Construction safety monitoring and early warning management box and method

Abstract

The application discloses a construction safety monitoring and early warning management box and method, and relates to the technical field of monitoring and early warning equipment.The management box comprises a box body, a detachable box cover, a supporting mechanism, a monitoring mechanism, an early warning indication mechanism and a pullable electric control module, realizes integrated integration of environment monitoring, early warning and communication, and the early warning indication mechanism can be lifted and stored, so that maintenance is convenient.The method comprises the following steps: obtaining historical data of a gas sensor to calculate a health index;calibrating the spatial distance of a leakage source through a camera;in combination with wind speed data, the real concentration of the leakage source is inversed;the inversed concentration is corrected by taking the health index as a confidence weight;and according to the difference between the corrected concentration and the health index, a differential alarm strategy is output.The application fuses the health degree of the sensor and the positioning compensation depth of the leakage source, avoids false positives and false negatives caused by sensor drift and response delay, improves the early warning accuracy, realizes the change from passive maintenance to predictive maintenance, and effectively guarantees the safety of tunnel construction.

Description

Technical Field

[0001] This invention relates to the field of monitoring and early warning equipment technology, specifically to a construction safety monitoring and early warning management box and a construction safety monitoring and early warning management method. Background Technology

[0002] Tunnel construction environments are complex, with poor ventilation, a tendency for toxic and harmful gases (such as CH4, CO, and H2S) to accumulate, and high dust levels, high humidity, and drastic temperature and humidity fluctuations, posing serious threats to the safety of construction workers. To ensure tunnel construction safety, various tunnel safety monitoring and early warning devices have emerged. These devices, such as gas sensors and cameras, are deployed inside the tunnel to collect environmental data in real time. When the monitored data exceeds a preset threshold, an audible and visual alarm is triggered to alert on-site personnel to take emergency measures.

[0003] However, existing tunnel safety monitoring and early warning equipment suffers from fragmented functions and low integration. Monitoring modules, early warning modules, communication modules, and control modules are often scattered, resulting in a large number of devices on-site, complex wiring, low deployment efficiency, and difficulty in unified management. Furthermore, control units or electrical control modules are typically fixed inside a cabinet, requiring complete disassembly of the cabinet or disconnection of external wiring during maintenance, which is cumbersome and impacts construction progress and equipment maintenance efficiency.

[0004] In terms of monitoring and early warning methods, the problem of "hidden failure" of sensors is prominent. After long-term operation, gas sensors in tunnels are generally affected by high dust and high humidity environments, resulting in performance degradation such as zero-point drift, response delay, and decreased sensitivity. However, this degradation is "hidden"—the sensor still outputs a signal, but the measurement data is distorted. Existing technologies only assess sensor health through periodic manual calibration, which cannot achieve real-time online diagnosis. When a sensor drifts, two dangerous situations may occur: "the environment exceeds the standard but the sensor reading does not" (false alarm) or "the environment is normal but the sensor reading exceeds the standard" (false alarm). At the same time, leak source location compensation algorithms ignore the reliability of the sensor itself. Some existing technologies attempt to use cameras to calibrate the spatial distance between the leak source and the sensor, and combine this with a gas diffusion model to invert the actual concentration of the leak source in order to more accurately assess the leak risk. However, these algorithms all implicitly assume that "the sensor measurement value is accurate" and do not take the sensor health into account. When the sensor drifts or has a response delay, the inverted leak source concentration deviates significantly from the actual value, leading to inaccurate risk assessment. Furthermore, sensor health status is disconnected from risk assessment. In existing technologies, sensor maintenance decisions and safety warning decisions are independent: maintenance personnel replace sensors according to fixed cycles, and the warning system assesses risk based on sensor readings; the two lack a linkage mechanism. On the one hand, a healthy sensor may continue to be used before its maintenance cycle is due (even though it has failed); on the other hand, a failed sensor may cause the warning system to draw incorrect conclusions when assessing risk.

[0005] Based on this, the present invention proposes a construction safety monitoring and early warning management box and method. Summary of the Invention

[0006] To solve the above-mentioned technical problems, the present invention provides a construction safety monitoring and early warning management box.

[0007] The technical solution adopted in this invention is as follows:

[0008] A construction safety monitoring and early warning management box, comprising:

[0009] The box body has a top cover;

[0010] The lid is detachably connected to the box body;

[0011] The support mechanism is installed inside the box, and its top is fixedly connected to the top cover.

[0012] The monitoring unit, installed inside the cover and on the support structure, is used to acquire tunnel environmental data;

[0013] The warning indicator mechanism is installed on the support mechanism and the top cover, and the support mechanism extends the warning indicator mechanism to the outside of the box.

[0014] The electrical control module is removable and located on one side of the enclosure. The monitoring mechanism and the early warning indicator mechanism are electrically connected to the electrical control module.

[0015] Furthermore, the monitoring system includes gas sensors, dust concentration sensors, temperature and humidity sensors, and wind speed and direction sensors, which are used to acquire gas concentration data, dust concentration data, temperature data, humidity data, and wind speed and direction data in the tunnel, respectively.

[0016] The box cover includes a cover body and a cover plate. The cover plate is detachably installed on the cover body. The cover plate has grid holes. The gas sensor, dust concentration sensor, and temperature and humidity sensor are installed in the cover body.

[0017] Furthermore, the support mechanism includes: support tube one, two support tubes one are symmetrically and fixedly connected to the inner wall of the box; support tube two, one end of support tube two is slidably connected inside support tube one, and the other end is fixedly connected to the top cover; support frame one, installed on one of the support tube two; support frame two, installed between support tube one and support tube two, and wind speed and direction sensors are installed on support frame two; electric push rod, installed inside support tube one, the output end of electric push rod is fixedly connected to the bottom of support tube two, and electric push rod two is electrically connected to the electric control module.

[0018] Furthermore, the early warning and indication mechanism includes an emergency light, an audible and visual alarm, and a camera; wherein, the emergency light is installed on support frame one, the camera is installed at the bottom of the top cover, and the audible and visual alarm is installed on support frame two.

[0019] Furthermore, the cover is equipped with a touch screen and control buttons, which are electrically connected to the electronic control module.

[0020] Furthermore, an explosion-proof telephone is also installed on the enclosure.

[0021] Furthermore, an opening is provided on one side of the enclosure, and the electrical control module includes a pull-out bracket and a controller. The controller is mounted on the pull-out bracket, which is slidably connected to the enclosure through the opening.

[0022] A construction safety monitoring and early warning management method includes the following steps:

[0023] S1: Obtain the real-time readings and historical operating data of the gas sensor, and calculate the health index of the gas sensor based on the historical operating data;

[0024] S2: Acquire tunnel images captured by the camera and determine the spatial distance between the leak source and the gas sensor;

[0025] S3: Obtain wind speed data, and invert the true concentration of the leakage source based on the spatial distance and the wind speed data to obtain the inverted true concentration;

[0026] S4: Use the health index as a confidence weight to correct the retrieved true concentration to obtain the corrected leakage source concentration;

[0027] S5: Output a differentiated alarm strategy based on the corrected leakage source concentration and the health index.

[0028] Further, in S1, the health index of the gas sensor is calculated based on the historical operating data, specifically including:

[0029] S11: Extract the degradation characteristic parameters of the gas sensor from the historical operating data. The degradation characteristic parameters include zero drift rate, response time, and noise level.

[0030] S12: Compare the zero-point drift rate, the response time, and the noise level with the corresponding preset thresholds, and calculate the score value of each feature parameter;

[0031] S13: The scores of each characteristic parameter are weighted and summed to obtain the health index of the gas sensor;

[0032] In S11: the zero-point drift rate is obtained by analyzing the linear regression slope of the daily minimum in the historical operating data; the response time is obtained by analyzing the rise time of the step response to blasting events or shotcreting operations in the historical operating data; and the noise level is obtained by analyzing the standard deviation of readings under stable conditions in the historical operating data.

[0033] Furthermore, in S2, calibrating the spatial distance between the leak source and the gas sensor specifically includes:

[0034] S21: Acquire the sequence of tunnel wall images captured by the camera;

[0035] S22: Identify the leakage source target in the tunnel wall image sequence using an image recognition algorithm. The leakage source target includes tunnel wall cracks, pipe joints, or equipment sealing surfaces.

[0036] S23: Calculate the spatial distance between the leak source target and the gas sensor based on monocular vision calibration technology.

[0037] Furthermore, in S3, the actual concentration of the inverted leakage source specifically includes:

[0038] S31: Acquire wind speed and wind direction data inside the tunnel collected by the wind speed and wind direction sensor;

[0039] S32: Based on the spatial distance, the wind speed data, and the wind direction data, establish a gas diffusion inversion model, which is used to characterize the functional relationship between the sensor-measured concentration and the actual concentration of the leak source;

[0040] S33: Input the real-time reading of the gas sensor and the spatial distance into the gas diffusion inversion model to calculate the inverted true concentration of the leakage source.

[0041] Furthermore, in S4, the health index is used as a confidence weight to correct the retrieved true concentration, specifically including:

[0042] S41: Obtain the health index and the retrieved true concentration;

[0043] S42: Establish a confidence mapping function to convert the health index into confidence weights. The confidence weights range from 0 to 1 and are positively correlated with the health index.

