Integrated Management System and Method for Hot Work Operations on Tunnel Waterproofing Membranes and Intelligent Fire Prevention

Abstract

This invention relates to the field of tunnel construction technology, specifically to an integrated management system and method for hot work operations and intelligent fire prevention on tunnel waterproofing membranes. It includes a hot work management subsystem and an intelligent fire prevention subsystem that communicate with each other. The hot work management subsystem includes a mobile terminal and a PC terminal; the intelligent fire prevention subsystem includes interconnected dual-light detectors, automatic water guns, a water tank, and a water pump. The dual-light detectors include a thermal imaging sensor and an infrared-ultraviolet dual-spectrum flame detector. The thermal imaging sensors are deployed at 3-meter intervals below the waterproofing membrane. The infrared-ultraviolet dual-spectrum flame detector covers the ultraviolet band (185-260nm) and the infrared band (760-1400nm). The automatic water gun integrates a rotating joint and a telescopic water hose, and is connected to the infrared-ultraviolet dual-spectrum flame detector. This system solves the problems of low efficiency and limited alarm equipment in traditional manual inspections, ensuring construction safety and is suitable for fire prevention scenarios in tunnel engineering.

Description

Technical Field

[0001] This invention relates to the field of tunnel construction technology, specifically to an integrated management system and method for hot work operations on tunnel waterproofing membranes and intelligent fire prevention. Background Technology

[0002] Fire prevention and control in the waterproofing membrane installation area is crucial during tunnel construction. Traditional fire prevention methods mainly rely on manual inspections and simple alarm devices, which have significant shortcomings: First, manual inspections are easily limited by the environment, making it difficult to achieve all-weather, all-round monitoring, leading to untimely detection of fire hazards; second, simple alarm devices have limited functionality, only triggering alarms when a fire reaches a certain scale, making it difficult to detect fires quickly in their early stages, and failing to provide detailed fire information or accurately locate the fire source, thus delaying fire response. These drawbacks make traditional prevention and control methods ineffective in preventing and responding to fires in a timely manner, failing to meet the high requirements of fire safety in tunnel construction. Therefore, there is an urgent need for a system that integrates hot work operations with intelligent fire prevention management to meet the high requirements of fire safety in tunnel construction. Summary of the Invention

[0003] The problem this invention aims to solve is to overcome the shortcomings of traditional fire prevention and control measures in tunnel waterproofing membrane construction areas, and to provide a professional system that integrates hot work management and intelligent fire prevention. By standardizing hot work procedures, achieving real-time fire monitoring, accurate detection, and rapid response, it effectively prevents fires from occurring, and takes swift measures in the early stages of a fire to reduce fire losses, ensuring the safe and smooth progress of tunnel construction and improving the level of fire safety management in tunnel construction.

[0004] To address the shortcomings of existing technologies, the technical solution adopted by this invention to solve its technical problems is: an integrated management system for hot work operations and intelligent fire prevention of tunnel waterproofing membranes, including a hot work management subsystem and an intelligent fire prevention subsystem that communicate with each other. The hot work management subsystem includes a mobile terminal and a PC terminal.

[0005] The intelligent fire prevention subsystem includes interconnected dual-light detectors, automatic water guns, water tanks, and water pumps. The dual-light detectors include thermal imaging sensors and infrared-ultraviolet dual-spectrum flame detectors. The thermal imaging sensors are deployed at 3-meter intervals below the waterproof membrane. The infrared-ultraviolet dual-spectrum flame detectors cover the ultraviolet band of 185-260nm and the infrared band of 760-1400nm. The automatic water guns integrate rotating joints and telescopic water hoses and are connected to the infrared-ultraviolet dual-spectrum flame detectors. The water tank is equipped with a water level sensor. The intelligent fire prevention subsystem also includes an audible and visual alarm, which is connected to the hot work management subsystem via a network.

[0006] Preferably, the automatic water gun includes an automatic water gun and a water gun controller, the automatic water gun and the water gun controller are connected via an RS-485 bus, and the water pump is connected to the water gun controller via an AC contactor.

[0007] Preferably, the water tank and the water pump are connected by a steel wire hose.

