Tunnel deformation real-time automatic monitoring device and system

Abstract

The invention provides a tunnel deformation real-time automatic monitoring device and system, and relates to the technical field of video monitoring and image data process.The tunnel deformation real-time automatic monitoring device comprises a monitoring mixing device and a monitoring processing system.The monitoring mixing device comprises a deformation sensing platform, a data efficient transmission module and an inspection moving frame; the tunnel deformation real-time automatic monitoring device comprises a deformation sensing platform, the deformation sensing platform is composed of a stainless steel frame and a sensor array, the sensor array comprises a photonic crystal sensor, a flexible piezoelectric film and thermal conductive fibers, and the sensor array is embedded into the inner wall of a tunnel lining through expansion bolts. The problems that a traditional tunnel monitoring system is insufficient in fixed sensor coverage, limited in data fusion capability, lack of dynamic adjustment and insufficient in closed-loop feedback in a complex environment are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of video monitoring and image data processing, and in particular to a real-time automatic monitoring device and system for tunnel deformation. Background Art

[0002] In recent years, with the rapid development of urbanization and infrastructure construction, tunnels, as an important component of railway, subway, and road transportation, play a key role in modern transportation networks. The structural safety of tunnels is directly related to the safety and reliability of traffic operations. Therefore, the demand for real-time monitoring of parameters such as tunnel wall deformation, stress, and temperature is increasing. Traditional tunnel monitoring systems typically use fixed sensors such as strain gauges, inclinometers, or temperature sensors. These can achieve regular monitoring of the tunnel structure status and provide certain support for maintenance and safety management. However, existing technologies still have significant limitations in practical applications and cannot meet the dynamic monitoring and rapid response requirements in complex tunnel environments.

[0003] First, traditional monitoring systems rely heavily on fixed sensors, which are limited in density and placement, making it difficult to fully cover deformation and stress changes on the tunnel's inner wall. This is especially true in joints, cracks, or high-stress areas, where blind spots are prone to occur. For example, fixed sensors cannot dynamically adjust their monitoring positions, making it difficult to capture sudden deformations or localized leakage risks, resulting in potential dangers not being discovered in a timely manner. Second, existing systems have limited data acquisition and analysis capabilities, typically only processing a single type of data (such as deformation or temperature) and lacking a fusion analysis mechanism for multi-source data (deformation, stress, temperature). This makes it difficult for the system to comprehensively assess the overall condition of the tunnel structure, especially in complex environments such as humidity, low temperatures, or high water pressure, where monitoring accuracy and reliability are reduced. Furthermore, traditional systems lack intelligent risk assessment and dynamic adjustment capabilities, and the monitoring frequency and range cannot be adaptively adjusted according to real-time risk levels, resulting in wasted resources or insufficient monitoring of key areas.

[0004] Although some monitoring systems have introduced mobile inspection equipment and preliminary intelligent analysis algorithms in recent years, such as deformation detection based on laser ranging or simple threshold alarm functions, these improvements have not completely solved the problem. First, the path planning and coordination capabilities of mobile inspection equipment are limited, making it difficult to achieve efficient coverage in complex tunnel geometries. In addition, path conflicts are prone to occur when multiple devices collaborate, affecting monitoring efficiency. Secondly, the risk assessment of existing systems relies heavily on manual judgment and lacks automated strategy generation and remedial measure recommendation mechanisms. The response speed is slow, making it difficult to take effective measures in emergency situations (such as rapid crack expansion). Finally, the system lacks closed-loop feedback and optimization mechanisms, and cannot use historical data or user feedback to continuously improve monitoring strategies and remedial plans, which limits its intelligence level and adaptability in long-term operation.

[0005] Therefore, a real-time automatic monitoring device and system for tunnel deformation is needed to solve the above problems. Summary of the Invention

[0006] Technical problems solved

[0007] In view of the deficiencies in the prior art, the present invention provides a real-time automatic monitoring device and system for tunnel deformation, which solves the problems in the above background technology.