[0044] S43: Correct the retrieved true concentration according to the confidence weight to obtain the corrected leakage source concentration, wherein the corrected leakage source concentration = the retrieved true concentration × the confidence weight.

[0045] Furthermore, in S5, based on the corrected leakage source concentration and the health index, a differentiated alarm strategy is output, specifically including:

[0046] S51: Compare the corrected leakage source concentration with a preset first alarm threshold;

[0047] S52: When the corrected leakage source concentration is greater than or equal to the first alarm threshold, further determine whether the health index is greater than or equal to a preset health threshold:

[0048] If the health index is greater than or equal to the health threshold, it is determined to be a high-confidence alarm, and a first alarm signal is output. The first alarm signal is used to trigger the sound and light alarm and the emergency light to start, and to instruct personnel to evacuate urgently.

[0049] If the health index is less than the health threshold, it is determined to be a medium confidence alarm, and a second alarm signal is output. The second alarm signal is used to trigger the sound and light alarm to start, and at the same time, sensor maintenance prompt information is output.

[0050] S53: When the corrected leakage source concentration is less than the first alarm threshold, further determine whether the health index is less than a preset maintenance threshold:

[0051] If the health index is less than the maintenance threshold, a third alarm signal is output. The third alarm signal is used to output a sensor forced replacement prompt message and does not trigger an environmental alarm.

[0052] If the health index is greater than or equal to the maintenance threshold, a normal status signal is output.

[0053] The beneficial effects of this invention are:

[0054] (1) High integration and modular design: The monitoring mechanism, early warning and indication mechanism, electrical control module and emergency communication module (explosion-proof telephone) are highly integrated into the box. Each module has a clear division of labor and electrical linkage, realizing the integrated management of tunnel environment monitoring, safety early warning, video surveillance and emergency communication, greatly reducing on-site wiring, improving the deployment and debugging efficiency of equipment, and adapting to the complex on-site environment of tunnel construction.

[0055] (2) Convenient inspection and maintenance: The electrical control module adopts a pull-out design. During inspection, the controller can be operated directly by pulling out the pull-out bracket from the box without disassembling the box and external wiring. The cover plate of the box is a detachable structure, which facilitates the calibration, replacement and maintenance of the monitoring mechanism inside the cover, effectively reducing the difficulty of equipment maintenance and reducing the impact on the tunnel construction progress.

[0056] (3) Good protection performance and long service life: The early warning indicator mechanism is installed on the support mechanism and the top cover. When not in operation, it can be retracted into the box with the support mechanism to avoid the risk of damage such as collision of construction equipment and dust erosion. When in operation, it extends to the outside of the box by the support mechanism to ensure the early warning, monitoring and lighting effects, and realizes the balance between protection and functionality.

[0057] (4) The method of this invention calculates the health index of the gas sensor in real time, quantifies the sensor's own degradation state into confidence weights, and combines the spatial distance of the leak source calibrated by the camera with wind speed and direction data to invert the true concentration of the leak source, thus realizing a dual fusion assessment of sensor health and external environmental risk. Based on this, a differentiated alarm strategy is output according to the corrected leak source concentration and health index: emergency evacuation is triggered when the confidence level is high, early warning and maintenance prompts are linked when the confidence level is medium, and sensor replacement is only prompted when the confidence level is low. This method effectively avoids false alarms and missed alarms caused by sensor aging drift and response delay, improves the accuracy of early warning, and realizes the transformation from "passive maintenance" to "predictive maintenance", reducing tunnel construction safety risks and equipment maintenance costs. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of the overall structure of the construction safety monitoring and early warning management box according to an embodiment of the present invention;

[0059] Figure 2 This is a schematic diagram of the working status of a construction safety monitoring and early warning management box according to an embodiment of the present invention;

[0060] Figure 3 This is a schematic diagram of the electronic control module being pulled out of the housing according to an embodiment of the present invention;

[0061] Figure 4 This is a schematic diagram of the internal structure of a construction safety monitoring and early warning management box according to an embodiment of the present invention;

[0062] Figure 5 This is an overall flowchart of the construction safety monitoring and early warning management method according to an embodiment of the present invention.

[0063] Explanation of reference numerals in the attached figures:

[0064] 1-Box body, 11-Top cover, 12-Opening;

[0065] 2-Box lid, 21-Lid body, 22-Lid plate, 23-Grate hole;

[0066] 3- Monitoring agencies;

[0067] 4-Support mechanism, 41-Support pipe one, 42-Support pipe two, 43-Support frame one, 44-Support frame two;

[0068] 5-Early warning indicator, 51-Emergency light, 52-Audible and visual alarm, 53-Camera;

[0069] 6-Electrical control module, 61-Drawer rack, 62-Controller, 63-Air switch, 64-Switch power supply one, 65-Switch power supply two, 66-Switch power supply three;

[0070] 7-Touchscreen display, 8-Control buttons, 9-Explosion-proof telephone. Detailed Implementation

[0071] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0072] like Figures 1-4 As shown, a construction safety monitoring and early warning management box according to an embodiment of the present invention includes a box body 1, a box cover 2, a monitoring mechanism 3, a support mechanism 4, an early warning indication mechanism 5, and an electrical control module 6.

[0073] The enclosure 1 has a top cover 11; the cover 2 is detachably connected to the enclosure 1; the support mechanism 4 is installed inside the enclosure 1, and its top is fixedly connected to the top cover 11; the monitoring mechanism 3 is installed on the support mechanism 4 and the cover 2, and the monitoring mechanism 3 is used to acquire tunnel environmental data; the early warning indication mechanism 5 is installed on the support mechanism 4 and the top cover 11, and when working, the support mechanism 4 extends the early warning indication mechanism 5 to the outside of the enclosure 1; the electrical control module 6 is removable and installed on one side of the enclosure 1, and the monitoring mechanism 3 and the early warning indication mechanism 5 are electrically connected to the electrical control module 6 respectively, and the electrical control module 6 completes data processing and command issuance.

[0074] In this invention, the housing 1 is a sealed metal housing, suitable for the dusty and humid environment inside tunnels. The top cover 11 is slightly larger than the top opening of the housing 1, serving as a dustproof protection. The cover 2 is detachably connected to the front end face of the housing 1 by bolts, facilitating the opening of the cover 2 to inspect the internal components of the housing 1.

[0075] In one embodiment of the present invention, such as Figures 1-4 As shown, monitoring unit 3 includes a gas sensor, a dust concentration sensor, a temperature and humidity sensor, and a wind speed and direction sensor, which are used to acquire gas concentration data, dust concentration data, temperature data, humidity data, and wind speed data in the tunnel, respectively.

[0076] The cover 2 includes a cover body 21 and a cover plate 22. The cover plate 22 is detachably installed on the cover body 21. The cover plate 22 is provided with a grid hole 23. The gas sensor, dust concentration sensor, and temperature and humidity sensor are installed inside the cover body 21. Specifically, the cover plate 22 can be detachably connected to the cover body 21 by a buckle. The detection probes of each monitoring module are all set facing the grid hole 23 to ensure the real-time performance and accuracy of the monitoring data.

[0077] In one embodiment of the present invention, such as Figure 2 and Figure 4 As shown, the support mechanism 4 includes a first support pipe 41, a second support pipe 42, a first support frame 43, and a second support frame 44. The two first support pipes 41 are symmetrically and fixedly connected to the inner wall of the housing 1; one end of the second support pipe 42 is slidably connected inside the first support pipe 41, and the other end is fixedly connected to the top cover 11; the first support frame 43 is installed on one of the second support pipes 42; the second support frame 44 is installed between the first support pipe 41 and the second support pipe 42, and a wind speed and direction sensor is installed on the second support frame 44.

[0078] An electric push rod is installed inside the support tube 41. The output end of the electric push rod is fixedly connected to the support tube 42. The electric push rod is electrically connected to the electric control module 6, which controls its extension and retraction to realize the automatic lifting and lowering of the support tube 42, thereby further improving the automation level of the equipment.

[0079] Specifically, the two support tubes 41 are symmetrically welded to the inner wall of the housing 1 and are rigid metal sleeves; the second support tube 42 is a rigid metal tube that is clearance-fitted with the first support tube 41, with one end slidably inserted into the first support tube 41 and the other end welded and fixed to the lower surface of the top cover 11. The second support tube 42 can slide up and down along the axial direction of the first support tube 41. A sealing ring is provided at the sliding fit to prevent dust from entering; the first support frame 43 is a metal bracket welded to the second support tube 42. The welding position of the first support frame 43 ensures that the emergency light has a good lighting angle after it extends out of the housing 1.

[0080] The gas sensor can detect the concentration of toxic, harmful, and flammable gases such as H2S, CO, SO2, CO2, NO2, NH3, H2, and CH4.

[0081] In one embodiment of the present invention, such as Figure 2 As shown, the warning indication mechanism 5 includes an emergency light 51, an audible and visual alarm 52, and a camera 53; wherein, the emergency light 51 is installed on the first support frame 43, the camera 53 is installed on the bottom of the top cover 11, and the audible and visual alarm 52 is installed on the second support frame 44.