[0008] Preferably, both the water level sensor and the water gun controller are connected to the hot work management subsystem via a MODBUS bus.

[0009] The method for integrating hot work operations on tunnel waterproofing membranes with intelligent fire prevention management system includes the following steps: a: Multi-source data acquisition and image preprocessing: including sensor deployment and image preprocessing;

[0010] b: Data fusion and feature extraction, including temperature-spectral data fusion and feature extraction;

[0011] c: Fire detection and location optimization, including detection criteria and fire source location;

[0012] d: Firefighting execution and closed-loop feedback, including automatic water gun control, secondary verification of ultraviolet light detectors, and linkage response;

[0013] e: Post-event analysis and system self-optimization.

[0014] Preferably, in step b, feature extraction uses Gaussian mixture model (GMM) to analyze the thermal image and extract edge features of high-temperature areas (such as circular diffusion patterns); the MLP model inputs the fused data and outputs a fire probability value (range 0-1).

[0015] Preferably, the determination condition in step c is that the fire probability is ≥0.85 (threshold adjustable) and the overlap rate between the GMM bounding box coverage area and the MLP high probability area is >80%.

[0016] Preferably, in step c, the fire source localization is divided into a PSO stage and a GA stage. In the PSO stage, the particle swarm (50 particles) is initialized with the high temperature point of the thermal imaging as the center, and the optimal coordinates are searched iteratively (10 iterations). In the GA stage, the PSO results are cross-mutated (cross-mutation rate 0.7, mutation rate 0.1), and the final coordinates are output (accuracy ±0.1m).

[0017] Preferably, the image preprocessing in step a includes image enhancement and filtering / denoising, and the specific implementation method is as follows:

[0018] Image enhancement employs histogram equalization to improve the contrast of the original images acquired by the thermal imaging sensor, highlighting the dark fire characteristics. The formula is as follows: Among them, P out(k) is the probability of the k-th gray level in the output image, nk is the number of pixels with the k-th gray level in the original image, and N is the total number of pixels in the image; Gaussian filtering is used to remove noise interference in the image, and the formula is as follows: Where g(x,y) is the value of the Gaussian filter at position (x,y), and σ is the standard deviation.

[0019] Preferably, the probability density function of the Gaussian mixture model (GMM) is p(I), p(I)=ω0·N(I|μ0,∑0)+ω1·N(I|μ1,∑1), where ω0 and ω1 are the weights of the dark fire and the background, respectively, N(I|μ,∑) represents a Gaussian distribution with mean μ and covariance Σ, and I is the image gray value.

[0020] The beneficial effects of this invention are as follows: The successful application of the integrated management system and method for hot work operations on tunnel waterproofing membranes and intelligent fire prevention has brought significant economic and social benefits. In terms of economic benefits, the system effectively prevents and promptly handles fires, successfully avoiding construction interruptions, equipment damage, and material losses caused by fires, significantly reducing construction costs. Taking a certain tunnel as an example, since the system was put into use, it has successfully monitored and handled multiple fire hazards, effectively ensuring the smooth progress of tunnel construction and avoiding potential huge economic losses. At the same time, the system reduces reliance on manual inspections, lowers human resource costs, and improves construction efficiency.

[0021] In terms of social benefits, the system has significantly improved the safety management level of tunnel construction, effectively protected the lives and health of construction workers, reduced the negative impact of fire accidents on the environment and society, and maintained social stability. The widespread application of the system has also enhanced the social image and reputation of construction companies, strengthened their competitiveness in the market, made a positive contribution to the sustainable development of the tunnel construction industry, and promoted the advancement of fire safety technology across the entire industry. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the network topology of the system of the present invention;

[0023] Figure 2 This is a schematic diagram of the hot work management subsystem module involved in the present invention;

[0024] Figure 3 This is a flowchart of the system operation of the present invention;

[0025] Figure 4 This invention relates to a schematic diagram of the entire process of hot work operation management. Detailed Implementation

[0026] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. The embodiments of the present invention are given for illustrative and descriptive purposes only, and are not intended to be exhaustive or to limit the invention to the forms disclosed. Many modifications and variations will be apparent to those skilled in the art. The embodiments were chosen and described to better illustrate the principles and practical application of the invention, and to enable those skilled in the art to understand the invention and design various embodiments with various modifications suitable for a particular purpose.

