Device and method for detecting suffocating gas in deep pit operation

Abstract

The invention discloses a device and a method for detecting suffocating gas in deep pit operation, and belongs to the technical field of gas detection. Comprising the following steps: constructing a detection model; constructing a real-time response system through starting-up self-calibration and setting of a dynamic threshold value; multi-modal data are collected and processed, gas collection is optimized by adopting a spiral flow guide gas chamber and a turbulent flow fan, the detection precision is improved by combining wavelet noise reduction and LSTM dynamic compensation, and a detection result is obtained. Compared with the prior art, through the MXene material sensor and the multi-modal data fusion technology, the detection error rate is reduced to be smaller than or equal to + / -2% FS, and the response speed is increased by four times; an enhanced convection air chamber and dynamic power consumption management (endurance time is longer than or equal to 48 hours) are adopted to adapt to a deep pit complex environment; through three-level linkage alarm (light / sound / touch) and cloud intelligent decision, full-chain safety prevention and control from real-time monitoring to emergency response are realized, the false alarm rate is reduced by 78%, and the safety guarantee efficiency of operators is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of gas detection, and more specifically, to an asphyxiating gas detection device and method for deep pit operation. Background Art

[0002] In the deep pit operation environment, due to the relatively enclosed space and limited ventilation conditions, the operation area is extremely likely to become a "hot spot" for the accumulation of harmful gases. These harmful gases include, but are not limited to, carbon dioxide ( ), carbon monoxide ( ), and hydrogen sulfide ( ), etc. They pose a great threat to the life safety of operators, and may even trigger dangerous accidents such as poisoning and asphyxiation in severe cases. Among them, although carbon dioxide is regarded as a non-toxic gas under normal conditions, its high concentration state will significantly displace oxygen in the air, resulting in a sharp drop in the oxygen content, and then triggering the risk of oxygen deficiency and asphyxiation of operators. It should be noted that the density of carbon dioxide is greater than that of air, and this physical property makes it easier to accumulate in low-lying and limited spaces such as deep pits, thus further exacerbating the danger of the operation environment.

[0003] Traditional carbon dioxide gas detection devices are unable to cope with the challenges of the deep pit operation environment. These devices are often bulky and inconvenient to carry to the operation site, severely limiting their flexibility in practical applications. More critically, they usually lack real-time data transmission and alarm functions, which means that operators cannot timely obtain the dynamic change information of the carbon dioxide gas concentration and are difficult to take necessary protective measures before the occurrence of danger. In addition, traditional detection devices also have obvious deficiencies in detection accuracy and response speed, and cannot meet the urgent needs of high-precision and fast-response in deep pit operations.

[0004] In related technologies, gas detection devices mostly adopt the single-point static sampling method, and both the sampling period and the detection response time exceed the safety threshold, unable to capture the sudden change process of gas concentration. Moreover, traditional detection systems adopt a discrete sensor design, and there is a lack of a data fusion mechanism between gas channels. In actual working conditions, the coexistence of other gases and carbon dioxide gas will significantly affect the selectivity of electrochemical sensors, and the single sensor calibration method cannot establish a multi-objective optimization model, resulting in a high detection error rate in a composite gas environment. In addition, the alarm method in traditional detection methods is single (only sound and light alarm), and the alarm threshold is often set layer by layer through physical buttons, lacking a dynamic configuration function, unable to quickly adapt to different deep pit operation scenarios, and having a high risk of false alarms in a noisy environment.

[0005] As can be seen from the above, how to design a miniaturized and intelligent carbon dioxide gas detection device to meet the functions of real-time monitoring and timely alarm at the deep pit operation site is a technical problem that urgently needs to be solved at present. Summary of the Invention

[0006] 1. Technical Problem to be Solved In view of the problems in the prior art that existing gas detection devices cannot achieve real-time multi-gas monitoring and intelligent early warning in the dynamic environment of deep pits, and there are problems such as response delay, cross-sensitivity, and data islands, the present invention provides a detection device and method for asphyxiating gases in deep pit operations, which can achieve real-time monitoring of carbon dioxide gas concentration and give timely alarms, quickly detect and warn of situations such as increased or short-term sudden increase in the content of asphyxiating gases during deep pit operations, resulting in difficulty in breathing and physical loss, etc., effectively reducing potential safety hazards for personnel in deep pit operations and ensuring the safety of workers.

[0007] 2. Technical Solution The object of the present invention is achieved through the following technical solutions.

[0008] This section of the application is used to briefly introduce concepts, which will be described in detail in the following detailed implementation section. This section of the application is not intended to identify the key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0009] Some embodiments of the present application propose a detection device and method for asphyxiating gases in deep pit operations to solve the technical problems mentioned in the above background section.

[0010] As a first aspect of the present application, some embodiments of the present application provide a detection method for asphyxiating gases in deep pit operations, including the following steps: Construct a detection model through standard gas calibration, multi-dimensional environmental parameter compensation, multi-modal sensor integration, and machine learning modeling; Construct a real-time response system through power-on self-calibration and setting of dynamic thresholds; Collect and process multi-modal data, optimize gas collection by using a spiral diversion air chamber and a turbulent fan, and improve detection accuracy by combining wavelet denoising and LSTM dynamic compensation to obtain detection results.

[0011] Furthermore, the steps of constructing the detection model include: obtaining initial concentration data, and removing noise and outliers; using a two-stage calibration method to compare and analyze the initial concentration data with the standard gas concentration, adjusting the output characteristics of the sensor to make the measured value of the detection device match the true value, and establishing a sensor reference model; constructing an interdigitated electrode of a carbon dioxide electrochemical sensor, and setting a multi-dimensional parameter space to construct a machine learning model for carbon dioxide concentration detection.

[0012] Further, the multi-dimensional parameters include carbon dioxide, temperature, humidity, and oxygen; the carbon dioxide concentration value is 1000 ppm - 10000 ppm; the temperature value is 10°C - 50°C; the humidity value is 20%RH - 90%RH; the oxygen concentration value is 100 ppm - 1000 ppm.

[0013] Further, the interdigital electrodes of the carbon dioxide electrochemical sensor are made of a composite material.

[0014] Further, in the process of constructing the detection model, the oxygen sensor, temperature sensor, humidity sensor, and carbon dioxide probe are encapsulated together, and data synchronization acquisition is realized through the SPI bus.

[0015] Further, the LSTM dynamic compensation includes temperature compensation and humidity compensation. The temperature compensation automatically corrects the concentration value according to the temperature change coefficient, and the humidity compensation adjusts the data based on the humidity change coefficient.

[0016] Further, the standard for temperature compensation is: for every 1°C increase in temperature, the compensation value increases by 0.3%; the standard for humidity compensation is: for every 10% increase in humidity, the compensation value decreases by 0.015%.

[0017] As a second aspect of the present application, some embodiments of the present application provide an asphyxiating gas detection device for deep pit operations, which is used to implement the above-mentioned asphyxiating gas detection method for deep pit operations, and includes a housing, a main controller, a sensor unit, a power supply, a fan, an alarm control unit, and an interaction unit; the main controller is electrically connected to the sensor unit, the power supply, the fan, and the interaction unit respectively to realize the control and data processing of each component; the main controller includes a display control module, a key control module, an alarm control module, and a storage module; the interaction unit is electrically connected to the display control module, and the power supply is electrically connected to the key control module.

[0018] Further, a centrifugal fan is built into the housing, and honeycomb-shaped air inlet and outlet holes are provided on the front and rear side walls of the housing.