[0082] In this embodiment of the invention, the emergency light 51 can be fixed to the support frame 43 with bolts. It is an explosion-proof emergency lighting fixture suitable for tunnel construction environment. The camera 53 can be installed on the lower surface of the top cover 11 with a bracket. It is a 360-degree panoramic explosion-proof camera that can realize real-time monitoring of the tunnel construction area without blind spots. The audible and visual alarm 52 can be installed on the support frame 44 with a bracket. It is located on the lower surface of the top cover 11 and can emit a high-decibel alarm sound and a red flashing warning signal. The audible and visual alarm 52, the emergency light 51, and the camera 53 are all electrically connected to the electronic control module 6 through wires, and their working status is controlled by the electronic control module 6.

[0083] In one embodiment of the present invention, such as Figure 1 As shown, the cover 21 is equipped with a touch screen display 7 and control buttons 8, which are electrically connected to the electronic control module 6. The touch screen display 7 is an LCD touch screen display, which can display environmental data such as gas concentration, dust concentration, temperature, humidity, and wind speed collected by the monitoring mechanism 3, as well as the equipment operating status in real time. The control buttons 8 include a power button, a power off button, a warning reset button, and a manual warning button. Both the touch screen display 7 and the control buttons 8 are electrically connected to the electronic control module 6 through wires to realize human-machine interaction control on site.

[0084] In one embodiment of the present invention, such as Figure 1 As shown, an explosion-proof telephone 9 is also installed on the enclosure 1. The explosion-proof telephone 9 is a special explosion-proof communication device for tunnel construction. It is connected to an external communication base station to ensure emergency communication between construction personnel inside the tunnel and the outside world, and to achieve real-time connection in emergency situations.

[0085] In one embodiment of the present invention, an opening 12 is provided on one side of the housing 1. The electronic control module 6 includes a pull-out bracket 61 and a controller 62. The controller 62 is mounted on the pull-out bracket 61, and the pull-out bracket 61 is slidably connected to the housing 1 through the opening 12. The monitoring mechanism 3, the touch screen 7, the control button 8, the electric push rod, the emergency light 51, the audible and visual alarm 52, and the camera 53 are electrically connected to the controller 62.

[0086] In a specific embodiment of the present invention, a slide rail is installed at the opening 12; the controller 62 is an embedded industrial computer or a high-performance industrial controller with a built-in image processing unit for performing image recognition and spatial calibration calculations. It can be fixed to the pull-out bracket 61 by bolts. The pull-out bracket 61 is slidably connected to the slide rail at the opening 12, and can be pulled out along the slide rail, so that the controller 62 can extend out of the box 1; a sealing gasket can be set between the pull-out bracket 61 and the opening 12 to prevent dust from entering the box 1.

[0087] In addition, the electronic control module also includes an air switch, a switching power supply 1, a switching power supply 2, and a switching power supply 3. The air switch is used to connect to a 220V power supply. Switching power supplies 1, 2, and 3 are each connected to the air switch. Switching power supply 1 is used to convert the 220V power supply to 24V DC. Switching power supply 2 is used to convert the 220V power supply to 12V DC. Switching power supply 3 is used to convert the 220V power supply to 9V DC. Switching power supply 1 is electrically connected to the controller. Switching power supply 2 is electrically connected to the camera. Switching power supply 3 is used to power the switch.

[0088] Work process:

[0089] In this embodiment, the construction safety monitoring and early warning management box is installed on the side wall or support of the tunnel construction area. When the equipment is turned on, the staff starts the system through the control button 8. The controller 62 of the electrical control module 6 issues an instruction to control the electric push rod to drive the second support pipe 42 to slide upward along the first support pipe 41, thereby driving the top cover 11 and the early warning indicator mechanism 5 to rise, so that the emergency light 51, the sound and light alarm 52, and the camera 53 extend to the outside of the box 1 and enter the working state.

[0090] The monitoring agency 3 collects environmental data such as gas concentration, dust concentration, temperature, humidity, and wind speed in the tunnel in real time through various modules, and transmits the data to the controller 62. The controller 62 analyzes and processes the data and displays the real-time data on the touch screen 7. When the monitoring data exceeds the preset warning threshold, the controller 62 automatically triggers the audible and visual alarm 52 to issue a dual warning signal of sound and light, and at the same time controls the emergency light 51 to turn on and the camera 53 to record video, reminding the on-site construction personnel to take emergency measures.

[0091] On-site personnel can manually control the system via control button 8, such as manually triggering the alarm, resetting the alarm, and shutting down the equipment. They can also communicate with the outside world in an emergency via the explosion-proof telephone 9.

[0092] When equipment maintenance is required, staff can directly pull out the drawer 61 of the electrical control module 6 along the slide rail to debug, set parameters, or repair faults of the controller 62 without disassembling the housing 1 and external wiring. If the monitoring mechanism 3 needs to be calibrated or replaced, the cover plate 22 of the housing cover 2 can be removed from the cover 21 to directly operate the monitoring mechanism 3 inside the cover 21. When the equipment is turned off, the controller 62 controls the electric push rod to control the second support tube 42 to slide down along the first support tube 41, driving the warning indicator mechanism 5 back into the housing 1 for storage, effectively protecting the warning indicator mechanism 5 from damage.

[0093] In another embodiment of the present invention, the bottom of the housing 1 may be provided with an installation structure, including symmetrical mounting plates and strip-shaped mounting holes, to fit the expansion bolts on the tunnel wall, so as to realize the rapid installation and fine-tuning of the equipment and improve the installation adaptability of the equipment;

[0094] In another embodiment of the present invention, the controller 62 may be configured with a wireless communication module to realize the remote transmission of monitoring data and video surveillance images, so that managers can monitor the construction environment and equipment operation status inside the tunnel in real time from the monitoring room outside the tunnel, further enhancing the remote management capability of the system.

[0095] Corresponding to the construction safety monitoring and early warning management box in the above embodiments, the present invention also proposes a construction safety monitoring and early warning management method.

[0096] A construction safety monitoring and early warning management method according to an embodiment of the present invention includes the following steps:

[0097] S1: Acquire real-time readings and historical operating data of the gas sensor, and calculate the health index of the gas sensor based on the historical operating data.

[0098] This step involves a systematic and quantitative analysis of historical operating data from the gas sensor to extract core characteristic parameters that accurately characterize the sensor's aging and degradation. Then, through standardized scoring of single indicators and weighted summation of multiple indicators, a quantitative model of the sensor's health index is established, ultimately yielding a health index that reflects the confidence level of the sensor's measurement data. This step transforms the sensor's operational status into calculable and comparable numerical values, providing crucial confidence criteria for subsequent leak source concentration correction and differentiated alarm strategy output. It also addresses issues such as measurement errors, false alarms, or missed alarms caused by sensor performance degradation in traditional monitoring methods.

[0099] Specifically, the health index of the gas sensor is calculated based on historical operating data, including:

[0100] S11: Extract degradation characteristic parameters of the gas sensor from historical operating data. These parameters include zero-point drift rate, response time, and noise level. The zero-point drift rate is obtained by analyzing the linear regression slope of the daily minimum in historical operating data; the response time is obtained by analyzing the rise time of the step response to blasting events or shotcreting operations in historical operating data; and the noise level is obtained by analyzing the standard deviation of readings under stable conditions in historical operating data.

[0101] To address the issue that gas sensors are prone to performance degradation and decreased measurement accuracy in harsh construction environments such as dusty tunnels with high humidity and large temperature and humidity fluctuations due to component aging, dust adhesion, and environmental corrosion, this invention extracts core degradation characteristic parameters from the sensor's historical operating data. The historical operating data used must be continuous, uninterrupted, and have outlier removed, with a statistical period of ≥30 days to ensure statistical significance. Finally, the zero-point drift rate is extracted. Response time Noise level Three core parameters comprehensively characterize the sensor's performance from three dimensions: static baseline, dynamic response, and measurement stability. All three are classic indicators that can be quantified by data and are strongly correlated with measurement accuracy. Among them, the zero-point drift rate... The response time is the quantization rate at which the measured value slowly deviates from the initial zero point over time when there is no target gas input to the sensor. It reflects the degree of systematic error of the sensor's static baseline and can be obtained by performing linear regression on the minimum value sequence of measurements taken during daily periods without gas leaks or construction work. This refers to the sensor's response speed to sudden changes in gas concentration caused by tunnel blasting and shotcreting operations. It reflects the degree of hysteresis in the sensor's dynamic performance and can be determined by capturing the time difference between the sensor's step response and 90% of its peak value; noise level. The degree of random fluctuation in the sensor's measured values ​​around the true value under stable conditions with no changes in gas concentration, no construction work, and constant temperature and humidity reflects the degree of random error in the sensor's measurement. This can be quantified by the standard deviation of the measured values ​​under stable conditions. It should be noted that the extraction process of all three parameters requires rigorous data acquisition and preprocessing. Through operations such as outlier removal and static baseline calibration, it is ensured that the extracted parameters accurately reflect the normal aging and degradation patterns of the sensor. All data processing and calculations are performed by an industrial control computer.

[0102] S12: Compare the zero-point drift rate, response time, and noise level with the corresponding preset thresholds, and calculate the score value of each feature parameter.