[0027] The system of this invention mainly consists of a hot work management subsystem and an intelligent fire prevention subsystem. For example... Figure 1 As shown, the system of this invention consists of an intelligent fire prevention subsystem and a hot work management subsystem. The intelligent fire prevention subsystem includes a dual-light detector, an automatic water gun, an audible and visual alarm, a water tank, and a water pump. The dual-light detector integrates a thermal imaging sensor and an infrared-ultraviolet dual-spectrum flame detector. The automatic water gun includes an automatic water gun and a water gun controller, which are connected via an RS-485 bus. The water tank is equipped with a water level sensor. The dual-light detector, automatic water gun, audible and visual alarm, water tank, and water pump are all connected to the hot work management subsystem via signal connections. The intelligent fire prevention subsystem is responsible for data acquisition and command execution. The water pump is connected to the water gun controller via an AC contactor. The water tank and water pump are connected via a steel wire hose. The water level sensor and water gun controller are both connected to the hot work management subsystem via a MODBUS bus. The dual-light detector and automatic water gun are connected to the network via a switch and an RTU. The RTU is an edge computing gateway and is also connected to the water gun controller, water tank, and audible and visual alarm. The hot work management subsystem runs on mobile and PC platforms, allowing operators to submit applications and administrators to approve and monitor them. It is responsible for data processing, analysis, and overall system coordination and control. The specific connection is as follows: the intelligent fire prevention subsystem transmits collected data to the hot work management subsystem via wired or wireless networks. After processing and analyzing the data, the hot work management subsystem sends instructions to the intelligent fire prevention subsystem based on the analysis results. Simultaneously, the hot work management subsystem interacts with the intelligent fire prevention subsystem in real time, enabling operators and administrators to monitor and operate the system status in real time.

[0028] The dual-light detector uses infrared technology to measure the temperature of the detection area in real time, detecting abnormal temperature rises and promptly identifying smoldering fires. An automatic water gun is connected to a dual-spectrum ultraviolet flame detector, which can quickly capture the unique spectral characteristics of flames, achieving efficient flame detection and accurately identifying open flames. When the dual-spectrum ultraviolet flame detector detects a smoldering or open flame, it sends an alarm message to the server via a 4G router. Simultaneously, it activates a water pump to draw fire-fighting water from the tank, and the automatic water gun automatically locates the fire point and sprays water to extinguish it. Upon receiving the alarm message, the server remotely activates the on-site audible and visual alarms to alert personnel to evacuate immediately.

[0029] In terms of hot work management, operators submit hot work applications via mobile software, detailing the location (kilometer marker), time, content, and fire prevention measures. Managers rigorously review the applications on PC software, assessing risks and verifying fire prevention measures. Upon approval, the system automatically generates an electronic permit, which operators use to conduct hot work. During operations, managers monitor the process in real-time via PC software, including whether operators are following regulations and whether work time has exceeded limits. If any abnormalities are detected, such as improper operation, managers can quickly issue instructions via mobile or PC software to halt operations, ensuring safe and compliant hot work. This invention innovatively develops a seamless integration and collaborative working mechanism between mobile and PC software. Operators can conveniently fill out and submit hot work applications via mobile software, recording key information such as the application date, time, location, content, and safety measures. Subsequently, managers rigorously review the applications on PC software, comprehensively assessing risks and deciding whether to approve or request modifications. Once approved, the system automatically generates an electronic permit, which workers can use to carry out hot work. Simultaneously, management personnel can monitor the operation in real time using PC software, including worker information, operation progress, and environmental parameters at the work site. If any abnormalities or violations are detected, management personnel can quickly issue instructions via mobile or PC software to stop the operation immediately, ensuring safety and compliance throughout the entire hot work process.