[0019] 3. Beneficial effects Compared with the prior art, the advantages of the present invention are as follows: (1)This detector uses an electrochemical sensor for detection and senses the surrounding environmental gases through diffusion. The instrument is equipped with multiple independent sensors for different gases to be detected, such as carbon dioxide. This design allows users to flexibly configure any number of sensors according to actual needs to achieve the mixed detection of multiple gases. At the same time, each sensor has an individual switch control function, effectively preventing the normal use of other sensors from being affected by the damage of a single sensor, ensuring the continuity and reliability of the detection work. In addition, the instrument also has extremely high sensitivity and can quickly and effectively detect asphyxiating gases in the deep pit operation environment, providing timely and accurate gas concentration information for the operators; (2)The display system of the detector uses a high-definition OLED display screen with a backlight design, which can stably display the on-site detection data under high brightness and large viewing angles, ensuring that the operators can clearly read important information such as gas concentration under various light conditions. In terms of power supply, the detector is equipped with a large-capacity lithium polymer battery pack that can be charged cyclically, which is not only energy-saving and environmentally friendly but also provides long-lasting battery life to meet the needs of long-term detection in deep pits and ensure that the detection work will not be interrupted due to insufficient power during the operation; (3)Convenient operation and durability: This detector is simple and convenient to operate, with a light body design, making it easy to carry and operate, reducing the burden on the operators. The instrument shell is made of high-strength materials, with good anti-pressure and anti-drop performance, and can effectively resist accidental situations such as collisions and drops that may occur in the deep pit operation environment, ensuring the integrity of the instrument. At the same time, the shell has excellent sealing performance and is waterproof, and can work normally in a humid environment. In addition, the shell also has excellent heat resistance and cold resistance, adapting to the operation requirements under different temperature conditions. It is worth mentioning that the shell is made of environmentally friendly materials, avoiding secondary pollution to the environment and conforming to modern environmental protection concepts; (4)Intelligent alarm system: The detector has an intelligent alarm function and can customize the alarm standard according to different environments and user needs. The instrument is set with three alarm methods, including light alarm, buzzer alarm, and vibration alarm. When the detected gas concentration is higher than the set value, the indicator light will show red, and at the same time, a high-decibel buzzer sound will be emitted and the vibration alarm will be activated, ensuring that in the noisy deep pit operation environment, the operators can promptly notice the situation of gas concentration exceeding the standard and take corresponding safety measures to effectively avoid the occurrence of asphyxiation accidents; (5)Through the MXene material sensor and multi-modal data fusion technology, the detection error rate is reduced to ≤±2% FS, and the response speed is increased by 4 times; by adopting an enhanced convection gas chamber and dynamic power consumption management (battery life ≥ 48 hours), it is adapted to the complex environment of deep pits; through three-level linkage alarm (light / sound / tactile) and cloud intelligent decision-making, a full-chain safety prevention and control from real-time monitoring to emergency response is realized, and the false alarm rate is reduced by 78%. Description of the Drawings

[0020] Figure 1 Schematic diagram of the steps of the asphyxiating gas detection method for deep pit operation in an embodiment of the present invention; Figure 2 Schematic diagram of the structure of the asphyxiating gas detection device for deep pit operation in an embodiment of the present invention; Figure 3 Schematic diagram of the external structure of the asphyxiating gas detection device for deep pit operation in an embodiment of the present invention; Figure 4 Schematic diagram of the side wall with air holes of the asphyxiating gas detection device for deep pit operation in an embodiment of the present invention; Figure 5 Schematic diagram of the logic of the asphyxiating gas detection method for deep pit operation in an embodiment of the present invention.

[0021] Description of the reference numerals in the figure: 1. Outer shell; 2. Main controller; 3. Sensor unit; 4. Power supply; 5. Fan; 6. Alarm control unit; 7. Interaction unit; 8. Air inlet and outlet holes; 9. Display control module; 10. Button control module; 11. Alarm control module; 12. Storage module; 13. Algorithm module; 14. Vibrator; 15. Gas delivery channel; 16. Liquid crystal display; 17. LED indicator light; 18. 95dB buzzer; 19. Quick display screen for network disconnection; 20. Wearable connection hole. Detailed implementation manners

[0022] The present invention will be described in detail below with reference to the accompanying drawings of the specification and specific embodiments.

[0023] Combined with Figures 1 to 5 , the asphyxiating gas detection device for deep pit operation of the present invention includes an outer shell 1, a main controller 2, a sensor unit 3, a power supply 4, a fan 5, an alarm control unit 6 and an interaction unit 7. The main controller 2, the sensor unit 3, the power supply 4, the fan 5 and the alarm control unit 6 are all arranged inside the outer shell 1, the interaction unit 7 is embedded in the outer shell 1, and the sensor unit 3, the power supply 4, the fan 5, the alarm control unit 6 and the interaction unit 7 are all electrically connected to the main controller 2.

[0024] Specifically, the outer shell 1 is made of explosion-proof aluminum alloy material, and honeycomb-shaped air inlet and outlet holes 8 are provided on the front and rear side walls to ensure smooth gas flow. The outer shell 1 adopts an IP67 (International Protection Marking Level 67) explosion-proof structure, and honeycomb-shaped air inlet and outlet holes 8 are provided on its front and rear side walls to ensure air flow circulation. The fan 5 is a centrifugal fan 5 and is built into the outer shell 1 to achieve forced ventilation.

[0025] The main controller 2 includes a display control module 9, a button control module 10, an alarm control module 11, and a storage module 12. The interaction unit 7 is electrically connected to the display control module 9, the power supply 4 is electrically connected to the button control module 10, and the alarm control module 11 is electrically connected to the storage module 12.

[0026] The sensor unit 3 is fixedly connected to the gas delivery channel 15. Both the liquid crystal display 16 and the offline quick display screen 19 are electrically connected to the display control module 9, and both the liquid crystal display 16 and the offline quick display screen 19 are disposed on the surface of the housing 1. The wearable connection holes 20 are used to be fixed on work clothes or the arm and are disposed on both the left and right sides of the housing 1.

[0027] In a specific embodiment, the main controller 2 uses a control system with the STM32F429 chip (STMicroelectronics 32-bit F4 Series Microcontroller) as the core. By integrating the WiFi / Bluetooth dual-mode communication function, wireless data transmission and remote interaction are realized. In addition, the main controller 2 also integrates edge computing capabilities and built-in threshold judgment algorithms. When the concentration exceeds the preset limit, a three-level early warning mechanism is synchronously triggered by a three-color LED (Light-Emitting Diode) warning light (the liquid crystal display synchronously displays the alarm code), a 95dB buzzer (95 Decibel Buzzer), and a vibrator 14. And the alarm data packet is pushed to the cloud platform through the MQTT protocol (Message Queuing Telemetry Transport).

[0028] In terms of data transmission, the deep pit operation asphyxiating gas detection device adopts a dual storage architecture of local TF card storage (circular overwrite mode) and cloud database, and realizes real-time data transmission through the WiFi / Bluetooth module, supporting HTTP / HTTPS dual-protocol communication. It not only ensures the integrity and security of data, but also provides a more convenient data management and query method for users.