[0103] Due to zero-point drift rate Response time Noise level The three degradation characteristic parameters have different dimensions and significantly different numerical ranges, making direct comprehensive comparison and analysis impossible. Therefore, this invention uses an inverse linear normalization model to unify the three parameters into a range of values ​​within a certain range. The scores are the zero-point drift rate scores. Response time score Noise level score This achieves dimensional unification for parameters with different dimensions. All three degradation characteristic parameters are deterioration parameters, meaning that a larger parameter value indicates worse sensor performance. Therefore, inverse linear normalization follows the principle that "the closer the parameter value is to 0, the closer the score is to 100; when the parameter value reaches a preset threshold, the score is 0; when the parameter value exceeds a preset threshold, the score is forced to 0." The preset thresholds for each parameter include the maximum zero-point drift rate threshold. Maximum response time threshold Maximum threshold for noise level The threshold settings for this type of measurement combine the maximum permissible error parameters calibrated by the sensor manufacturer with adaptability adjustments for the tunnel construction environment. These thresholds represent the maximum limits at which the sensor can function normally in a tunnel construction environment; exceeding these thresholds will cause the sensor's measurement error to exceed the permissible range for tunnel construction safety monitoring. The calculation formulas for the scoring values ​​corresponding to each characteristic parameter are as follows:

[0104] (when ,otherwise );

[0105] (when ,otherwise );

[0106] (when ,otherwise );

[0107] The zero-point drift rate score Calculation The absolute value is used because the direction of the zero-point drift (rising or falling) does not affect the performance judgment; only the drift rate is the core criterion. If the score of any parameter is 0, it means that the sensor has completely failed in that performance dimension. At this time, the sensor can be directly determined to be in a state of overall failure, and a sensor replacement prompt will be output simultaneously. There is no need to continue the calculation of the subsequent health index, thus avoiding invalid concentration inversion and correction operations.

[0108] S13: The weighted sum of the scores of each characteristic parameter is used to obtain the health index of the gas sensor.

[0109] This step uses a weighted linear summation method to... , , The three standardized score values ​​are combined to obtain a health index that reflects the overall health status of the sensor. The core of this approach is to assign corresponding weight coefficients to each score value based on the varying degrees of influence of three degradation characteristic parameters on the sensor's measurement accuracy. This reflects the weighted impact of different characteristic parameters on the sensor's health status. The weight coefficients for each score value are, respectively, the zero-point drift rate weight. Response time weight Noise level weight The magnitude of the weight coefficients directly represents the degree of influence of the corresponding feature parameters on the sensor's health status, and all weight coefficients must satisfy the normalization condition, i.e. Meanwhile, the values ​​of each weighting coefficient are all within Within the range. Health Index The calculation formula is:

[0110] ;

[0111] Calculated using this formula The range of values ​​is , is a dimensionless number. A higher value indicates a better overall health status of the sensor, and a higher confidence level in its measurement data. The weighting coefficients are not fixed and need to be flexibly adjusted based on the actual monitoring needs of tunnel construction. For example, in general tunnel monitoring scenarios, the zero-point drift rate, as a systematic error, has the greatest impact on measurement accuracy. Response time is crucial for timely early warning of sudden gas leaks. The noise level is the random error that can be partially eliminated by data filtering. During tunnel blasting operations, frequent sudden changes in gas concentration necessitate a higher weighting for response time. Conversely, during long-term tunnel surrounding rock monitoring, where continuous and stable monitoring is paramount, a higher weighting for noise levels is appropriate. To facilitate quick and intuitive sensor status assessment by on-site personnel, this invention can display a health index. Divided into four practical judgment intervals, The sensor is in excellent condition, the measurement data has high confidence, and it can work normally without maintenance. The sensor is in good condition, and the measurement data has a medium to high confidence level. It needs to be calibrated regularly and checked weekly. If the sensor is in a state of medium confidence, it needs to be calibrated immediately and vulnerable parts replaced. The sensor is in poor condition, resulting in low-confidence measurement data. Maintenance is required, and sensor replacement is recommended. Furthermore, considering that sensor health status changes dynamically over time, the health index... A sliding window method can be used for real-time updates. For example, a basic statistical window of 30 days can be used. For each new day of sensor operation data, the industrial control computer will remove the earliest historical data and recalculate each degradation characteristic parameter and health index to ensure that the health index can always reflect the latest working status of the sensor and avoid the problem of sensor status judgment lag caused by static statistics.

[0112] S2: Acquire tunnel images captured by the camera and determine the spatial distance between the leak source and the gas sensor.

[0113] This step relies on the 360-degree panoramic explosion-proof camera configured in the construction safety monitoring and early warning management box. Through a continuous automated process of image sequence acquisition, leak source target identification, and monocular visual spatial calibration, the leak source target in the two-dimensional image of the tunnel wall is converted into the three-dimensional spatial straight-line distance between it and the gas sensor. This parameter is the core spatial input of the gas diffusion inversion model, and its measurement accuracy determines the inversion accuracy of the true concentration of the leak source. This invention uses monocular vision calibration technology to achieve distance measurement. Unlike the dual-camera distance measurement mode of binocular vision, monocular vision completes three-dimensional distance measurement using only a single camera. This not only fits the compact hardware structure of the management box but also avoids the problem of distance measurement failure caused by dust obscuring a single lens in binocular cameras. The key lies in utilizing the engineering characteristic of the tunnel wall as a rigid plane, with the industrial control computer pre-calibrating the axial distance from each position on the tunnel wall to the gas sensor. This addresses the issue of scale uncertainty in monocular visual ranging. The 360-degree panoramic explosion-proof camera used in this invention must meet the requirements of resolution ≥1080P and frame rate ≥25fps. The camera is installed at the bottom of the top cover of the management box, with the lens facing the tunnel wall. Under normal operating conditions, the camera and gas sensor must be installed at the same height to ensure the initial fit between the visual coordinate system and the sensor's world coordinate system. If the heights are not aligned, the industrial control computer can pre-calibrate the vertical height difference between the two and introduce compensation parameters for correction calculations, eliminating spatial coordinate conversion errors caused by height deviations. Furthermore, considering the characteristics of the tunnel construction environment—high dust levels, dim lighting, and a monotonous tunnel wall background—all image data acquired by the camera must be pre-processed in real-time by the industrial control computer. This involves image dehazing, Gaussian filtering for noise reduction, and histogram equalization to enhance contrast, effectively eliminating environmental interference and ensuring that the features of the leak source target can be accurately captured by subsequent identification algorithms. All spatial coordinate calculations in this step are performed by the industrial control computer based on the mutual transformation between the pixel coordinate system, camera coordinate system, and world coordinate system. The world coordinate system is the core coordinate system, with the installation center of the gas sensor as its origin. , The axis points horizontally towards the tunnel wall along the tunnel's radial direction. The axis runs along the height of the tunnel. The axis, along the tunnel axis, is used to characterize the position of the leak source target in actual space; the transformation between the camera coordinate system and the pixel coordinate system is achieved by a pre-calibrated camera intrinsic parameter matrix. To achieve this, the matrix is ​​a fixed value, calibrated, and stored in the industrial computer's storage module. Its expression is: ,in , For the camera in pixel coordinate system , The effective focal length of the axis, measured in pixels. , These are the pixel coordinates of the camera's principal point, expressed in pixels, providing a basis for subsequent spatial coordinate transformations.

[0114] Specifically, calibrating the spatial distance between the leak source and the gas sensor includes:

[0115] S21: Obtain the sequence of tunnel wall images captured by the camera.

[0116] As the data acquisition stage for distance calibration, step S21 continuously acquires a sequence of images of the tunnel walls using a camera, replacing the single-frame image acquisition method. This solves the problem of missing leak source targets or blurred images caused by momentary dust obstruction or sudden changes in light in single-frame images, ensuring the robustness of subsequent target identification. The acquisition of the image sequence is synchronously triggered with the concentration monitoring of the gas sensor. When the gas sensor detects an abnormal gas concentration in the tunnel that exceeds a preset low threshold, the system will automatically detect the leak. At that time, the industrial control computer will automatically control the camera to start the continuous data acquisition process, in which... Below the first alarm threshold in the subsequent differentiated alarm strategy This signal serves only as a trigger for visual ranging, avoiding invalid image acquisition operations. To balance the device's computational efficiency and target recognition accuracy, acquisition parameters adapted to tunnel construction can be set. The camera continuously acquires data for 35 seconds at a time, generating a sequence of 75,125 images. Simultaneously, the panoramic images acquired by the camera are cropped, retaining only the tunnel wall area within a 5-meter detection radius of the gas sensor, significantly reducing invalid background data and improving the running speed of subsequent algorithms. Subsequently, the industrial control computer performs frame quality screening on the acquired image sequence, calculating the blur of each image using the Laplace variance method, discarding low-quality frames with blur ≥0.5, and retaining only clear frames to form a valid image sequence, which is stored in the local cache of the industrial control computer. If the concentration detected by the gas sensor returns to normal, the camera will immediately stop continuous acquisition and return to a low-power standby state, retaining only the mode of capturing one frame every 10 seconds, effectively saving equipment energy while ensuring monitoring needs are met.

[0117] S22: Identify leakage source targets in tunnel wall image sequences using image recognition algorithms. Leakage source targets include tunnel wall cracks, pipe joints, or equipment sealing surfaces.

