarm rate. Temperature and humidity sensors are used to dynamically adjust the detection threshold to adapt to different climatic conditions. The edge computing nodes are equipped with lightweight machine learning models. The network layer adopts dual-mode communication, which includes local short-range communication (Zigbee 3.0, Bluetooth Mesh, Wi-Fi) and wide-area IoT communication (NB-IoT). The control layer includes an intelligent linkage engine and multi-level actuators. The intelligent linkage engine has a multi-level emergency strategy based on concentration thresholds. The multi-level actuators include a high-speed solenoid valve (response time ≤ 50ms) installed on the gas pipeline, an intelligent ventilation system (using a brushless DC fan or exhaust fan), and an intelligent window actuator (push-pull force ≥ 200N). The platform layer is a cloud-based data analysis and remote management platform. The remote management platform is built on the Alibaba Cloud IoT platform and has digital twin modeling and intelligent operation and maintenance functions. Digital twin modeling uses the Unity3D engine to build a three-dimensional state model of the equipment and displays a gas concentration heat map in real time (generated using the Kriging interpolation algorithm). Intelligent operation and maintenance is based on LSTM for predictive maintenance.

[0020] In the above scheme, when a gas leak occurs in the kitchen, the multi-sensor array collects data every 100ms. Temperature and humidity sensors simultaneously collect environmental parameters, such as a baseline threshold of 10% LEL at 25℃ and 50%RH; and a corrected threshold of 10% × (1 + 0.005 × 10 + 0.003 × 30) = 10% × 1.14 = 11.4% LEL at 35℃ and 80%RH. Sensor data is transmitted to edge nodes and fused using a weighted algorithm (C = 0.4 × C1 + 0.4 × C2 + 0.2 × C3). The random forest model takes the fused concentration, concentration change rate, and temperature and humidity correction values ​​as input and outputs the leak probability. The values ​​of 0.005 (temperature) and 0.003 (humidity) are determined through sensor temperature drift experiments. Within the range of -20℃ to 60℃, the sensitivity of the catalytic combustion sensor decreases linearly with increasing temperature, with a slope of -0.005% LEL / ℃. For every 10% increase in humidity (RH), the semiconductor sensor baseline drifts by +0.03% LEL. In an environment of 35℃ and 80%RH, the baseline threshold of 10%LEL was corrected to 11.4%LEL. Actual measurements show that the corrected sensor response deviation is <±2%LEL, while the deviation without correction reaches ±8%LEL, significantly improving environmental adaptability.

[0021] If the concentration is 15% LEL, the probability is 0.85. The intelligent linkage engine controls the exhaust fan to turn on (PWM duty cycle 80%), the APP pushes a "gas leak" prompt, and at the same time, the intelligent sound and light alarm is triggered, the solenoid valve is closed, and the intelligent ventilation system (exhaust fan / range hood) is turned on.

[0022] If the gas concentration rises to 30% LEL and increases by 10% LEL within 5 seconds, the engine control solenoid valve closes (cutting off the gas supply within 10ms), the exhaust fan runs at full speed, and the app and SMS push a "gas leak" notification. At the same time, the smart sound and light alarm is triggered, the solenoid valve closes, the smart ventilation system (exhaust fan / range hood) is activated, and the smart window actuator is controlled to open.

[0023] If the concentration suddenly rises to 60% LEL, in addition to shutting off the engine valve and turning on the exhaust fan, the window actuator will be opened (300mm travel), the audible and visual alarm will be activated (85dB + flashing), an emergency notification will be pushed to the user and property management, the nearest maintenance personnel will be dispatched via the cloud, and an insurance claim link will be pushed simultaneously.

[0024] The weighted algorithm is: the concentration value after fusion C = W1×C1 + W2×C2 + W3×C3; Where C1 is catalytic combustion sensor data, C2 is infrared sensor data, and C3 is semiconductor sensor data, with weights W1=0.4, W2=0.4, and W3=0.2.

[0025] In the dynamic adjustment detection threshold of temperature and humidity, the corrected threshold = baseline threshold × [1 + 0.005 × (T - T0) + 0.003 × (H - H0)]; Where T0 = 25℃, H0 = 50%, T0 and H0 are the reference temperature and humidity, T and H are the actual temperature and humidity, and the reference threshold is 10% LEL.

[0026] Lightweight machine learning models, such as random forest (10 decision trees, with input features including fused concentration, concentration change rate, and temperature and humidity correction values) or LSTM (3 hidden layers, 32 neurons per layer), are used on edge devices to analyze data patterns in real time, distinguish between real leaks and false alarms, and output the leak probability or classification result. When the probability is ≥0.8, it is determined to be a real leak.