[0029] In addition, the main controller 2 also integrates the WiFi / Bluetooth dual-mode communication protocol (supporting MQTT / HTTP / HTTPS), and can process and analyze the data collected by the sensors in real time. Combining the adaptive threshold algorithm and the LSTM neural network algorithm (Long Short-Term Memory Neural Network Algorithm), the main controller 2 can call the LSTM neural network algorithm through the algorithm module 13 to dynamically compensate for the cross-sensitivity of the sensors, and adjust the alarm threshold in real time according to the environmental parameters (such as temperature, humidity, air pressure) through the adaptive threshold algorithm, thereby improving the accuracy and reliability of detection.

[0030] The sensor unit 3 includes a carbon dioxide electrochemical sensor and a vibration sensor. The carbon dioxide electrochemical sensor is used to detect the concentration of carbon dioxide. The sensor unit 3 is electrically connected to the algorithm module 13, and the main controller 2 is electrically connected to the algorithm module 13 to receive the electrical signals sent from the sensor unit 3.

[0031] Specifically, the carbon dioxide electrochemical sensor is based on the self-developed titanium carbide (Ti 3 C 2 T x substrate material, and realizes the dual detection of carbon dioxide concentration through the chemical resistance principle, ensuring high-precision and stable detection. The carbon dioxide electrochemical sensor uses SMD (Surface-Mounted Device) packaging to improve its stability and anti-interference ability.

[0032] The alarm control module 11 includes a three-color LED indicator 17 and a high-decibel buzzer 18. The LED indicator 17 is an adjustable-color lamp and can display different colors according to different alarm levels. The high-decibel buzzer 18 uses the high-decibel buzzer 18 of the SFM-27 model. This model buzzer is an active sounder with a working voltage of DC3-24V and can emit a sound of 95dB, ensuring that it can be clearly heard in a noisy environment.

[0033] Specifically, the vibration sensor is a 590C embedded three-axis vibration chip sensor. The vibration sensor works in coordination with the three-color LED indicator 17 and the high-decibel buzzer 18 to jointly form a three-level early warning mechanism. When it is detected that the concentration of harmful gases exceeds the standard, the vibration sensor will emit a slight vibration, the LED indicator 17 will flash a specific color, and the buzzer will emit a loud alarm sound, thereby prompting to take corresponding measures in time.

[0034] To ensure the reliability and stability of the device, each sensor adopts an independent power supply design and supports plug-and-play expansion. At the same time, each sensor is equipped with a separate switch control function, effectively avoiding the normal operation of the entire detection system being affected by the failure of a single sensor and ensuring the continuity and reliability of the detection work.

[0035] The interaction unit 7 includes a display module and a button module. The display module includes an OLED (Organic Light-Emitting Diode) display screen, using an OLED module of the SSD1306 model, supporting two communication methods of I2C (Inter-Integrated Circuit) and SPI (Serial Peripheral Interface), with a resolution of 128×64 or 132×32 selectable. The display screen has a built-in multi-level menu system, and different menus can be switched through touch buttons to realize the dynamic configuration of detection parameters.

[0036] The button module includes touch function keys, using a GTX312L (Capacitive Touch Sensor IC) capacitive touch button chip, which supports single-point and multi-point touch, I2C communication protocol, and 32QFN package. The OLED display screen selects an OLED module of the SSD1306 model, which supports two communication methods of I2C and SPI. The resolution is generally selected as 128×64 or 132×32. The display screen has a built-in multi-level menu system, and users can switch between different menus through touch buttons to realize the dynamic configuration of detection parameters, such as setting the type of detected gas, concentration unit, alarm threshold, etc.

[0037] In a specific embodiment, the power supply 4 adopts a high-density lithium polymer battery pack, combined with a dynamic power consumption management algorithm, and automatically switches to the Burst mode (Burst Power Saving Mode) to reduce energy consumption during wireless transmission. A 5000mAh lithium polymer battery pack is used to achieve a battery life of ≥48 hours through dynamic power consumption management (Burst mode); data is circularly stored in a local TF card (FAT32 format) and synchronized with the cloud (Alibaba Cloud / Amazon Web Services), supporting resume from breakpoint and export of OSHA standard (Occupational Safety and Health Administration Standards) reports.

[0038] Such as Figure 5As shown in the figure, a method for detecting asphyxiating gases in deep pit operations according to the present invention uses the above-mentioned deep pit operation asphyxiating gas detection device, and includes the following steps: S1. Build a detection model In this step, a reference calibration system is built through standard gas calibration, multi-dimensional environmental parameter compensation, multi-modal sensor integration, and machine learning modeling, so as to realize the construction of the detection model of the detection device. First, precise original data is obtained through hardware filtering and algorithm denoising, and temperature and humidity interference is eliminated by combining two-stage linear / non-linear calibration; subsequently, a random forest model is trained based on the full-factor experimental data to achieve high-precision dynamic compensation of carbon dioxide concentration in complex environments, and a reliable standardized reference is established for real-time monitoring of deep pit operations.

[0039] S101. Data collection and preprocessing: Under a sealed state, connect the detection device to a carbon dioxide standard gas with a concentration of 2000 ppm, and start the detection device.

[0040] After the detection device is started, the carbon dioxide electrochemical sensor in its sensor unit will output an analog current signal. After the analog current signal is denoised by the hardware filtering circuit, it is transmitted to a 24-bit high-precision ADS1256 analog-to-digital conversion module for analog-to-digital conversion to complete the digital processing of the signal, so as to obtain the initial concentration data.

[0041] Preprocess the obtained initial concentration data, including removing noise and outliers, etc., to ensure the accuracy and reliability of the data, and provide a basis for subsequent processing and analysis.

[0042] Specifically, a filtering algorithm is used to eliminate noise and identify and exclude outliers to ensure the accuracy and high reliability of the initial concentration data. The sliding window median filtering (window width = 5) and 3σ criterion outlier rejection algorithm are adopted for the initial concentration data to eliminate the influence of environmental electromagnetic interference and transient noise.

[0043] In a specific embodiment, in the face of significant noise values introduced by environmental electromagnetic interference in the initial concentration data sequence, random jump noises of ±500 ppm (such as jumping data such as 1500 ppm and 2500 ppm) appear near the 2000 ppm reference value in the initial concentration data sequence. These noises can be effectively removed through the preset filtering algorithm, making the initial concentration data smooth and stable. After being processed by the filtering algorithm, the output standard deviation drops from ±2.5% FS to ±0.8% FS, and the data stability is significantly improved. At the same time, the initial concentration data is preliminarily processed to remove outliers to ensure the accuracy and reliability of the data.

[0044] S102. Establish a sensor reference model: Adopt a two-stage calibration method to compare and analyze the preprocessed initial concentration data with the known standard gas concentration, adjust the output characteristics of the sensor, and make the measured value of the detection device match the true value, thereby initially establishing a reference model for the sensor.

[0045] Specifically, based on the least squares method to fit the sensor output characteristic curve, the established linear calibration model, that is, the reference model of the sensor, can be expressed as: , where C calibrated represents the calibrated concentration value, C measured represents the preprocessed initial concentration data, and a and b are calibration parameters obtained by least squares fitting.

[0046] In a specific embodiment, in a laboratory environment, a 2000ppm carbon dioxide standard gas is used for calibration. After two-stage calibration, the measured value can stably remain within the range of 2000ppm ± 2%. At this time, the specific values of a and b can be obtained through calculation. For example, a = 1.005 and b = -25, so that a high degree of consistency is achieved between the calibrated concentration value and the true value.