[0008] Technical Solution

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions: a real-time automatic monitoring device and system for tunnel deformation, including a monitoring hybrid device and a monitoring processing system, wherein the monitoring hybrid device comprises a deformation sensing platform, a data efficient transmission module and an inspection mobile frame, wherein the deformation sensing platform is composed of a stainless steel frame and a sensor array, wherein the sensor array comprises a phononic crystal sensor, a flexible piezoelectric film and a thermal conductive fiber, wherein the sensor array is embedded in the inner wall of the tunnel lining through expansion bolts and is distributed along the axial direction of the tunnel every 5 meters; the top of the deformation sensing platform is connected to the data efficient transmission module through a buckle, and the data efficient transmission module is composed of a corrosion-resistant aluminum alloy shell and an unprotected The data transmission module is fixed to the arch structure at the top of the tunnel by stainless steel bolts, and one is set every 10 meters; the inspection mobile frame is composed of a centimeter-level wall-climbing robot and a laser rangefinder, and is adsorbed on the inner wall of the tunnel by an electro-permanent magnetic suction cup and moves along the axial direction of the tunnel; the monitoring and processing system includes an analysis and processing unit, a dynamic control module and an interactive display panel. The analysis and processing unit is composed of a high-performance embedded computer, and the dynamic control module is composed of a microcontroller and a drive circuit. They are all installed in the tunnel entrance control room through a steel fixed bracket. The interactive display panel is composed of a touch screen and a display, and is fixed to the wall of the control room through a wall-mounted bracket.

[0010] Preferably, a data acquisition subsystem is provided in the deformation perception platform, and the data acquisition subsystem includes a data acquisition module, a data preprocessing module and a data transmission module; an auxiliary acquisition subsystem is provided in the inspection mobile frame, and the auxiliary acquisition subsystem includes a deformation measurement module and a data processing module; a deformation analysis subsystem, a risk assessment subsystem and a feedback optimization subsystem are provided in the analysis and processing unit, and the deformation analysis subsystem includes a feature extraction module, a data fusion module and a state analysis module, the risk assessment subsystem includes a task scheduling module, a deformation assessment module, a strategy generation module and a remediation generation module, and the feedback optimization subsystem includes a feedback analysis module; a dynamic adjustment subsystem is provided in the dynamic control module, and includes an instruction parsing module, a drive control module and a collaborative scheduling module; a user interaction subsystem is provided in the interactive display panel, and includes a visualization module and an interactive control module;

[0011] The data acquisition subsystem transmits deformation, stress and temperature data to the deformation analysis subsystem through the data efficient transmission module. The deformation analysis subsystem generates a deformation status report and pushes it to the risk assessment subsystem. The risk assessment subsystem generates adjustment strategies and remedial measures. The adjustment strategy is transmitted to the dynamic adjustment subsystem, and the remedial measures are transmitted to the user interaction subsystem. The interactive display panel displays the remedial measures and transmits them to the staff. The feedback optimization subsystem generates optimization instructions based on the user instructions and adjustment results. The phononic crystal sensor of the deformation sensing platform is pre-buried and installed inside the tunnel lining, every 2 meters along the circumference of the tunnel. The sensor array is arranged to collect submillimeter deformation data for evaluating the stability of the tunnel structure; the flexible piezoelectric film is made of polyvinylidene fluoride material, covered with epoxy resin on the inner wall surface of the tunnel, and laid along the axial direction of the tunnel every 3 meters to collect distributed stress data for monitoring the force distribution; the thermal conductive fiber is made of carbon nanotube material, installed at the joints of the tunnel lining by embedding, distributed along the axial direction of the tunnel every 5 meters to collect deformation and temperature coupling data for identifying leakage risks; the stainless steel frame is a rectangular structure, fixed to the inner wall of the tunnel lining by expansion bolts, and coated with a waterproof coating on the surface to support the sensor array and ensure long-term stability.