[0030] This system employs a combination of intelligent algorithms to achieve accurate identification and efficient handling of fire hazards. In terms of data processing, the system first performs data cleaning to remove noise and outliers from the sensor-collected data. Statistical analysis-based cleaning rules are used to discard or correct data that significantly deviates from the normal range. For example, the Z-score method is used to identify and process outlier data points, ensuring data reliability. Next, data fusion is performed, combining temperature data from thermal imaging sensors with spectral characteristic data from infrared-ultraviolet dual-spectrum flame detectors. Weighted averaging and Kalman filtering methods are used to improve data integrity and accuracy, enabling a more comprehensive analysis of fire hazards. For instance, when determining the source of an ignition, the system comprehensively considers the matching degree of areas with abnormally high temperatures and flame spectral characteristics; only when both conditions are met simultaneously is a fire hazard identified, effectively distinguishing between open flames and smoldering fires.

[0031] In fire hazard identification, the system utilizes neural network algorithms to analyze the fused data. A multilayer perceptron (MLP) model is constructed and trained on a large amount of historical fire data to learn characteristic patterns during fires. In practical applications, real-time collected data is input into the trained neural network model, and the model's forward propagation calculates the probability prediction of fire occurrence. If the predicted probability exceeds a set threshold, a fire hazard is identified, thus achieving rapid and accurate identification of fire hazards.

[0032] In addition, the system employs a Gaussian Mixture Model (GMM) to analyze the thermal image data. By creating bounding boxes to highlight detected hotspot areas, the accuracy of fire hazard identification is further improved. The GMM model can automatically learn features in thermal images, such as temperature distribution patterns and edge features, thereby more effectively identifying potential fire areas.

[0033] For fire source localization, the system combines Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) to optimize the localization process. PSO simulates the social behavior of bird flocks to find the optimal solution in the search space, quickly determining the possible location of the fire source. GA simulates natural selection, crossover, and mutation processes to select superior individuals from the population, further improving the accuracy and reliability of the localization. The combination of these two algorithms enables rapid and accurate location of fire sources in complex tunnel environments, providing a reliable basis for the precise spraying of automatic water cannons.

[0034] The thermal imaging sensor has a temperature measurement range of -20℃ to 500℃, adapting to the complex temperature environment inside tunnels. Its temperature measurement accuracy reaches ±2℃, ensuring precise monitoring of temperature changes. The infrared-ultraviolet dual-spectrum flame detector has a carefully set sensitivity threshold, exhibiting high sensitivity to specific bands of the flame spectrum, enabling rapid capture of weak flame signals. Its ultraviolet detection band is 185nm-260nm, and its infrared detection band is 760nm-1400nm, demonstrating strong selectivity for the spectral characteristics of flames. Furthermore, these sensors possess excellent anti-interference capabilities in the complex environment of tunnels. By employing shielding and filtering technologies, the impact of high humidity and dust on sensor performance is effectively reduced, ensuring stable and reliable operation even under harsh conditions.

[0035] The automatic water nozzle employs an advanced mechanical structure design, possessing high-precision positioning capabilities. Its mechanical structure includes a rotating joint, a telescopic water hose, and a nozzle. A motor-driven rotating joint achieves precise horizontal and vertical rotation, and combined with the telescopic water hose's extension and retraction control, it can quickly and accurately target the fire source. Regarding positioning technology, the automatic water nozzle receives fire source location information from the automatic fire management subsystem. This information is derived from the fusion analysis of multi-source sensor data by an intelligent algorithm, including temperature anomaly areas detected by thermal imaging sensors and the spectral characteristics of flames captured by an infrared-ultraviolet dual-spectrum flame detector. Based on this location information, the automatic water nozzle, through its internal control algorithm and drive system, achieves precise location of the fire source and sprays water to extinguish it, ensuring rapid and effective fire control in the early stages and minimizing fire damage. Furthermore, the automatic water nozzle is equipped with an ultraviolet flame detector to further confirm the presence of a flame at the designated location after the nozzle reaches it, reducing the occurrence of accidental spraying. In terms of intelligent fire prevention, the system integrates multiple advanced sensor technologies to achieve comprehensive, blind-spot-free fire monitoring. Thermal imaging sensors are deployed inside the tunnel, installed at the bottom of the waterproofing liner construction area, one every 3 meters. They monitor the temperature distribution in the construction area in real time, accurately identifying areas of abnormal temperature increases and detecting abnormal temperature changes caused by smoldering fires. Infrared and ultraviolet dual-spectrum flame detectors are connected to automatic water cannons to quickly capture the unique spectral characteristics of flames, achieving efficient flame detection. One automatic water cannon is installed in each protection zone, effectively covering the fire protection needs of the construction area. The system employs advanced intelligent algorithms to fuse multi-source data. Through data cleaning to remove noise, data fusion to improve data integrity, and data analysis to automatically identify smoldering and open flames, it automatically sprays water to extinguish fires, greatly improving the timeliness and accuracy of fire detection.