[0047] Specifically, on the basis of two-stage calibration, further consider the influence of environmental factors in depth. Environmental factors include temperature and humidity, etc. By sending a storage instruction through the touch button, the main controller will quickly call the LSTM neural network algorithm built into the control system, and combine multi-dimensional data such as the temperature and humidity collected in real time to perform non-linear fitting and multi-dimensional calibration on the sensor reference model. In this process, comprehensive multi-dimensional calibration parameters including temperature coefficients and humidity coefficients will be generated, and these parameters will be written into the Flash memory of the TF card, that is, the complete standard curve will be written into the TF card Flash memory for subsequent use.

[0048] In a specific embodiment, in a laboratory environment, when the temperature is 25°C and the humidity is 60%RH (Relative Humidity), a 2000ppm carbon dioxide standard gas is measured by the detection device. After two-stage calibration, the measured value is stable within the range of 5000ppm ± 2%. At this time, by sending a storage instruction through the touch button, the main controller calls the LSTM neural network algorithm, combines multi-dimensional data such as the temperature and humidity collected in real time, and performs non-linear fitting and multi-dimensional calibration on the sensor reference model. After calibration, multi-dimensional calibration parameters including temperature coefficients and humidity coefficients are generated, and these parameters are stored in the Flash memory of the TF card to ensure accurate carbon dioxide concentration detection under different environmental conditions.

[0049] During the multi-dimensional parameter calibration process, a polynomial regression model can be adopted, expressed as: , where T represents temperature, H represents humidity, and a, b, c, d, e, f are multi-dimensional calibration parameters obtained by fitting training data.

[0050] In a specific embodiment, experiments are conducted under different temperature and humidity conditions, and a large amount of data is collected for fitting to obtain specific parameter values: a = 0.998, b = 1.2, c = 0.5, d = 0.002, e = 0.003, f = -10, so as to be able to calibrate sensor data more accurately and significantly improve detection accuracy.

[0051] S103. Design of multi-modal sensing probe and synchronous data acquisition: Use Ti 3 C 2 T X -MXene composite material to construct the interdigital electrode of a carbon dioxide electrochemical sensor. Specifically, a micro-nano structure with a pitch of 80 μm and a width of 50 μm is formed through a lithography process, and Pt nanoparticles are loaded on the surface of the micro-nano structure to enhance the electrochemical activity, so as to construct the interdigital electrode, thereby reducing the detection limit of carbon dioxide to 50 ppm. Experimental data shows that the sensor with this innovative electrode structure has higher sensitivity and accuracy in a low-concentration carbon dioxide detection environment. Compared with the traditional electrode structure, its detection limit value is reduced from 200 ppm to 50 ppm, and it can more effectively detect low-concentration carbon dioxide changes and is suitable for various precision detection scenarios.

[0052] At the same time, an oxygen sensor, a temperature sensor, and a humidity sensor are co-packaged with the carbon dioxide probe and packaged in an explosion-proof enclosure. Synchronous data acquisition is achieved through the SPI bus to eliminate the timing error of traditional split probes and ensure the synchrony and consistency of multi-parameter data.

[0053] Specifically, the oxygen sensor uses ZrO 2Solid electrolyte (Zirconium Dioxide Solid Electrolyte), the temperature sensor uses a Pt1000 thin film (Platinum 1000 Ohm Thin-Film Resistance Temperature Detector, with a resistance value of 1000 Ω at 0 °C and a temperature coefficient α = 0.00385 / °C), and the humidity sensor uses an interdigital capacitive type. In this embodiment, the housing adopts the IP68 rating. Thus, the sampling rate for synchronous data acquisition is as high as 10 Hz, completely eliminating the timing error of traditional split probes and ensuring the synchronization and consistency of multi-parameter data.

[0054] In actual application scenarios, the multi-parameter sensor can synchronously collect data, and can obtain 10 sets of complete and accurate multi-parameter data per second, including carbon dioxide concentration, oxygen concentration, temperature, humidity, etc., providing rich and accurate data support for subsequent data analysis and environmental assessment, and effectively avoiding analysis errors caused by asynchronous data collection.

[0055] S104, Dynamic Environment Compensation and Machine Learning Modeling: Set up a multi-dimensional parameter space in the environmental simulation chamber, and the multi-dimensional parameters include carbon dioxide, temperature, humidity, and oxygen.

[0056] Specifically, among the multi-dimensional parameters, the carbon dioxide concentration value is 1000 ppm - 10000 ppm, and a total of 20 gradients can be designed using uniform division and equal spacing; the temperature value is 10 °C - 50 °C, and a total of 6 gradients can be designed using uniform division and equal spacing; the humidity value is 20%RH - 90%RH, and a total of 5 gradients can be designed using uniform division and equal spacing; and the oxygen value is 100 ppm - 1000 ppm, and a total of 4 gradients can be designed using uniform division and equal spacing.

[0057] A total of 2,400 sets of comprehensive calibration data were obtained through a full factorial experiment. Through experimental verification: when the carbon dioxide concentration endpoint value is 1,000 ppm, the sensor measurement error rate is ±1.2% (linearity R2 = 0.995), and the deviation after compensation < ±0.8% FS; when it is 10,000 ppm, the measurement error rate is ±1.8% (linearity R2 = 0.988), and the deviation after compensation < +1.5% FS, both meeting the ±2% FS accuracy requirement. When the temperature endpoint value is 10 °C, the concentration deviation after temperature compensation is -0.5% (the actual detected value is 1,990 ppm when the standard value is 2,000 ppm); when it is 50 °C, the deviation after temperature compensation is +0.4% (the actual detected value is 2,008 ppm), and the compensation stability > 98%. When the humidity endpoint value is 20% RH, the concentration deviation after humidity compensation is +0.3% (the detected value is 2,006 ppm when the standard value is 2,000 ppm); when it is 90% RH, the deviation after compensation is -0.6% (the detected value is 1,988 ppm), and the humidity interference suppression rate > 95%. When the oxygen concentration endpoint value is 100 ppm, the sensor response time is 10 seconds (error ±1.0%), meeting the rapid detection requirement; when it is 1,000 ppm, the response time is 12 seconds (error ±1.3%), and the response speed is 4.2 times faster than that of traditional sensors. For example, in an environmental simulation chamber, the multi-dimensional parameters are gradually adjusted according to the set gradient changes and kept in a stable state for a period of time after each adjustment to ensure that the sensor can accurately capture the corresponding values. Finally, 2,400 sets of calibration data covering different environmental conditions were obtained, providing a rich sample basis for the training of subsequent detection models.

[0058] Specifically, feature engineering is performed on the normalized (Min-Max Scaling) data, and variables with a Gini coefficient > 0.3 are selected as the input features of the detection model. During the training process of the detection model, the hyperparameters are set as follows: the number of decision trees is 100, the maximum depth is 10, and the minimum number of samples in a leaf is 5.

[0059] After a rigorous 5-fold cross-validation process, the R² of the model on the test set reached 0.983, which fully demonstrates the model's high-efficiency compensation ability for complex environmental factors and its powerful modeling ability. Finally, a machine learning model for carbon dioxide concentration detection, namely the detection model, was successfully constructed. Among them, R² (Coefficient of Determination) is an index to measure the fitting degree between the predicted value and the true value of the model, ranging from 0 to 1.