[0012] Preferably, the corrosion-resistant aluminum alloy shell of the data efficient transmission module is arc-shaped, with a diameter of 10 cm and a height of 5 cm. It is fixed to the arch structure on the top of the tunnel by stainless steel bolts, and is used to protect the wireless communication module and transmit multi-source data; the wireless communication module includes a low-power wide-area wireless network transmitter and a short-range wireless communication transmitter, which transmit standardized deformation data and compression deformation data to the analysis and processing unit respectively; a waterproof communication interface is set on the side of the data efficient transmission module, which is connected to the deformation sensing platform and the inspection mobile frame through corrosion-resistant wires; the surface of the data efficient transmission module is coated with an anti-corrosion coating to adapt to the humid environment of the tunnel.

[0013] Preferably, the inspection mobile frame consists of an aluminum alloy body, a rubber anti-skid track and a laser rangefinder. The body size is 20 cm × 15 cm × 10 cm and weighs 1.5 kg. It is adsorbed on the inner wall of the tunnel through an electro-permanent magnetic suction cup, moves along the axial direction of the tunnel, and collects dynamic deformation data to supplement the blind spots of fixed sensors; the anti-skid track is 5 cm wide and coated with a wear-resistant coating on the surface to adapt to high temperature, high humidity and complex terrain; the laser rangefinder is fixed to the top of the robot by bolts to collect tunnel surface deformation data; the robot surface is coated with a waterproof coating to ensure long-term operation.

[0014] Preferably, the data acquisition subsystem collects data through the deformation sensing platform, specifically including: the data acquisition module receives the original data of the phononic crystal sensor, the flexible piezoelectric film and the thermal conductive fiber, and generates deformation, stress and temperature data; the data preprocessing module uses wavelet transform to denoise, and the unified format is JSON to generate standardized deformation data; the data transmission module transmits the data to the deformation analysis subsystem through a low-power wide-area wireless network; the auxiliary acquisition subsystem generates compression deformation data through the laser rangefinder of the patrol mobile frame, and transmits it to the deformation analysis subsystem through a short-range wireless communication protocol.

[0015] The data utilization and transmission method is as follows: normalized deformation data is used for feature extraction, and compressed deformation data is used for supplementary deformation analysis;

[0016] The control flow is designed as follows: the data acquisition subsystem and the auxiliary acquisition subsystem push data to the deformation analysis subsystem every 30 seconds. If data missing or abnormality is detected (such as the data value exceeds the normal range by 2 times), the control instructions will be used to require re-collection or adjust the frequency to once every 10 seconds.

[0017] Preferably, the deformation analysis subsystem processes multi-source data through an analysis and processing unit, specifically including: a feature extraction module receives standardized deformation data and compression deformation data, and uses a time-frequency analysis method to extract deformation rate, stress distribution and temperature change characteristics to generate a deformation feature set; a data fusion module fuses the deformation feature set through a weighted average method (weights: phononic crystal 0.4, piezoelectric film 0.3, thermal conductivity fiber 0.2, laser 0.1) to generate a comprehensive deformation feature vector; a state analysis module evaluates deformation trends through a time series analysis method, detects high-risk areas through a threshold comparison method, generates a deformation status report, stores it in a local database, and transmits it to the risk assessment subsystem.

[0018] The data utilization and transmission methods are as follows: deformation feature sets are used for risk assessment, comprehensive deformation feature vectors are used for trend analysis, and deformation status reports are used for task optimization;

[0019] The control flow is designed as follows: the deformation analysis subsystem generates a deformation status report and pushes it to the risk assessment subsystem. If specific features are required, re-extraction is requested through control instructions.

[0020] Preferably, the risk assessment subsystem performs deformation assessment and task optimization through an analysis and processing unit, specifically including: the task scheduling module receives deformation status reports and adjustment feedback data, and uses a priority scheduling method to generate a task allocation table for the inspection mobile frame; the deformation assessment module identifies deformation areas through a geometric modeling method, tracks deformation trends through a motion prediction method, constructs a deformation distribution model, and generates a deformation distribution heat map; the remediation generation module generates remediation measures (such as reinforcing seams, spraying a waterproof layer) based on the deformation distribution model and the heat map, combined with a preset reinforcement solution database, and transmits them to the user interaction subsystem;

[0021] Data utilization and transmission methods are as follows: deformation status reports are used for task allocation, compressed deformation data are used for area identification, and thermal maps and remedial measures are used to optimize inspection and maintenance;

[0022] The control flow is designed as follows: the risk assessment subsystem triggers the deformation analysis subsystem to adjust feature extraction based on the deformation assessment results. If the assessment error exceeds 5%, reanalysis is requested.