[0118] This invention utilizes an improved machine vision target detection algorithm to automatically identify leakage source targets such as tunnel wall cracks, pipe joints, and equipment sealing surfaces from preprocessed effective image sequences, and accurately extracts the center pixel coordinates of these targets. These leakage source targets are all high-frequency locations of gas leaks during tunnel construction, and each possesses unique visual characteristics. Tunnel wall cracks appear as linear or branching gray-scale abrupt changes, pipe joints exhibit circular or rectangular outlines, and equipment sealing surfaces show square or annular outlines, exhibiting significant visual differences from the tunnel wall background, providing clear feature basis for algorithm recognition. The industrial control computer is equipped with a lightweight improved YOLOv5 algorithm combined with morphological processing, balancing recognition speed and accuracy while avoiding computational delays caused by complex algorithms. During algorithm execution, the industrial control computer first extracts the visual features of the leak source target in the image through the YOLOv5 backbone network, effectively eliminating interfering features such as tunnel wall scratches and stains; then, it clusters the extracted effective features to generate target detection boxes and outputs the initial pixel coordinates of the targets; subsequently, it performs dilation-erosion morphological processing on the region within the detection box to eliminate the target edge blurring caused by dust adhesion and correct the position of the detection box; and finally, it calculates the center pixel coordinates of the target based on the corrected detection box, using the following formula:

[0119] , ;

[0120] in , To detect the pixel coordinates of the left and right boundaries of the bounding box, , The pixel coordinates of the upper and lower boundaries of the detection box are determined; finally, mean filtering is applied to the target center coordinates of all frames in the valid image sequence to obtain the final pixel coordinates of the leakage source center. To eliminate pixel errors in single-frame images and improve the accuracy of coordinate extraction, a dual exclusion mechanism can be set to avoid target misidentification caused by interference from the complex tunnel environment. First, the identified target features are matched with the leak source target feature library pre-stored in the electronic control module. Only when the matching degree is ≥80% is it determined to be a valid leak source. Second, if a target is identified but the concentration detected by the gas sensor returns to normal within 5 seconds, it is directly determined to be a false target and the identification result is discarded, ensuring that the target identified by the algorithm is a real gas leak source.

[0121] S23: Calculate the spatial distance between the leak source target and the gas sensor based on monocular vision calibration technology.

[0122] This step, through camera intrinsic parameter calibration and spatial constraints of the world coordinate system, extracts the two-dimensional pixel coordinates of the leakage source. Convert to 3D world coordinates Then, the actual spatial distance between the leak source and the gas sensor is calculated using the formula for the straight-line distance between two points in space. The core of this step lies in utilizing the characteristic that the tunnel wall is a rigid plane to pre-determine the axial distance. A mapping database is established to address the scale uncertainty issue in monocular vision ranging. Before calculation, the axial distances within the tunnel wall participating in the camera's operation need to be determined. The initial calibration work involved using the Zhang Zhengyou calibration method, a common practice in the machine vision industry, to measure camera intrinsic parameters. A checkerboard calibration board was fixed to the tunnel wall. The board was moved within the camera's field of view, and 15-20 images of the board at different angles and positions were captured. The physical focal length and pixel size of the camera were calculated using a calibration algorithm and converted into the effective focal length. , Simultaneously determine the principal pixel coordinates of the camera. , Finally, the intrinsic parameter matrix is ​​generated. The calibrated reprojection error is ≤0.5px to ensure the accuracy of subsequent coordinate transformations; axial distance of the tunnel wall The calibration was performed using a laser rangefinder with an accuracy of ±1mm. Using the installation center of the gas sensor as the origin, calibration points were marked every 1m along the tunnel axis and every 0.5m along the tunnel height. The distance from each calibration point to the sensor was measured using the laser rangefinder. The value is obtained by simultaneously collecting the pixel coordinates of each point through the camera and establishing the pixel coordinates. axial distance The mapping database is then used to perform interpolation and fitting on the database to generate the entire field of view of the tunnel wall. The interpolation model ensures that the corresponding pixel coordinates within the camera's field of view can be queried. If the installation location of the management box is changed, it needs to be readjusted. The calibration is performed and the database is updated. The core coordinate transformation is based on the homogeneous coordinate formula of the perspective projection model, which is:

[0123] ;

[0124] in As a scaling factor, because For pre-calibrated known values, they can be directly derived from the formula. Then, the scale factor is eliminated and the radial coordinates of the tunnel are obtained. with height coordinates The specific calculation formula is as follows: , Thus, the three-dimensional spatial coordinates of the leakage source target in the world coordinate system can be obtained. .

[0125] Based on the world coordinate system origin being the gas sensor installation center, the spatial straight-line distance between the leak source target and the gas sensor is... It can be directly calculated using the formula for the straight-line distance between two points in three-dimensional space. The formula is: The calculation results are rounded to two decimal places and the unit is meters, which fully meets the accuracy requirements of subsequent gas diffusion inversion. For tunnels with curved cross-sections, directly solving the problem using the calculation method for planar tunnels would be ineffective. This will result in a measurement error of ≤5%, therefore a curvature correction is required for the calculation results. The correction formula is as follows: ,in This is the corrected actual spatial distance. The curvature correction factor is 1 for curved sections and 0 for planar sections. The radius of curvature of the tunnel cross section can be measured in advance and then entered into the formula for correction.

[0126] In this invention, all operations in step S2 are uniformly controlled by the industrial control computer in the management box, achieving fully automatic linkage with the gas sensors of the monitoring mechanism and the camera of the early warning indication mechanism, requiring no manual operation and forming a complete and efficient automated execution logic. When the gas sensor detects an abnormal gas concentration in the tunnel and Upon receiving the trigger signal, the industrial control computer will immediately send a trigger signal to initiate the entire ranging process. After receiving the trigger signal, the industrial control computer will immediately control the camera to begin continuous image acquisition, while simultaneously performing real-time preprocessing and frame quality filtering on the acquired images to generate a valid image sequence. Subsequently, the industrial control computer will call its built-in improved YOLOv5 algorithm to identify the leakage source target in the valid image sequence, extracting and optimizing the target's center pixel coordinates. Next, the industrial control computer retrieves the pre-stored camera intrinsic parameter matrix. and The mapping database uses a perspective projection model to transform two-dimensional pixel coordinates into three-dimensional world coordinates, calculating the world coordinates of the leak source and its spatial distance from the gas sensor. The calculated results will then be used to... The values ​​are transmitted in real time to the gas diffusion inversion model in step S3, serving as the core input parameters for that model. Finally, the industrial control computer will perform subsequent operations based on the concentration status of the gas sensor. If the concentration returns to normal, the camera will immediately return to a low-power standby state. If the concentration remains abnormal and reaches the alarm threshold, the camera will continue to acquire images and update in real time. The value provides real-time and accurate spatial data for subsequent concentration inversion and differential alarm. This linkage logic ensures the seamless integration of step S2 with the entire construction safety monitoring and warning system. All operations are automatically completed in the background of the industrial control computer. On-site workers only need to view the location and distance information of the leakage source intuitively through the touch display screen of the management box, achieving the goals of automation, intelligence, and unmanned operation for tunnel construction gas leakage monitoring.

[0127] S3: Obtain wind speed data, and invert the true concentration of the leakage source based on the spatial distance and wind speed data to obtain the inverted true concentration.

[0128] This step is executed by the explosion-proof industrial control computer in the electric control module. Its core is to establish a modified exponential decay gas diffusion inversion model by abandoning the traditional free-space diffusion model according to the gas diffusion characteristics of the long and narrow confined space in the tunnel. The industrial control computer realizes the integrated operation of multi-source data acquisition, preprocessing, mathematical operation, data storage, and linkage control. The industrial control computer establishes a unique storage mapping relationship for the original acquisition parameters, pre-stored calibration parameters, intermediate operation parameters, and final result parameters involved in this step, and completes the unit normalization process. The core associated parameters include the real-time measured concentration of the gas sensor , the real-time wind speed in the tunnel , the included angle between the wind direction and the diffusion direction , the spatial distance between the leakage source and the sensor , the gas diffusion attenuation coefficient in the tunnel , the adsorption correction coefficient of the tunnel wall , and finally, the true concentration of the leakage source is obtained by inverting and solving through the model . [[ID=​​The data obtained from step S2 is directly retrieved from the industrial control computer's own storage module, eliminating external transmission errors and achieving complete synchronization with the timestamps of other acquired data. Data acquisition employs a "fixed-frequency acquisition and triggered supplementary acquisition" mode. Under normal conditions, the acquisition frequency is 10Hz, consistent with the calculation frequency in steps S1 and S2. When the concentration value acquired by the gas sensor reaches a preset low threshold, the industrial control computer automatically increases the acquisition frequency to 50Hz to capture dynamic changes in the leak source concentration. Simultaneously, the industrial control computer performs real-time validity verification during data acquisition, sequentially completing triple checks on numerical range, communication status, and abrupt changes. Invalid data caused by sensor malfunctions, communication interruptions, and environmental interference is eliminated. The verified valid data is then integrated into the input dataset. .