[0036] In actual operation, the system has successfully monitored and handled fire hazards on multiple occasions. For example, during a construction project, the intelligent fire prevention subsystem detected an abnormal temperature rise near a waterproofing slab inside the tunnel, which was identified as an early-stage fire hazard by intelligent algorithms. The system immediately issued an alarm and activated the audible and visual alarm devices via automated control technology, alerting personnel to evacuate in time. Simultaneously, automatic water cannons were automatically activated to spray water at the fire source, quickly controlling the spread of the fire. Management personnel could promptly check the system's operational status and monitoring data via PC software. After confirming that the fire hazard had been eliminated, appropriate measures were taken to resume construction, effectively preventing a potential fire accident and ensuring the smooth progress of tunnel construction. Currently, the system is operating stably, with all functions performing well. It is expected to be fully operational soon, providing a solid guarantee for tunnel construction safety.

[0037] Figure 2This is a schematic diagram of the hot work management subsystem involved in this invention. The hot work management subsystem is presented as a web application and a mini-program. The web version is operated on a PC, and the mini-program is operated on a mobile device. Business interactions are carried out at the hot work management subsystem layer, including hot work application, hot work approval, fire prevention and extinguishing, and equipment communication. The interactive information is stored in the data storage module.

[0038] The entire process of hot work management is as follows: Figure 4 As shown, the specific user roles are categorized as follows: hot work applicants, supervisors, construction units, supervision units, hot work personnel, monitors, and safety officers.

[0039] Step 1: Submit Hot Work Application

[0040] Workers fill out a hot work application form via mobile device, which includes: work location (accurate to the tunnel kilometer marker), time (start and end time periods), work content (welding, cutting, etc.); fire prevention measures (fire extinguisher configuration, isolation materials, personnel division of labor); the system automatically verifies the completeness of the form and prompts for completion if any data is missing.

[0041] Step 2: Hot Work Application Stage: Multi-level Approval and Risk Assessment

[0042] After the application is submitted, a multi-level approval process is triggered on the PC:

[0043] Preliminary review: Supervisors and construction unit personnel review the feasibility of fire prevention measures (e.g., whether the number of fire extinguishers meets the standards);

[0044] Final review: The supervisory unit assesses the risk level of the operation (based on the type of operation, ambient temperature and humidity, and historical accident data);

[0045] Approval result: If approved, the system will generate an encrypted electronic license (including a QR code) and synchronize it to the operator's mobile device; if rejected, the specific reasons will be noted (such as insufficient fire prevention measures or conflicting work hours).

[0046] Step 3: Hot work clearance phase: The hot work operator, supervisor, and safety officer sign in sequence, and the operation is monitored in real time.

[0047] When operators scan the permit QR code to start the operation, the system activates the following monitoring functions:

[0048] Environmental parameter monitoring: Real-time collection of temperature, humidity, and oxygen concentration in the work area (via sensors).

[0049] Operational compliance checks: whether the work time exceeds the permitted time (compare with the permit period), whether the workers are wearing protective equipment (through camera image recognition), and abnormal behavior warnings: if unauthorized personnel are detected entering the work area, trigger an audible and visual alarm.

[0050] Step 4: End, work terminated and records archived

[0051] Normal termination: After the work is completed, the workers submit a completion report (including on-site photos) via mobile device;

[0052] Abnormal Termination: If the system detects a violation (such as failure to shut down the welding machine), the PC will forcibly lock the device and generate an event log.

[0053] All data (application records, sensor data, alarm records) are encrypted and stored in a cloud database.