[0060] In a specific embodiment, during the feature engineering process, by analyzing the normalized data, it is found that there is a significant correlation between multiple variables such as carbon dioxide concentration, temperature, humidity, and oxygen concentration and the detection results. Among them, the Gini coefficients of carbon dioxide concentration is 0.45, temperature is 0.35, humidity is 0.32, and oxygen concentration is 0.33, etc., all meet the condition that the Gini coefficient > 0.3, so they are selected as important input features of the model. After training and optimizing the random forest model, in the test set, the coefficient of determination R² between the predicted value and the true value is as high as 0.983, which indicates that the model can well fit the data and accurately reflect the influence of environmental factors on the carbon dioxide concentration detection results, thus realizing effective compensation and accurate modeling of complex environments.

[0061] S2. Calibration and Setting In this step, a real-time response system of the detection device is constructed through power-on self-calibration and dynamic threshold setting. First, a three-level calibration protocol is executed in a clean environment of ISO 14644-1 (International Organization for Standardization 14644-1 Cleanrooms Standards), combined with wavelet denoising and dynamic power management; then, a four-level alarm interval is divided, integrating sound-light-touch composite alarm and Modbus RTU (Modbus Remote Terminal Unit Protocol) remote threshold synchronization function to ensure that the ventilation equipment is quickly linked when the carbon dioxide concentration exceeds the limit in the deep pit environment, and improve the end-to-end response efficiency.

[0062] S201. Power-on Self-calibration Process: Each time the device is powered on, the detection device starts the precision calibration process. The detection device is started in the clean air specified by the ISO 14644-1 standard (that is, the number of particles greater than or equal to 0.5 microns in the air does not exceed 3520 per cubic meter). After power-on, the main controller immediately receives the current signal output by the gas detection sensor and automatically performs zeroing processing to lay a precise foundation for the detection work. After the calibration is completed, the calibration result will be transmitted to the display screen.

[0063] In a specific embodiment, the detection device performs a three-step adaptive calibration when starting up: First, initialize the main controller (embedded with the clock source of the STM32F429 main control chip), detect the initial states of the BMP280 (Bosch BMP280 Digital Pressure Sensor) pressure sensor and the SHT35 (Sensirion SHT35 Humidity and Temperature Sensor) humidity and temperature sensor, and construct an environmental feature matrix containing temperature, relative humidity, and atmospheric pressure.

[0064] Subsequently, start the gas flow. The main controller reads the sensor baseline data through the SPI interface, and at the same time calls the Daubechies 5-level wavelet transform algorithm to suppress the noise of the current signal output by the gas detection sensor to ensure the purity of the signal. During this process, the dynamic power management module is synchronously activated to optimize energy utilization. After calibration is completed, the calibration results are output to the display screen through the SSD1306 driver.

[0065] S202. Dynamic threshold setting: According to the characteristics and safety requirements of the deep pit operation environment, the alarm value of the detection device can be flexibly set through the interaction module. Once the main controller receives the instruction to set the alarm value, the detection device enters the alarm setting state. In this state, the detection device selects the alarm category and the alarm setting value according to the instruction.

[0066] Specifically, the alarm categories include light alarm, buzzer alarm, and vibration alarm. The alarm setting values include low alarm value, high alarm value, extreme concentration alarm value, and safety concentration alarm value. According to the different detected concentrations, the detection device can issue different levels of alarms. When the detected concentration is within the safe range, that is, when the detected concentration is lower than the alarm setting value, the indicator light shows green; when the detected concentration is higher than the alarm setting value, the indicator light turns red, the buzzer sounds, and the vibration alarm is synchronously activated to form an all-round warning.

[0067] In a specific embodiment, alarm parameters are input through the GTX312L (GTX312L Capacitive Touch Sensor, IC capacitive touch sensor integrated circuit) capacitive touch keys. The main controller calls the pre-compiled threshold setting program to divide the safe area (green), warning area (yellow), danger area (red), and emergency area within the voltage range of 0.5 - 4.5V respectively. In addition, a dynamic threshold algorithm is adopted to achieve automatic adjustment of the day and night mode, enabling the detection device to automatically adjust the alarm threshold according to the day and night mode. When the concentration exceeds the alarm set value, the hardware cascade mechanism will respond quickly and immediately trigger the three-color LED indicator lights (with wavelengths of 520nm green, 590nm yellow, and 620nm red), the SFM-27 high-loudness buzzer, and the three-axis vibrator. At the same time, an external ventilation device is directly driven through the GPIO (General-Purpose Input / Output) port. Among them, the system reserves a Modbus RTU interface, supporting remote synchronization of the alarm threshold with the SCADA system (Supervisory Control and Data Acquisition System) to achieve multi-level safety protection.

[0068] Through the above process, each time the device is powered on, the detection device will perform a calibration process to ensure startup in the clean air specified by the ISO 14644-1 standard, and receive and process the signals of the gas detection sensors through the main controller. A three-step adaptive calibration is adopted to construct an environmental feature matrix, the wavelet transform algorithm is applied for noise suppression, and dynamic power management is synchronously activated. The calibration results will be displayed on the display screen.

[0069] In addition, according to the safety requirements of the deep pit operation environment, users can flexibly set the alarm values of the detection device through the interaction module. The alarm categories include light, buzzer, and vibration alarms, and the alarm set values cover multiple levels. Alarm parameters are input through the touch keys, and the dynamic threshold algorithm is applied to achieve automatic adjustment of the day and night mode. When the concentration exceeds the set value, multi-level alarms are triggered and an external device is driven through the GPIO port. The system also supports remote synchronization of the alarm threshold with the SCADA system to achieve multi-level safety protection.

[0070] S3. Multi-modal data collection and processing This step realizes the precise monitoring and rapid response of the deep pit gas environment through multi-source sensing fusion and intelligent analysis. The spiral diversion gas chamber and turbulent fan (3-second gas replacement) are used to optimize gas collection, combined with wavelet noise reduction and LSTM dynamic compensation to achieve a detection accuracy of ±1.1% in high-temperature and high-humidity environments; a four-level alarm system and remote threshold linkage mechanism are constructed to support real-time adjustment of Modbus TCP (Modbus Transmission Control Protocol, a Modbus protocol based on TCP / IP); dual-redundant storage (128GB TF card + W25Q128 flash memory) and AES-256-GCM (Advanced Encryption Standard 256-bit Galois / Counter Mode) encrypted transmission are designed to reduce the data loss rate. In case of anomalies, the supercapacitor bank maintains emergency power supply to ensure the robustness and continuity of the system under complex working conditions.

[0071] S301. Gas collection and pretreatment: Based on the shell of the detection device with spiral diversion air inlet and outlet holes, the gas to be measured is guided into the enhanced gas chamber.

[0072] Specifically, the detection device is equipped with a built-in turbulent fan (adjustable speed ≥ 2000 RPM), which can accelerate gas flow. Collaborating with the shell design of spiral diversion honeycomb air inlet and outlet holes, it can ensure the gas replacement in the gas chamber within 3 seconds, improving the detection efficiency.

[0073] The carbon dioxide electrochemical sensor ionizes gas based on the principle of chemical resistance, accurately ionizes gas, and outputs an analog current signal of 0.5 - 4.5V. The sensitivity of the analog current signal is not less than 0.1mV / ppm. At the same time, the oxygen sensor adopts a physically isolated cavity design to collect data synchronously, effectively avoiding cross-interference between gases.

[0074] In addition, the sensor unit of the detection device also integrates an SHT35 temperature and humidity sensor and a BMP280 pressure sensor to record the temperature (T), relative humidity (RH), and pressure value in real time at a sampling rate of 10Hz (10 times per second), providing basic data for subsequent dynamic compensation.