[0130] In tunnel construction environments, sensor-collected data is susceptible to factors such as dust accumulation, airflow disturbances, and electromagnetic interference, resulting in random noise and outliers. Directly inputting these into model calculations can lead to excessively biased inversion results. Therefore, the industrial control computer performs lightweight real-time preprocessing on the input dataset after data acquisition to minimize interference while maintaining processing speed. This preprocessing process only applies to… , , Three sets of parameters were used, including spatial distance. After error correction and accuracy verification in step S2, no further processing is needed. The industrial control computer uses the 10 most recent valid data sets as a statistical window, removes outliers using the 3σ criterion, and then applies a 5-point moving average filter to eliminate random noise from the remaining valid data, finally obtaining a smoothed valid dataset. The industrial control computer stores the data in its local cache as the final input data for the gas diffusion inversion model. At the same time, the industrial control computer retains the correspondence between the original collected data and the preprocessed data and stores it in the storage module to facilitate subsequent troubleshooting and data traceability.

[0131] Specifically, the actual concentration of the leak source is inverted to include:

[0132] S31: Acquire wind speed and wind direction data inside the tunnel collected by the wind speed and direction sensor.

[0133] The industrial control computer collects the raw real-time wind speed inside the tunnel through wind speed and direction sensors. The original angle between the wind direction and the diffusion direction of the leak source-sensor. After the two sets of data are transmitted to the industrial control computer via RS485 bus, invalid data is first removed through validity verification, and then preprocessed by 3σ outlier removal and moving average filtering to obtain smoothed valid wind speed data. And effective angle data These two sets of valid data will serve as core wind field parameters, directly participating in the construction of subsequent gas diffusion inversion models, providing accurate numerical basis for quantitatively characterizing the impact of the wind field on gas diffusion within the tunnel and completing wind field-related corrections.

[0134] S32: Based on spatial distance, wind speed data, and wind direction data, establish a gas diffusion inversion model. The gas diffusion inversion model is used to characterize the functional relationship between the sensor-measured concentration and the actual concentration at the leakage source.

[0135] Step S32 involves establishing a gas diffusion inversion model based on spatial distance, wind speed, and wind direction data. This model is the core model characterizing the functional relationship between the sensor-measured concentration and the actual concentration at the leakage source. It is a modified exponential decay gas diffusion model adapted to the confined space characteristics of tunnels. Specifically, the industrial control computer first establishes the model based on the preprocessed effective angle... Calculate the wind direction correction factor , The coefficient ranges from -1 to 1. A positive value indicates a downwind, which accelerates gas diffusion attenuation; a value of 0 indicates a crosswind, which has no significant impact on gas diffusion; and a negative value indicates a headwind, which slows down gas diffusion attenuation. This achieves a quantitative correction of the impact of wind direction on gas diffusion. Subsequently, the industrial control computer combines the tunnel gas diffusion attenuation coefficient, calibrated on-site and pre-stored in the local storage module, with the data. Tunnel wall adsorption correction coefficient and the effective wind speed after pretreatment The spatial distance obtained in step S2 A positive gas diffusion model for confined spaces in tunnels is constructed, and the model is as follows:

[0136] ;

[0137] in For real-time concentration measurement of pre-treated gas by sensors, The unit is It is related to the tunnel cross-sectional dimensions, the gas molecule diffusion coefficient, and the degree of turbulence within the tunnel. This is a dimensionless number, ranging from (0,1], representing the adsorption of gas by the tunnel wall. The rougher the tunnel wall, the better. The smaller the value, the better.

[0138] S33: Input the real-time readings and spatial distance of the gas sensor into the gas diffusion inversion model to calculate the inversion true concentration of the leak source.

[0139] This step involves the industrial control computer performing a reverse calculation on the gas diffusion forward model to obtain the final inversion result of the true concentration at the leakage source. The industrial control computer performs a reverse calculation on the constructed gas diffusion forward model to obtain the inversion calculation formula for the true concentration at the leakage source:

[0140] ;

[0141] in This is the reciprocal of the tunnel wall adsorption correction coefficient. To reduce the real-time computation load on the industrial control computer, this value can be compared with... Synchronously pre-stored in the local storage module of the industrial control computer, The natural exponent term is calculated quickly by the industrial control computer using a dedicated floating-point arithmetic module.

[0142] When the industrial control computer executes the inversion formula calculation, it can adopt the method of "step-by-step calculation and intermediate value verification", first calculating the wind field correction terms in sequence. Distance-wind field coupling term diffusion attenuation term Then, the natural exponentiation and subsequent multiplication operations are performed to ensure the accuracy of the results at each step. Simultaneously, the industrial control computer performs dual physical law verification on the final inversion result: firstly, it verifies... First, whether it is within the reasonable concentration range of the corresponding monitored gas; second, verification. Is it greater than or equal to? Because gas attenuates during diffusion, the actual concentration at the leak source will inevitably be no lower than the concentration measured by the sensor, after passing dual verification. The actual concentration of the leakage source is then inverted and stored in a dedicated storage module by the industrial control computer as the final calculation result of step S3. A timestamp is added to this result to ensure time synchronization with all previous data.

[0143] The industrial control computer employs a dual-layer storage architecture—a local storage module and an external SD card—for all types of data generated throughout the S3 process. This architecture enables layered, categorized, and persistent data storage, balancing high-speed data retrieval for real-time computation with long-term traceability analysis of historical data. The industrial control computer's local storage module includes a temporary cache and a fixed storage area. As a high-speed storage area, it primarily stores the raw acquired data, preprocessed valid data, intermediate computational parameters from each stage, and the final inversion results. Only the most recent 100 data sets are retained to meet the high-speed retrieval requirements of the industrial control computer for real-time calculation, data interaction between preceding and subsequent steps, and real-time display on the touch screen. External storage uses industrial-grade explosion-proof SD cards with a capacity of ≥8GB, which are vibration-resistant, dust-resistant, and data is retained even after power loss. As a persistent storage area, the industrial control computer saves the timestamped S3 full-process data, calculation logs, and fault records in a common CSV format. Under normal monitoring conditions, the storage frequency is 1 minute / time. When the gas concentration reaches the warning threshold, the storage frequency is automatically increased to 1 second / time, recording all data during the alarm period in detail. Data uses a circular storage mode; when the SD card is full, it automatically overwrites the oldest historical data. Data can also be permanently saved by manually replacing the SD card. Simultaneously, the industrial control computer will invert the actual concentration from the leak source in the storage module. The data is transmitted in real time to the touchscreen display, which then displays the data as numerical values ​​plus a curve showing the changes. The dynamic trend of change is displayed simultaneously. , , These key parameters allow on-site staff to intuitively understand changes in the concentration of the leak source.

[0144] S4: Use the health index as a confidence weight to correct the inverted true concentration, and obtain the corrected leakage source concentration.

[0145] This step is used to deeply couple the sensor's own health status with the concentration inversion result. The gas sensor health index measured in step S1 is quantified into a confidence weight that can intuitively reflect the reliability of the data. This is used to correct the true concentration of the leak source obtained in step S3, eliminating the concentration inversion deviation caused by problems such as sensor zero drift, response lag, and on-site measurement noise. It also eliminates data errors caused by equipment deterioration, obtaining a high-confidence corrected leak source concentration. This provides an accurate and reliable basis for subsequent differentiated alarm procedures, achieving dual control over monitoring data quality and concentration values, and avoiding false alarms and missed alarms caused by sensor performance degradation.

[0146] Specifically, using the health index as a confidence weight to correct for the retrieved true concentration includes:

[0147] S41: Obtain health index and invert true concentration.

[0148] This step is for acquiring the basic data source for the concentration correction process. All data is directly retrieved through a dedicated storage module inside the industrial control computer, eliminating the need for external sensors or industrial bus transmission, thus preventing errors caused by data transmission delays and electromagnetic interference. The industrial control computer simultaneously retrieves the gas sensor health index calculated in step S1 and the actual leakage source concentration obtained from step S3. The retrieval frequency of these two types of data is completely consistent with the monitoring frequency of the entire system. Under normal monitoring conditions, the frequency is synchronously retrieved at 10Hz, and it automatically increases to 50Hz under alarm and warning conditions, meeting the monitoring requirements for dynamic concentration changes. At the same time, the industrial control computer verifies the millisecond-level timestamps of the two types of data, retaining only synchronous data with a timestamp deviation of no more than 10ms, ensuring that the health index and the retrieved concentration correspond to the same monitoring cycle, and preventing correction deviations caused by data misalignment. After the data retrieval is completed, the industrial control computer immediately performs a dual validity check. On the one hand, it checks that the health index is within a reasonable range of 0-100 and that the retrieved true concentration conforms to the standard range of the corresponding monitored gas. On the other hand, it verifies the completion status of the previous steps. Only after confirming that both types of data are valid calculation results are they temporarily stored in the industrial control computer's dedicated temporary buffer area as the original data for subsequent correction calculations. If invalid data is detected, it will be automatically retrieved again, ensuring the compliance and accuracy of the input data in all aspects.

[0149] S42: Establish a confidence mapping function to convert the health index into confidence weights. The confidence weights range from 0 to 1 and are positively correlated with the health index.