[0054] Figure 3 This is a flowchart of the system operation of the present invention, which specifically includes the following steps:

[0055] a: Multi-source data acquisition and image preprocessing:

[0056] Sensor deployment: Thermal imaging sensors (installed 3 meters apart under the waterproof membrane) to collect temperature distribution data (resolution 640×480); Infrared and ultraviolet dual-spectrum flame detector (linked with automatic water gun) to detect the ultraviolet (185-260nm) and infrared (760-1400nm) spectra of the flame;

[0057] Image preprocessing includes image enhancement and filtering / denoising, and the specific implementation methods are as follows:

[0058] Image enhancement: Histogram equalization is used to enhance the contrast of the original images acquired by the thermal imaging sensor, highlighting the dark fire features. The formula is as follows: Among them, P out (k) is the probability of the k-th gray level in the output image, nk is the number of pixels with the k-th gray level in the original image, and N is the total number of pixels in the image.

[0059] Noise Reduction Filtering: Gaussian filtering is used to remove noise interference from the image. The formula is as follows: Where g(x,y) is the value of the Gaussian filter at position (x,y), and σ is the standard deviation, which determines the smoothness of the filter.

[0060] b: Data fusion and feature extraction to determine the presence of open or smoldering flames, including temperature-spectral data fusion and feature extraction.

[0061] Temperature-spectral data fusion employs a weighted average method and Kalman filtering: Weighted average method: Temperature data has a weight of 70%, and spectral data has a weight of 30% (because smoldering fires are easier to detect due to temperature anomalies); Kalman filtering: The fusion weights are dynamically adjusted to adapt to dust interference within the tunnel;

[0062] Feature extraction employs a Gaussian Mixture Model (GMM) to analyze thermal images and extract edge features of high-temperature areas (such as circular diffusion patterns). An MLP model takes the fused data as input and outputs fire probability values ​​(range 0-1). Feature extraction is implemented using the Local Autocorrelation Function (LAF) and the Gaussian Mixture Model (GMM).

[0063] Local Autocorrelation Function (LAF): Calculates the autocorrelation function of a local region to capture the variation of thermal radiation in smoldering fires. The formula is as follows: Where LAF(i,j) is the local autocorrelation function value at position (i,j), I is the image gray value, and w is the half-width of the local window.

[0064] Gaussian Mixture Model (GMM): This model models the grayscale values ​​of image pixels to distinguish between dark and background pixels. Assuming dark pixels follow a Gaussian distribution and background pixels follow another Gaussian distribution, the probability density function of the mixture model is p(I), p(I)=ω0·N(I|μ0,∑0)+ω1·N(I|μ1,∑1), where ω0 and ω1 are the weights of the dark and background pixels, respectively, and N(I|μ,∑) represents a Gaussian distribution with mean μ and covariance Σ.

[0065] c: Fire detection and location optimization, including detection conditions and fire source location. If there is an open flame or smoldering fire, determine whether it is within the scope of the hot work application. Otherwise, activate the audible and visual alarm to locate the fire point.

[0066] d: Firefighting execution and closed-loop feedback, including automatic water cannon control, secondary verification of ultraviolet detectors, and linkage response.

[0067] The automatic water gun control system receives the fire point location coordinates and drives the rotating joint (stepper motor accuracy 0.01°) and the telescopic water pipe (stroke error <2cm) to align with the fire source; the ultraviolet light detector verifies whether the fire has been extinguished. If not, the water pump (flow rate 30L / s) is started to spray; if yes, the spraying is stopped and the audible and visual alarm is turned off.

[0068] Linkage response refers to triggering an audible and visual alarm (decibel ≥110dB, light intensity ≥2000 lumens); and pushing the fire source location, temperature curve, and handling record to the PC terminal.

[0069] e: Post-event analysis and system self-optimization

[0070] The system generates fire incident reports in the background (including raw sensor data, intermediate algorithm results, and handling effects); it also optimizes the weights of MLP and Gaussian mixture model (GMM) through reinforcement learning (DQN algorithm) to improve subsequent detection accuracy.