[0075] To further improve the measurement accuracy, a replaceable filter screen is added at the air inlet to effectively block large particle impurities from entering. The inner wall of the gas chamber is coated with an anti-adsorption material, which can significantly reduce gas residue and adsorption, ensuring the accuracy and reliability of the measurement results.

[0076] S302. Signal processing and dynamic compensation: The analog current signal is first converted into a digital signal by a 24-bit high-precision analog-to-digital conversion module (ADS1256). Aiming at the complex and variable noise interference in the deep pit environment, a wavelet transform algorithm is used for noise reduction to effectively eliminate baseline drift. At the same time, the above-mentioned pre-trained LSTM neural network model is called to perform deep learning and dynamic compensation on the sensor measurement data, significantly improving the data accuracy. The calibration module of the detection device regularly calls the standard curve to perform baseline calibration to ensure the measurement accuracy.

[0077] To optimize the analog-to-digital conversion module, a reference voltage source with higher accuracy than the general reference voltage source (model REF3125, output voltage 2.5V) is adopted and a shielding cover is added to effectively reduce external electromagnetic interference. At the same time, an adaptive filtering technology is introduced to further improve the signal-to-noise ratio and stability of the signal, ensuring that the digital signal resolution reaches 0.01ppm.

[0078] The wavelet transform algorithm is based on the Daubechies5 basis function, and signal noise reduction is achieved through the wavelet transform algorithm to eliminate baseline drift.

[0079] Specifically, the calibration module of the detection device automatically calls the standard curve pre-stored in the TF card every 30 seconds to perform baseline calibration, ensuring that the measurement zero-point drift correction amount is controlled within ±1% of the full scale. In a specific embodiment, the STM32F429 main controller calls the pre-trained LSTM neural network model to perform dynamic compensation on the sensor data. The LSTM dynamic compensation includes temperature compensation and humidity compensation. The temperature compensation automatically corrects the concentration value according to the temperature change coefficient, and the humidity compensation adjusts the data based on the humidity change coefficient.

[0080] Specifically, the temperature change coefficient α = 0.003 / °C; the standard for the temperature compensation to automatically correct the concentration value according to the temperature change coefficient is: when the temperature rises by 1°C, the compensation value increases by 0.3%; the humidity change coefficient β = 0.0015 / %RH; the standard for the humidity compensation to adjust the data based on the humidity change coefficient is: when the humidity increases by 10%, the compensation value decreases by 0.015%.

[0081] S303. Threshold determination and alarm trigger: Combined with historical data and environmental parameters, the alarm logic is dynamically adjusted to ensure that the alarm mechanism is more accurate and effective. When the gas concentration exceeds the set threshold, the corresponding alarm mechanism is immediately triggered. For example, the threshold is automatically reduced by 10% during night operations to adapt to environmental changes.

[0082] In a specific embodiment, when the carbon dioxide concentration exceeds the low alarm value (500 ppm), the yellow indicator light and intermittent beeping (1 time per second) are triggered; when it exceeds the high alarm value (1500 ppm), the red indicator light and continuous beeping are started; when the limit concentration (3000 ppm) is reached, the triaxial vibrator (amplitude ≥ 2G) is activated, and an emergency alarm code (such as "E01") is pushed to the cloud platform through the Internet of Things module. At the same time, the detection device adds a linkage mechanism for multiple alarm methods, reserves an interface for external devices, and designs a remote adjustment function for the alarm threshold, greatly improving the reliability and flexibility of the alarm system.

[0083] S304. Data storage and exception handling: The local storage module cyclically writes data to the TF card in the CSV (Comma-Separated Values) format, and the cloud communication module encrypts and uploads the data to the cloud database.

[0084] Specifically, the local storage module cyclically writes the original data and processing results to the TF card in the CSV format, and the maximum capacity of the TF card can reach 128GB. When recording key events (such as high-concentration carbon dioxide), the timestamp and GPS positioning information are added for subsequent analysis.

[0085] The cloud communication module uses ZIP compression and AES256 encryption technologies to securely upload data to the cloud database (supporting Alibaba Cloud / AWS) through the MQTT protocol, realizing offline breakpoint resumption and automatic generation of OSHA (Occupational Safety and Health Administration) standard reports.

[0086] The sensor impedance is monitored through the system self-check module (the normal impedance range of the carbon dioxide sensor is 50 - 200Ω). Once an abnormality is found (such as when the impedance > 500Ω), it is considered that an abnormality has occurred at this time. The main controller immediately automatically switches to the standby channel and displays a fault code (such as "F02") on the OLED screen. The data verification module uses the CRC-32 cyclic redundancy check algorithm to ensure data integrity, and immediately starts the bad block isolation and repair process when an abnormality is found. When the main battery voltage is lower than 3.3V, the power management module automatically switches to the supercapacitor bank for power supply to ensure that the data storage and alarm functions continue to operate for at least 10 minutes.

[0087] To further enhance the system reliability, a spare flash chip is added to ensure that the TF card does not lose data in case of failure; the data encryption algorithm is upgraded to the AES-256-GCM mode to add a data integrity verification function; at the same time, a spare battery is added to the power management to further extend the system working time.

[0088] In this step, through the built-in turbulent flow fan and enhanced air chamber design of the detection device, rapid gas replacement is ensured and cross-interference is reduced. The sensor unit integrates temperature and humidity, air pressure, and gas concentration sensors to record environmental parameters in real time. The signal processing module uses high-precision analog-to-digital conversion and wavelet transform algorithms for noise reduction, and combines with the LSTM neural network model for dynamic compensation to improve data accuracy. The intelligent threshold determination and multi-level alarm trigger mechanism are dynamically adjusted according to environmental parameters to ensure timely response. The data storage and exception handling mechanism realizes the secure storage and backup of data through a local TF card and a cloud database. At the same time, functions such as system self-check, data verification, and power management ensure the stable operation of the system and the integrity of the data.

[0089] S4. Obtain the detection result After the detection device completes data processing, the conversion channels built into its main controller quickly convert the current analog signal into a digital signal. Subsequently, this digital signal data is sent by the chip to the OLED display screen and presented to the user. Through the function keys on the display screen, a detection report can be further generated, processing methods can be obtained, and the recorded playback can be viewed.

[0090] The display content on the OLED display screen includes key information such as the concentration of various detected gases and the pollution level, ensuring that operators can understand the safety status of the working environment in a timely and intuitive manner. After the detection is completed, the user can generate a detailed detection report through the function keys on the display screen. The report not only includes gas concentration data but also provides corresponding treatment suggestions to help the user take effective countermeasures in a timely manner.

[0091] After completing the detection task, simply press the function key on the display screen to send a shutdown command to the detection device. After receiving this command, the main controller will immediately control the intake port sensor unit and the main controller itself to stop working, ensuring that the device is in a safe and low-power state when not in use.

[0092] In this step, after the detection device completes data processing, the current analog signal is converted into a digital signal through the conversion channel of the main controller, and key information such as the concentration of the detected gas and the pollution level is clearly displayed on the OLED display screen. The user can easily generate a detection report containing detailed data and treatment suggestions through the function keys on the display screen. After the detection is completed, the user can send a shutdown command, and the main controller controls the relevant units to stop working to ensure the safety of the device. In addition, the system also provides a complete solution for gas safety monitoring in deep pit operations, including a distributed detection network, an intelligent analysis and decision-making module, and an emergency response linkage mechanism, further improving the detection efficiency and safety.

[0093] In a specific embodiment, the present application also provides a suffocating gas detection system for deep pit operations, including a distributed detection network, an analysis and decision-making module, and an emergency response linkage mechanism.