[0150] This step standardizes the conversion of the health index to confidence weights using a linear confidence mapping function. The core mapping relationship is that the confidence weight equals the health index divided by the maximum value of the health index, where the maximum health index is fixed at 100. Through this linear mapping rule, the health index, ranging from 0 to 100, can be normalized into confidence weights within the 0-1 range. The two are strictly positively correlated; that is, the higher the sensor health index, the better the equipment's working condition, the larger the corresponding confidence weight, and the higher the reliability of the inverted concentration. Before the formal calculation, the industrial control computer first performs a 3-point moving average filter on the health index to eliminate minor computational fluctuations caused by previous steps, ensuring stable basic data. Then, it completes the conversion calculation using a 32-bit floating-point division instruction, while simultaneously performing a mandatory range check on the calculation results. If abnormal values ​​exceeding the 0-1 range are found, the confidence weights are automatically clamped to a reasonable range to ensure that the weight values ​​are compliant and effective. When the health index is 100, the sensor is in optimal working condition, the confidence weight is 1, and the inverted concentration does not need to be reduced in weight; when the health index is 0, the sensor is completely ineffective, the confidence weight is 0, and the inverted concentration has no reference value; for health indices between the two, the confidence weight is calculated according to the actual health level of the device.

[0151] S43: Correct the inverted true concentration according to the confidence weight to obtain the corrected leakage source concentration. Corrected leakage source concentration = inverted true concentration × confidence weight.

[0152] This step relies on confidence level weights to perform weighted correction of the inverted concentration. Specifically, the industrial control computer executes the correction calculation via floating-point multiplication instructions, multiplying the previously filtered and calibrated true inverted concentration by the corresponding confidence level weight to obtain the final corrected leakage source concentration. During the calculation, the industrial control computer fixes the floating-point operation precision to 6 decimal places, without truncating or rounding intermediate calculation values, retaining only 2 decimal places for the final corrected concentration. This ensures both calculation accuracy and meets the practical viewing needs of engineering sites. Simultaneously, the industrial control computer performs dual result verification: firstly, it verifies that the corrected concentration is not greater than the true inverted concentration, conforming to the physical law of weight reduction correction; secondly, it verifies the rationality of the data change trend to avoid calculation errors leading to distorted results. After the calculation is completed, the industrial control computer stores the corrected concentration in a dedicated fixed storage module. On the one hand, it pushes the corrected concentration to subsequent differentiated alarm steps in real time, serving as the sole core basis for alarm determination. On the other hand, it works with an external industrial explosion-proof SD card to achieve dual-layer data storage: timed storage in normal conditions and high-frequency storage in alarm conditions, completely retaining the corrected data with timestamps for easy troubleshooting and data traceability. Simultaneously, the industrial control computer will implement reverse linkage with previous steps based on the correction result and confidence weight value. If the confidence weight is too low, it will automatically trigger the sensor health index recalculation and leakage source concentration inversion verification process, forming a closed-loop control for data correction, further ensuring the accuracy and reliability of the entire system's monitoring data.

[0153] S5: Output differentiated alarm strategies based on the corrected leak source concentration and health index.

[0154] This step utilizes a modified mathematical model that nests thresholds for both leak source concentration and health index to achieve a dual quantitative assessment of tunnel construction environmental risks and gas sensor operational status. This results in differentiated alarm signals that directly drive the hardware of the construction safety monitoring and early warning management box to execute corresponding alarm and alert actions. This step transforms the comprehensive quantitative analysis results of preliminary sensor health status assessment, leak source spatial location, concentration inversion, and correction into engineered safety early warning actions. It addresses the technical shortcomings of traditional monitoring and early warning methods, such as a single alarm strategy, failure to differentiate between sensor status and environmental risk, and inaccurate hardware linkage. It achieves tiered control, including high-confidence emergency alarms, medium-confidence early warnings and sensor maintenance, and low-confidence sensor alerts, forming a complete technical closed loop from data quantification to safety action in the entire monitoring and early warning method.

[0155] Specifically, based on the corrected leak source concentration and health index, the differentiated alarm strategies output include:

[0156] S51: Compare the corrected leakage source concentration with the preset first alarm threshold.

[0157] This step is the first level of environmental risk assessment in the differentiated alarm strategy, and it is the foundation of the entire nested threshold determination process. All operations are executed rapidly by the industrial control computer using native commands. The industrial control computer first synchronously retrieves the corrected leakage source concentration calculated in step S4 from the dedicated storage module. and the first alarm threshold pre-stored in the local storage module. ,in Based on tunnel construction safety regulations, the system is matched to the types of toxic, hazardous, and flammable gases monitored, and supports flexible on-site modification via the touchscreen display of the management box without disassembly. Before formal judgment, the industrial control machine will retrieve... Lightweight preprocessing is performed: first, a two-point moving average filter is used to eliminate minute instantaneous fluctuations in the data; then, three consecutive data consistency checks are performed to avoid brief "flash alarms" caused by single data fluctuations. Finally, the filtered corrected concentration is obtained. This serves as the basis for judgment in this step. Subsequently, the industrial control computer executes a real number comparison instruction. The mathematical judgment is as follows: if the judgment result is true, it is determined that there is an environmental risk in the tunnel construction area, and the process proceeds to the subsequent detailed judgment stage of sensor health status; if the judgment result is false, it is determined that there is no environmental risk in the area, and the process proceeds to the detailed judgment stage of sensor maintenance status.

[0158] S52: When the corrected leakage source concentration is greater than or equal to the first alarm threshold, further determine whether the health index is greater than or equal to the preset health threshold:

[0159] If the health index is greater than or equal to the health threshold, it is determined to be a high-confidence alarm, and the first alarm signal is output. The first alarm signal is used to trigger the sound and light alarm and the emergency lights to start, and to instruct personnel to evacuate urgently.

[0160] If the health index is less than the health threshold, it is determined to be a medium confidence alarm, and a second alarm signal is output. The second alarm signal is used to trigger the audible and visual alarm to start, and at the same time, sensor maintenance prompt information is output.

[0161] This step is the second level of sensor health status refinement under environmental risk conditions, and it is the core link in realizing differentiated environmental alarms. After completing the initial environmental risk assessment in S51, the industrial control computer immediately retrieves the gas sensor health index calculated in step S1 from the dedicated storage module. And sensor health thresholds pre-stored in the local storage module , This is the minimum health value required for the sensor to function properly, with a range of [50, 80]. It can be modified on-site based on the sensor type and tunnel construction requirements. Simultaneously, the industrial control machine will monitor... Execution and The same preprocessing procedure is used, with two-point moving average filtering and three consecutive data consistency checks to obtain the filtered health index. This serves as the basis for judgment in this step. Subsequently, the industrial control computer executes a real number comparison instruction. The mathematical judgment indicates that if the judgment is valid, it means that the concentration of the leakage source in the tunnel construction area has indeed exceeded the standard and the sensor is working normally, with high confidence in the monitoring data. Therefore, a high-confidence alarm is triggered, and the first alarm signal S1 is output. The industrial control computer assigns a unique status bit and digital code to S1. Through an independently isolated hardware output point with overload protection, it drives the entire set of emergency alarm hardware in the management box, controlling the audible and visual alarm to emit a continuous high-decibel sound and flashing red light. The emergency lights immediately remain on to provide illumination for personnel evacuation. The explosion-proof telephone automatically dials the preset phone numbers of the tunnel monitoring room and the construction supervisor and plays an emergency evacuation voice prompt. The touch screen displays a full-screen red emergency prompt and highlights all monitoring data. If... The initial assessment is invalid, indicating that although the leakage source concentration exceeds the standard, the sensor's operating status is abnormal, and the monitoring data has limited confidence. Therefore, a medium-confidence alarm is determined, and a second alarm signal S2 is output. The industrial control computer assigns an independent status bit and digital code to S2, controlling the audible and visual alarm to operate intermittently, keeping the emergency light constantly on, and preventing the explosion-proof telephone from automatically dialing. Simultaneously, an orange half-screen prompt box pops up on the touch screen, clearly indicating the sensor's abnormal status and outputting specific sensor maintenance prompts. This is distinct from a high-confidence emergency alarm, avoiding excessive warnings that could interfere with on-site construction. Furthermore, the industrial control computer performs a uniqueness check on the assessment result of this step, ensuring that only one alarm signal, S1 or S2, is output. The assessment result is latched into a dedicated storage module and triggers subsequent data storage. Under either S1 or S2 alarm conditions, the industrial control computer automatically increases the system's acquisition, processing, and assessment frequency from the conventional 10Hz to 50Hz, achieving real-time tracking of concentration and sensor status changes.

[0162] S53: When the corrected leakage source concentration is less than the first alarm threshold, further determine whether the health index is less than the preset maintenance threshold:

[0163] If the health index is less than the maintenance threshold, a third alarm signal is output. The third alarm signal is used to output a sensor forced replacement prompt message and does not trigger an environmental alarm.

[0164] If the health index is greater than or equal to the maintenance threshold, a normal status signal is output.