[0027] In some possible implementations, the edge computing node is an STM32U5 MCU (160MHz clock speed, 512KB memory) used for local leak detection with a response latency of <1 second. Leak detection is completed on the device side through edge computing to avoid cloud latency and ensure that the linkage is triggered within <1 second.

[0028] The multi-level emergency response strategy is as follows: Primary leak (low concentration, 10%-20% LEL): triggers the intelligent sound and light alarm (85dB+), closes the solenoid valve, and turns on the intelligent ventilation system (exhaust fan / range hood). If the problem cannot be resolved, the work order is automatically forwarded to the safety inspector for on-site handling. Once the alarm is cleared, a notification is sent to the user that the problem has been resolved. Intermediate leak (concentration continues to rise, 20%-50% LEL): triggers the intelligent sound and light alarm (85dB+), closes the solenoid valve, turns on the intelligent ventilation system (exhaust fan / range hood), and controls the intelligent window driver to open. If the problem cannot be resolved, the work order is automatically forwarded to the safety inspector for on-site handling. After the alarm is cleared, a notification is sent to the user that the problem has been resolved. Severe leakage (explosion risk, >50% LEL or concentration rise ≥30% LEL within 10 seconds): The solenoid valve is closed, the audible and visual alarm is activated (85dB+), and an emergency notification is pushed to the user and safety inspector. After the alarm is cleared, a notification is sent to the user that the problem has been resolved.

[0029] The platform layer supports integration with HomeKit, Mi Home, and Huawei HarmonyOS platforms, enabling collaborative control with multi-level actuators. Device collaborative control is achieved through open APIs, and it has dual-mode redundancy of cloud and local (the policy library stored in local Flash will continue to run for 72 hours when the network is disconnected). Real-time alarm information is pushed via APP / SMS, and historical data review and leakage trend analysis are supported.

[0030] The perception layer, network layer, and control layer achieve low-latency response through a timing coordination mechanism: the sensor array of the perception layer outputs sampling data at a frequency of 100ms / time, the edge computing node takes less than 200ms for fusion processing of the sampling data and AI inference, and the actuator of the control layer completes the action response in less than 500ms after receiving the trigger command. At the same time, it reports data to the platform layer synchronously through the network layer, forming a closed-loop control timing chain.

[0031] The workflow of the indoor gas leak monitoring and intelligent control system provided in this application embodiment is as follows: (1) The sensor array collects environmental data in real time at a frequency of 100ms / time; (2) Edge nodes run fusion algorithms and AI inference, taking <200ms; (3) Upon confirmation of the leak, the local actuator is immediately triggered, taking less than 500ms, and the cloud is simultaneously reported and an alarm is pushed. (4) Cloud-based handling suggestions, including linkage control, maintenance personnel scheduling, planning of the nearest maintenance point based on Gaode Map API, and insurance push assistance.

[0032] Security and intelligence Intelligent early warning: The system intelligently adjusts the early warning threshold based on the changing trend of natural gas concentration, and performs self-optimization through the accumulation of historical data to improve the accuracy and sensitivity of the early warning system.

[0033] False alarm and false alarm protection: The optical sensor has a strong ability to identify the characteristics of gas molecules, effectively avoiding false alarms caused by environmental changes (such as temperature and humidity). In addition, the system also has real-time data analysis capabilities, which can identify and avoid potential false alarms.

[0034] By integrating high-precision gas detection (NDIR + catalytic combustion), edge AI inference (1D CNN model), and IoT communication (6LoWPAN / NB-IoT) into a single device, the following was achieved: 1. Detection accuracy: Multi-sensor cross-validation reduces the false alarm rate to <0.1% / year (compared to approximately 3-5% for traditional solutions).

[0035] 2. Response speed: Local decision-making closed loop <500ms (cloud-dependent solutions are usually >2s).

[0036] 3. Adaptability: The dynamic environmental compensation algorithm can maintain ±2% LEL detection accuracy in environments ranging from -20℃ to 60℃.

[0037] Intelligent linkage dimension upgrade Spatial linkage; enabling multi-room collaboration through mesh networking; Time-based optimization; learning strategies based on historical data; Cross-system integration: Integration with fire protection systems and security systems; This patent constructs a complete closed loop of "detection-decision-response" in the field of gas safety by integrating vertical technology (full-stack optimization of sensing-communication-control) and horizontal scenario coverage (home-industrial-commercial). Its integrated advantages are reflected in: more accurate detection, faster response, more intelligent linkage, and lower overall cost.

[0038] Scope of application The indoor gas leak monitoring and intelligent control system provided in this application embodiment can achieve full coverage from ordinary residential buildings to extreme industrial environments. Its applicable scope is more than 300% larger than that of traditional gas alarms, and it maintains system compatibility in various scenarios through the Internet of Things architecture.