[0094] Specifically, the distributed detection network consists of multiple detection devices, and realizes efficient data transmission through LoRa self-organizing network (Long Range Self-Organizing Network Technology). These detection devices are evenly distributed within the deep pit operation area (the node spacing does not exceed 50 meters) to ensure comprehensive monitoring of the operation environment. Multiple detection devices cover the deep pit operation area through LoRa self-organizing network and transmit data to the cloud platform in real time.

[0095] The cloud platform integrates the random forest algorithm through the analysis and decision-making module, and can predict the mutation risk of gas concentration in real time. At the same time, through the Hololens 2 glasses (Microsoft Hololens 2 Mixed Reality Headset), it realizes the AR visualization function, and superimposes the concentration heat map and the optimal escape path on the field of vision of the operators in real time, providing strong support for their safe operation. When the system detects the limit concentration, it will automatically trigger the emergency response linkage mechanism. This mechanism will not only automatically accelerate the operation of the ventilation system (by PID controlling the fan speed), but also push alarm information to the management personnel in a timely manner, and start the emergency plan database to provide scientific and reasonable decision-making support for the management personnel. Thus, a complete closed loop of "detection - early warning - decision - response" is formed to ensure the safety and stability of the deep pit operation environment.

[0096] In a specific example, that is, in the experiment of the suffocating gas detection device for deep pit operations, by adopting the MXene material sensor and the multi-modal data fusion technology, the detection error rate of the device is significantly reduced, reaching a high-precision level of ≤±2% FS. Taking carbon dioxide detection as an example, when the standard concentration is 2000 ppm, the error rate without using the new technology is 5.2%, and after using the new technology, the error rate is reduced to 1.8%; for oxygen detection, when the standard concentration is 500 ppm, the error rate without using the new technology is 4.7%, and after using the new technology, the error rate is reduced to 1.6%. This shows that the detection accuracy has been significantly improved, and it can more accurately reflect the actual gas concentration.

[0097] In terms of response speed, the MXene material sensor demonstrates obvious advantages, with the response speed increased by 4 times. The response time of the traditional sensor for detecting carbon dioxide is 60 seconds, while with the MXene material sensor, the response time is shortened to 15 seconds; for oxygen detection, the response time of the traditional sensor is 55 seconds, and after adopting the new technology, the response time is shortened to 13.5 seconds. This rapid response ability enables the device to promptly capture changes in gas concentration and provide more timely warnings for operators.

[0098] The battery life of the device has also been significantly improved. After adopting the enhanced convection gas chamber and dynamic power consumption management technology, the battery life reaches ≥48 hours. This enables the device to operate stably for a long time in complex environments such as deep pits, reducing the risk of monitoring interruption caused by insufficient power and ensuring the continuity and reliability of monitoring.

[0099] In terms of the false alarm rate, through three - level linked alarms (light / sound / touch) and cloud - based intelligent decision - making, the false alarm rate of the device is reduced by 78%. When cloud - based intelligent decision - making is not adopted, the false alarm rates of low - level alarm, high - level alarm, and extreme alarm are 15%, 20%, and 25% respectively. After adopting cloud - based intelligent decision - making and the three - level linked alarm mechanism, the false alarm rate of low - level alarm is reduced to 3.2%, the false alarm rate of high - level alarm is reduced to 4.3%, and the false alarm rate of extreme alarm is reduced to 5.5%. This significantly reduced false alarm rate effectively reduces unnecessary alarms, improves the accuracy and reliability of alarms, and thus significantly enhances the safety protection efficiency of operators.

[0100] In a specific embodiment, according to various scenarios in actual applications, including industrial deep pits, urban underground spaces, agricultural farms, and medical facilities, etc., this technical solution can be designed differently as follows to meet the gas monitoring requirements in specific environments and provide full - chain safety protection.

[0101] In the industrial deep pit scenario, the device is installed in a harsh environment such as a mine or a chemical plant pit. The gas to be tested is introduced through a long tube. The air inlet design takes into account the explosion-proof requirements, and special materials are used to prevent sparks. The signal processing optimization algorithm improves the anti-interference ability, and an additional filtering circuit is added to improve the signal stability. The alarm trigger is linked with the on-site ventilation system, and the ventilation equipment is automatically started to reduce the concentration of harmful gases. The alarm signal is transmitted through both wired and wireless methods. The local storage adds waterproof and shockproof design, the cloud communication uses an industrial-grade communication module, the main battery uses a high-capacity industrial battery, and the capacity of the supercapacitor group is increased to ensure that the continuous operation time is greater than 15 minutes in the event of a power outage. The solar charging function is added to extend the battery life. Specifically, in the industrial deep pit scenario, the device is installed in a harsh environment such as a mine or a chemical plant pit. The gas to be tested is introduced through a long tube. The air inlet design takes into account the explosion-proof requirements, and special materials are used to prevent sparks. The signal processing optimization algorithm improves the anti-interference ability, and an additional filtering circuit is added to improve the signal stability. The alarm trigger is linked with the on-site ventilation system, and the ventilation equipment is automatically started to reduce the concentration of harmful gases. The alarm signal is transmitted through both wired and wireless methods. The local storage has added waterproof and shockproof design, the cloud communication adopts industrial-grade communication module, the main battery uses high-capacity industrial battery, the capacity of the supercapacitor group has been increased to ensure that the continuous operation time is greater than 15 minutes in the event of a power outage, and the solar charging function has been added to extend the battery life.

[0102] In urban underground space scenarios, such as underground parking lots, subway tunnels and other crowded places, the device is installed in the ventilation duct, and the existing ventilation system is used for gas collection. The air inlet design considers compatibility with the ventilation system. The signal processing circuit optimizes the shielding design to improve the anti-interference ability, and the digital filtering algorithm is added to improve the signal accuracy. The alarm trigger is linked with the building fire protection system and ventilation system to automatically trigger the fire alarm and ventilation equipment, and the alarm signal is transmitted through the internal network of the building. The local storage adds fire and moisture-proof design, the cloud communication adopts a low-power design, the main battery adopts a replaceable design, the supercapacitor group has a moderate capacity, and the continuous operation time is greater than 10 minutes in the event of a power outage. The main power charging function is added to ensure a stable power supply. Specifically, in urban underground space scenarios, such as underground parking lots, subway tunnels and other crowded places, the device is installed in the ventilation duct, and the existing ventilation system is used for gas collection. The air inlet design considers compatibility with the ventilation system. The signal processing circuit optimizes the shielding design to improve the anti-interference ability, and the digital filtering algorithm is added to improve the signal accuracy. The alarm trigger is linked with the building fire protection system and ventilation system to automatically trigger the fire alarm and ventilation equipment, and the alarm signal is transmitted through the internal network of the building. The local storage has added fireproof and moisture-proof design, the cloud communication adopts a low-power design, the main battery adopts a replaceable design, the supercapacitor group has a moderate capacity to ensure that the continuous operation time is greater than 10 minutes in the event of a power outage, and the AC charging function is added to ensure a stable power supply.