[0165] This step is the second level of sensor maintenance status refinement judgment under the condition of no environmental risk. Its core is to achieve independent early warning of sensor status, avoiding the impact of sensor failure on the accuracy of subsequent monitoring and early warning, while also not interfering with normal tunnel construction. After the S51 determines that there is no environmental risk, the industrial control computer retrieves the sensor maintenance thresholds pre-stored in the local storage module. This threshold is the minimum health value at which the sensor needs to be forcibly replaced, and its value ranges from [20, 50]. The industrial control computer has this built-in. The system employs a mandatory logic mechanism. If the threshold is modified on-site and this rule is violated, the industrial control machine will automatically refuse to save the change and display a prompt on the touchscreen, ensuring the rationality of the judgment logic from a program perspective. This step still uses the pre-processed filtered health index. As the basis for judgment, the industrial control computer executes a real number comparison instruction. The mathematical judgment indicates that if the judgment is valid, it means there is no environmental risk in the tunnel construction area, but the sensor's working status has severely failed, making it impossible to guarantee the accuracy of subsequent monitoring data. Based on this, a third alarm signal S3 is output. The industrial control computer assigns an independent status bit and digital code to S3, without triggering any environmental alarm hardware action. Only a fixed yellow prompt box pops up in the lower right corner of the touch screen, highlighting and flashing the sensor health index and outputting a clear prompt message for forced sensor replacement. The prompt box does not obstruct the display of the main monitoring data to avoid interfering with on-site construction operations. The invalidation of the initial judgment indicates that there is no environmental risk in the tunnel construction area and the sensors are functioning normally, ensuring the effective implementation of monitoring. Based on this, a normal status signal S0 is output. The industrial control computer assigns a dedicated status bit and digital code to S0. All alarm and notification hardware in the control management box is in standby mode, only displaying real-time environmental monitoring data such as gas concentration, temperature, humidity, and wind speed within the tunnel, as well as the sensor health index and other equipment operating statuses on the touch screen. Simultaneously, the entire system maintains a regular acquisition and judgment frequency of 10Hz to achieve continuous monitoring. The industrial control computer performs a uniqueness verification on the judgment result of this step. The latched judgment result triggers the data storage process, marking the S3 alarm signal as a separate "alarm event." In alarm mode, the entire process data is stored at a frequency of 1 second / time; in normal mode, data is stored at a frequency of 1 minute / time. In addition, the industrial control computer is designed with a dual reset mechanism for all alarm signals S1, S2, and S3: local manual reset and remote monitoring terminal reset. On-site personnel can complete local reset via physical or virtual buttons on the touch screen, while monitoring room personnel can complete remote reset by sending commands through the remote monitoring system. After reset, the industrial control computer will control all hardware to return to standby state and automatically record information such as reset time and reset method to an external industrial SD card. If the industrial control computer detects a fault in core hardware such as an audible and visual alarm or an explosion-proof telephone during alarm hardware linkage, it will immediately switch the drive signal to the backup output point and simultaneously display a blue fault prompt box on the touch screen, recording the faulty device and the switching situation to ensure the reliability of the early warning function.

[0166] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A construction safety monitoring and early warning management box, characterized in that, include: The box body has a top cover; The lid is detachably connected to the box body; The support mechanism is installed inside the box, and its top is fixedly connected to the top cover. The monitoring unit, installed inside the cover and on the support structure, is used to acquire tunnel environmental data; The warning indication mechanism is installed on the support mechanism and the top cover. The support mechanism is used to extend the warning indication mechanism to the outside of the housing. The electrical control module is removable and located on one side of the enclosure. The monitoring mechanism and the early warning indicator mechanism are electrically connected to the electrical control module.

2. The construction safety monitoring and early warning management box according to claim 1, characterized in that, Supporting institutions include: One support tube and two support tubes are symmetrically and fixedly connected to the inner wall of the box. Support tube 2, one end of which is slidably connected inside support tube 1, and the other end is fixedly connected to the top cover; Support frame one is installed on one of the support tubes two; Support frame two is installed between support pipe one and support pipe two, and the wind speed and direction sensor is installed on support frame two; An electric actuator is installed inside support tube one. The output end of the electric actuator is fixedly connected to the bottom of support tube two. The electric actuator two is electrically connected to the electric control module.

3. The construction safety monitoring and early warning management box according to claim 1, characterized in that, The monitoring system includes gas sensors, dust concentration sensors, temperature and humidity sensors, and wind speed and direction sensors, which are used to acquire gas concentration data, dust concentration data, temperature data, humidity data, and wind speed and direction data in the tunnel, respectively. The cover includes a cover body and a cover plate. The cover plate is detachably installed on the cover body. The cover plate has grid holes. The gas sensor, dust concentration sensor, and temperature and humidity sensor are installed in the cover body. The cover body is equipped with a touch screen and control buttons, which are electrically connected to the electronic control module.

4. The construction safety monitoring and early warning management box according to claim 2, characterized in that, The early warning and indication mechanism includes an emergency light, an audible and visual alarm, and a camera; the emergency light is installed on support frame one, the camera is installed at the bottom of the top cover, and the audible and visual alarm is installed on support frame two. The cabinet has an opening on one side. The electrical control module includes a pull-out bracket and a controller. The controller is installed on the pull-out bracket, which is slidably connected to the cabinet through the opening. An explosion-proof telephone is also installed on the enclosure.

5. A construction safety monitoring and early warning management method, characterized in that, Includes the following steps: S1: Obtain the real-time readings and historical operating data of the gas sensor, and calculate the health index of the gas sensor based on the historical operating data; S2: Acquire tunnel images captured by the camera and determine the spatial distance between the leak source and the gas sensor; S3: Obtain wind speed data, and invert the true concentration of the leakage source based on the spatial distance and the wind speed data to obtain the inverted true concentration; S4: Use the health index as a confidence weight to correct the retrieved true concentration to obtain the corrected leakage source concentration; S5: Output a differentiated alarm strategy based on the corrected leakage source concentration and the health index.

6. The construction safety monitoring and early warning management method according to claim 5, characterized in that, In S1, the health index of the gas sensor is calculated based on the historical operating data, specifically including: S11: Extract the degradation characteristic parameters of the gas sensor from the historical operating data. The degradation characteristic parameters include zero drift rate, response time, and noise level. S12: Compare the zero-point drift rate, the response time, and the noise level with the corresponding preset thresholds, and calculate the score value of each feature parameter; S13: The scores of each characteristic parameter are weighted and summed to obtain the health index of the gas sensor; In S11: the zero-point drift rate is obtained by analyzing the linear regression slope of the daily minimum in the historical operating data; the response time is obtained by analyzing the rise time of the step response to blasting events or shotcreting operations in the historical operating data; and the noise level is obtained by analyzing the standard deviation of readings under stable conditions in the historical operating data.

7. The construction safety monitoring and early warning management method according to claim 5, characterized in that, In S2, the spatial distance between the leak source and the gas sensor is calibrated, specifically including: S21: Acquire the sequence of tunnel wall images captured by the camera; S22: Identify the leakage source target in the tunnel wall image sequence using an image recognition algorithm. The leakage source target includes tunnel wall cracks, pipe joints, or equipment sealing surfaces. S23: Calculate the spatial distance between the leak source target and the gas sensor based on monocular vision calibration technology.

8. The construction safety monitoring and early warning management method according to claim 5, characterized in that, In S3, the actual concentration of the inverted leakage source specifically includes: S31: Acquire wind speed and wind direction data inside the tunnel collected by the wind speed and wind direction sensor; S32: Based on the spatial distance, the wind speed data, and the wind direction data, establish a gas diffusion inversion model, which is used to characterize the functional relationship between the sensor-measured concentration and the actual concentration of the leak source; S33: Input the real-time reading of the gas sensor and the spatial distance into the gas diffusion inversion model to calculate the inverted true concentration of the leakage source.

9. The construction safety monitoring and early warning management method according to claim 5, characterized in that, In S4, the health index is used as a confidence weight to correct the retrieved true concentration, specifically including: S41: Obtain the health index and the retrieved true concentration; S42: Establish a confidence mapping function to convert the health index into confidence weights. The confidence weights range from 0 to 1 and are positively correlated with the health index. S43: Correct the retrieved true concentration according to the confidence weight to obtain the corrected leakage source concentration, wherein the corrected leakage source concentration = the retrieved true concentration × the confidence weight.

10. The construction safety monitoring and early warning management method according to claim 5, characterized in that, In S5, based on the corrected leakage source concentration and the health index, a differentiated alarm strategy is output, specifically including: S51: Compare the corrected leakage source concentration with a preset first alarm threshold; S52: When the corrected leakage source concentration is greater than or equal to the first alarm threshold, further determine whether the health index is greater than or equal to a preset health threshold: If the health index is greater than or equal to the health threshold, it is determined to be a high-confidence alarm, and a first alarm signal is output. The first alarm signal is used to trigger the sound and light alarm and the emergency light to start, and to instruct personnel to evacuate urgently. If the health index is less than the health threshold, it is determined to be a medium confidence alarm, and a second alarm signal is output. The second alarm signal is used to trigger the sound and light alarm to start, and at the same time, sensor maintenance prompt information is output. S53: When the corrected leakage source concentration is less than the first alarm threshold, further determine whether the health index is less than a preset maintenance threshold: If the health index is less than the maintenance threshold, a third alarm signal is output. The third alarm signal is used to output a sensor forced replacement prompt message and does not trigger an environmental alarm. If the health index is greater than or equal to the maintenance threshold, a normal status signal is output.

Images