[0071] This invention addresses the pain points of fire prevention and control in tunnel construction by integrating standardized hot work procedures with intelligent fire prevention technology. The system employs a collaborative approval mechanism between mobile and PC terminals, combining multi-source sensor data fusion and deep learning models (MLP, GMM) to achieve rapid fire source identification and precise location, significantly improving fire response speed and accuracy. In practical applications, the system has successfully reduced construction costs by 20%, with a false fire alarm rate of less than 3%, effectively ensuring personnel safety and project progress, and promoting the intelligent upgrade of fire safety technology in tunnel construction.

Claims

1. A method for an integrated management system for hot work operations and intelligent fire prevention of tunnel waterproofing membranes, wherein the integrated management system includes a hot work management subsystem and an intelligent fire prevention subsystem that communicate with each other, and the hot work management subsystem includes a mobile terminal and a PC terminal; The intelligent fire protection subsystem includes interconnected dual-light detectors, automatic water guns, water tanks, and water pumps. The dual-light detectors include thermal imaging sensors and infrared-ultraviolet dual-spectrum flame detectors. The thermal imaging sensors are deployed at 3-meter intervals under the waterproof membrane. The infrared-ultraviolet dual-spectrum flame detector covers the ultraviolet band (185-260nm) and the infrared band (760-1400nm); the automatic water gun integrates a rotating joint, a telescopic water hose, and is connected to the infrared-ultraviolet dual-spectrum flame detector; the water tank is equipped with a water level sensor; the intelligent fire prevention subsystem also includes an audible and visual alarm, which is connected to the hot work management subsystem via a network; its features are: The method includes the following steps: a: Multi-source data acquisition and image preprocessing: including sensor deployment and image preprocessing; b: Data fusion and feature extraction, including temperature-spectral data fusion and feature extraction; c: Fire detection and location optimization, including detection criteria and fire source location; d: Firefighting execution and closed-loop feedback, including automatic water gun control, secondary verification of ultraviolet light detectors, and linkage response; e: Post-event analysis and system self-optimization; In step b, feature extraction uses a Gaussian mixture model to analyze the thermal image and extract edge features of high-temperature areas; the MLP model takes the fused data as input and outputs the fire probability value. In step c, the fire source localization is divided into the PSO stage and the GA stage. In the PSO stage, the particle swarm is initialized with 50 particles centered on the high-temperature point of the thermal imaging; the optimal coordinates are searched iteratively 10 times; in the GA stage, the PSO results are cross-mutated with a cross-mutation rate of 0.7 and a mutation rate of 0.1; the final coordinates are output with an accuracy of ±0.1m. Step a involves image preprocessing, including image enhancement and filtering / denoising. The specific implementation method is as follows: Image enhancement employs histogram equalization to improve the contrast of the original images acquired by the thermal imaging sensor, highlighting the dark fire characteristics. The formula is as follows: ,in, It is the first output image Gray level probability, It is the first in the original image The number of pixels at each gray level. N It is the total number of pixels in the image; Gaussian filtering is used to remove noise from the image, and the formula is as follows: ;in, Is the Gaussian filter at position The value at that location, It is the standard deviation; The probability density function of a Gaussian mixture model (GMM) is: , ,in, and These are the weights of the dark fire and the background, respectively. The mean is μ A Gaussian distribution with covariance Σ. I It is the grayscale value of the image.

2. The method for the integrated management system of hot work operation and intelligent fire prevention for tunnel waterproofing membrane according to claim 1, characterized in that: The automatic water gun and water gun controller are connected via an RS-485 bus, and the water pump is connected to the water gun controller via an AC contactor.

3. The method for the integrated management system of hot work operation and intelligent fire prevention for tunnel waterproofing membrane according to claim 1, characterized in that: The water tank and the water pump are connected by a steel wire hose.

4. The method for the integrated management system of hot work operation and intelligent fire prevention for tunnel waterproofing membrane according to claim 3, characterized in that: Both the water level sensor and the water gun controller are connected to the hot work management subsystem via the MODBUS bus.

5. The method of the integrated management system for hot work on tunnel waterproofing membrane and intelligent fire prevention as described in claim 1, characterized in that: The determination criteria in step c are: fire probability ≥ 0.85, and overlap rate between the GMM bounding box coverage area and the MLP high probability area > 80%.

Images