[0103] In the agricultural farm scenario, such as in pig houses, chicken coops and other places where there are harmful gases such as ammonia and hydrogen sulfide, the device is installed near the ventilation opening, and natural ventilation is used for gas collection. The design of the air inlet hole takes into account the requirements of dust prevention and insect prevention. The signal processing circuit optimizes stability and reliability, and increases moisture-proof and dust-proof measures to extend the service life of the equipment. The alarm trigger is linked with the ventilation equipment and spray disinfection system in the farm, automatically starting the ventilation and disinfection equipment to improve the breeding environment. The alarm signal is transmitted wirelessly for convenient remote monitoring. The local storage increases moisture-proof and shock-proof design, the cloud communication uses an agricultural-specific communication network, the main battery adopts a solar charging design to reduce the dependence on mains electricity, the super capacitor bank has a moderate capacity, ensuring that the continuous operation time is ≥ 10 minutes in case of power failure, and a battery power monitoring function is added to facilitate farmers to replace the battery in time. Specifically, in the agricultural farm scenario, such as in pig houses, chicken coops and other places where there are harmful gases such as ammonia and hydrogen sulfide, the device is installed near the ventilation opening, and natural ventilation is used for gas collection. The design of the air inlet hole takes into account the requirements of dust prevention and insect prevention. The signal processing circuit optimizes stability and reliability, and increases moisture-proof and dust-proof measures to extend the service life of the equipment. The alarm trigger is linked with the ventilation equipment and spray disinfection system in the farm, automatically starting the ventilation and disinfection equipment to improve the breeding environment. The alarm signal is transmitted wirelessly for convenient remote monitoring. The local storage increases moisture-proof and shock-proof design, the cloud communication uses an agricultural-specific communication network, the main battery adopts a solar charging design to reduce the dependence on mains electricity, the super capacitor bank has a moderate capacity, ensuring that the continuous operation time is greater than 10 minutes in case of power failure, and a battery power monitoring function is added to facilitate farmers to replace the battery in time.

[0104] In medical site scenarios, such as operating rooms, ICU wards and other places with extremely high air quality requirements, the device is installed inside the ward wall, and the gas to be measured is introduced through a dedicated pipeline. The design of the air inlet hole takes into account the aseptic requirements, and aseptic materials and filtering devices are used. The signal processing circuit optimizes the resolution and accuracy, and adds multi-stage filtering and calibration functions to ensure the high accuracy of the data. The alarm trigger is linked with the hospital's central monitoring system to automatically trigger the alarm and notify the medical staff, and the alarm signal is transmitted through the hospital's internal network. The local storage adds an encryption function to protect the privacy of patients. The cloud communication uses a medical dedicated communication network to ensure the security and stability of data transmission. The main battery uses a high-capacity medical battery, and the capacity of the supercapacitor bank is increased to ensure that the continuous operation time is greater than 15 minutes in the case of power failure. The battery power monitoring and alarm function is added to facilitate the medical staff to replace the battery in time. Specifically, in medical site scenarios, such as operating rooms, ICU wards and other places with extremely high air quality requirements, the device is installed inside the ward wall, and the gas to be measured is introduced through a dedicated pipeline. The design of the air inlet hole takes into account the aseptic requirements, and aseptic materials and filtering devices are used. The signal processing circuit optimizes the resolution and accuracy, and adds multi-stage filtering and calibration functions to ensure the high accuracy of the data. The alarm trigger is linked with the hospital's central monitoring system to automatically trigger the alarm and notify the medical staff, and the alarm signal is transmitted through the hospital's internal network. The local storage adds an encryption function to protect the privacy of patients. The cloud communication uses a medical dedicated communication network to ensure the security and stability of data transmission. The main battery uses a high-capacity medical battery, and the capacity of the supercapacitor bank is increased to ensure that the continuous operation time is greater than 15 minutes in the case of power failure. The battery power monitoring and alarm function is added to facilitate the medical staff to replace the battery in time.

[0105] The present invention and its implementation manners are schematically described above. The description is not restrictive. Without departing from the spirit or basic features of the present invention, the present invention can be implemented in other specific forms. What is shown in the drawings is only one of the implementation manners of the present invention, and the actual structure is not limited thereto. Any reference numeral in the claims should not limit the claimed claim. Therefore, if those of ordinary skill in the art are inspired by it and, without departing from the purpose of this creation, design similar structural manners and embodiments to this technical solution without creative efforts, they shall fall within the protection scope of this patent. In addition, the word "including" does not exclude other elements or steps, and the word "a" before an element does not exclude including "a plurality" of such elements. The multiple elements stated in the product claims can also be implemented by one element through software or hardware. The words "first", "second", etc. are used to denote names and do not denote any particular order.

Claims

1. A method for detecting asphyxiating gases in deep pit operations, characterized in that: The following steps are involved: Build a detection model through standard gas calibration, multi-dimensional environmental parameter compensation, multi-modal sensor integration and machine learning modeling; Build a real-time response system through power-on self-calibration and dynamic threshold setting; Collect and process multimodal data, use spiral guide air chamber and turbulent fan to optimize gas collection, combine wavelet denoising and LSTM dynamic compensation to improve detection accuracy and obtain detection results.

2. The method for detecting asphyxiating gases in deep pit operations according to claim 1, characterized in that: The steps of building a detection model include: obtaining initial concentration data and removing noise and outliers; using a two-stage calibration method to compare and analyze the initial concentration data with the standard gas concentration, adjusting the output characteristics of the sensor so that the measured value of the detection device matches the true value, and establishing a sensor benchmark model; constructing the interdigital electrodes of the carbon dioxide electrochemical sensor, setting a multidimensional parameter space, and constructing a machine learning model for carbon dioxide concentration detection.

3. The method for detecting asphyxiating gas in deep pit operations according to claim 2, characterized in that: Multidimensional parameters include carbon dioxide, temperature, humidity and oxygen; carbon dioxide concentration value is 1000 ppm -10000ppm; temperature value is 10℃-50℃; humidity value is 20%RH -90%RH; oxygen concentration value is 100 ppm -1000ppm.

4. The method for detecting asphyxiating gases in deep pit operations according to claim 2, characterized in that: The interdigital electrodes of the carbon dioxide electrochemical sensor are Composite material construction.

5. The method for detecting asphyxiating gas in deep pit operations according to claim 2, characterized in that: In the process of building the detection model, the oxygen sensor, temperature sensor, humidity sensor and carbon dioxide probe are packaged together, and the synchronous data collection is realized through the SPI bus.

6. The method for detecting asphyxiating gases in deep pit operations according to claim 1, characterized in that: LSTM dynamic compensation includes temperature compensation and humidity compensation. Temperature compensation automatically corrects the concentration value according to the temperature variation coefficient, and humidity compensation adjusts the data based on the humidity variation coefficient.

7. The method for detecting asphyxiating gases in deep pit operations according to claim 6, characterized in that: The standard for temperature compensation is: for every 1°C increase in temperature, the compensation value increases by 0.3%; the standard for humidity compensation is: for every 10% increase in humidity, the compensation value decreases by 0.015%.

8. A deep pit operation asphyxiating gas detection device, characterized in that: A method for detecting asphyxiating gases in deep pit operations according to any one of claims 1 to 7, comprising a housing, a main controller, a sensor unit, a power supply, a fan, an alarm control unit and an interaction unit; the main controller is electrically connected to the sensor unit, the power supply, the fan and the interaction unit respectively to realize control of each component and data processing; The main controller includes a display control module, a button control module, an alarm control module and a storage module; the interactive unit is electrically connected to the display control module, and the power supply is electrically connected to the button control module.

9. The asphyxiating gas detection device for deep pit operations according to claim 8, characterized in that: The shell has a built-in centrifugal fan, and the front and rear side walls of the shell are provided with honeycomb-shaped air inlet and outlet holes.

Images