[0039] Core Applicable Scenarios (1) Residential buildings Urban apartments: Addressing the risk of gas leaks in enclosed kitchen spaces by integrating range hoods and window openers; Villa residences: Multi-story Mesh networking ensures safe gas coverage throughout the entire house; Renovation of old residential communities: Wireless installation avoids the need for wall-breaking wiring and is compatible with traditional gas meters; (2) Commercial premises Restaurant kitchens: To address the high leakage risk from high-powered stoves, explosion-proof detectors are installed; Hotel boiler room: monitored 24 hours a day and linked to the central air conditioning and fresh air system; Shopping mall food court: centralized monitoring of multiple merchants, zoned solenoid valve control; (3) Industrial facilities LNG vaporization station: Explosion-proof upgraded monitoring pipeline valve assembly; Chemical plant: Expand multi-gas detection capabilities.

[0040] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An indoor gas leak monitoring and intelligent control system, characterized in that, It adopts a three-layer intelligent architecture of "perception-decision-execution" and combines it with an Internet of Things cloud platform to build a complete gas safety closed-loop control system, including a perception layer, network layer, control layer and platform layer; The perception layer includes a multi-sensor array and edge computing nodes. The multi-sensor array combines catalytic combustion sensors, infrared sensors, semiconductor sensors, and temperature and humidity sensors. Sensor data is fused through a weighted algorithm to reduce the false alarm rate. The temperature and humidity sensors dynamically adjust the detection threshold to adapt to different climatic conditions. The edge computing nodes are equipped with lightweight machine learning models. The network layer adopts dual-mode communication, which includes local short-range communication and wide-area Internet of Things (NB-IoT) communication. The control layer includes an intelligent linkage engine and multi-level actuators. The intelligent linkage engine has a multi-level emergency strategy based on concentration thresholds. The multi-level actuators include high-speed solenoid valves, intelligent ventilation systems, and intelligent window actuators. The platform layer is a cloud-based data analysis and remote management platform, equipped with digital twin modeling and intelligent operation and maintenance.

2. The indoor gas leak monitoring and intelligent control system according to claim 1, characterized in that, The weighting algorithm is as follows: the fused concentration value C = W1×C1 + W2×C2 + W3×C3; Where C1 is catalytic combustion sensor data, C2 is infrared sensor data, and C3 is semiconductor sensor data, with weights W1=0.4, W2=0.4, and W3=0.

2.

3. The indoor gas leak monitoring and intelligent control system according to claim 2, characterized in that, In the aforementioned dynamic temperature and humidity adjustment detection threshold, the corrected threshold = baseline threshold × [1 + 0.005 × (T - T0) + 0.003 × (H - H0)]; Where T0 = 25℃, H0 = 50%, T0 and H0 are the reference temperature and humidity, T and H are the actual temperature and humidity, and the reference threshold is 10% LEL.

4. The indoor gas leak monitoring and intelligent control system according to claim 3, characterized in that, The lightweight machine learning model, such as Random Forest or LSTM, is used at the edge device to analyze data patterns in real time, distinguish between real leaks and false alarms, and output the leak probability or classification result. When the probability is ≥0.8, it is determined to be a real leak.

5. The indoor gas leak monitoring and intelligent control system according to claim 4, characterized in that, The edge computing node is an STM32U5 MCU, used for local leak detection, with a response latency of <1 second.

6. The indoor gas leak monitoring and intelligent control system according to claim 5, characterized in that, The multi-level emergency response strategy is as follows: Primary leakage: triggers the intelligent audible and visual alarm, closes the solenoid valve, and activates the intelligent ventilation system; Intermediate leakage: triggers the intelligent sound and light alarm, closes the solenoid valve, activates the intelligent ventilation system, and controls the intelligent window actuator to open; Severe leakage: The solenoid valve is shut off, the audible and visual alarm is activated, and an emergency notification is sent to the user and safety inspector.

7. The indoor gas leak monitoring and intelligent control system according to claim 6, characterized in that, The platform layer supports access to HomeKit, Mi Home, and Huawei HarmonyOS platform, enabling collaborative control with multi-level actuators; It enables collaborative device control through open APIs and features dual-mode redundancy between cloud and local operation. Real-time alarm information is pushed via APP / SMS, and historical data review and leakage trend analysis are supported.

8. The indoor gas leak monitoring and intelligent control system according to claim 7, characterized in that, The perception layer, network layer, and control layer achieve low-latency response through a timing coordination mechanism: the sensor array of the perception layer outputs sampling data at a frequency of 100ms / time, the edge computing node takes less than 200ms for fusion processing of the sampling data and AI inference, and the execution mechanism of the control layer completes the action response in less than 500ms after receiving the trigger command. At the same time, it reports data to the platform layer synchronously through the network layer, forming a closed-loop control timing chain.