[0023] Preferably, the risk assessment subsystem generates an adjustment strategy through an analysis and processing unit, specifically including: a strategy generation module receives a task allocation table and a deformation distribution heat map, adjusts the monitoring range through a fuzzy control method, optimizes the path and posture of the inspection mobile frame through a linear programming method, generates an adjustment strategy, and transmits it to the dynamic adjustment subsystem; a remedial generation module generates remedial measures (such as the amount of reinforcement materials and the construction location), and transmits them to the staff through an interactive display panel; a verification module uses historical deformation data and simulated environment data to verify the strategy and remedial measures, generates a verification report, and feeds it back to the deformation analysis subsystem and the feedback optimization subsystem;

[0024] The data utilization and transmission methods are as follows: task allocation tables and heat maps are used to generate adjustment strategies, and remedial measures are used to guide maintenance;

[0025] The control flow is designed as follows: the risk assessment subsystem sends the adjustment strategy to the dynamic adjustment subsystem. If the execution deviation exceeds 10%, it requests to regenerate the strategy.

[0026] Preferably, the dynamic adjustment subsystem realizes dynamic adjustment of the inspection mobile frame through a dynamic control module, specifically comprising: the instruction parsing module parsing the adjustment strategy into path offset and posture adjustment parameters; the drive control module driving the inspection mobile frame through a stepper motor to generate an adjustment execution result; the collaborative scheduling module coordinates multiple groups of inspection mobile frames through a time synchronization method to generate a final adjustment execution result, which is fed back to the risk assessment subsystem;

[0027] The data utilization and transmission method is as follows: control instructions are used to drive the robot to move, and the execution results are used to verify the effectiveness of the strategy;

[0028] The control flow is designed as follows: the dynamic adjustment subsystem pushes the execution results to the risk assessment subsystem. If the deviation exceeds 0.5 meters, it requests readjustment.

[0029] Preferably, the analysis and processing unit is composed of a high-performance embedded computer, installed in the control room, which processes multi-source data and generates deformation status reports and remedial measures; the interactive display panel is composed of a 50 cm × 30 cm touch screen and a display, fixed to the wall of the control room, which displays three-dimensional deformation mapping and transmits remedial measures (such as reinforcement plans) to staff; all hardware surfaces are coated with a waterproof and corrosion-resistant coating to adapt to high temperature and high humidity environments.

[0030] The data utilization and transmission methods are as follows: deformation status reports are used for visual mapping, and remedial measures are displayed and transmitted via the touch screen;

[0031] The control flow is designed as follows: after the user interaction subsystem generates user instructions, they are fed back to the feedback optimization subsystem. If the optimization effect error exceeds 5%, re-analysis is requested.

[0032] Beneficial effects

[0033] The present invention provides a real-time automatic monitoring device and system for tunnel deformation, which has the following beneficial effects:

[0034] This invention proposes a real-time automatic tunnel deformation monitoring device and system, addressing the challenges of traditional tunnel monitoring systems in complex environments, such as insufficient fixed sensor coverage, limited data fusion capabilities, a lack of dynamic adjustment, and insufficient closed-loop feedback. Through the collaborative design of multi-source sensors and mobile inspection equipment, as well as optimized data analysis and user interaction, the device significantly improves the coverage, accuracy, and response efficiency of tunnel deformation, stress, and temperature monitoring, providing reliable assurance for tunnel structural safety.

[0035] By combining a deformation sensing platform with multiple sensors, this device enables high-density, wide-coverage data collection of deformation, stress, and temperature data on tunnel interior walls, effectively overcoming the density limitations of traditional fixed sensor deployments. The combined use of fixed sensors and mobile inspection equipment allows for the precise capture of anomalies in joints, cracks, or high-stress areas, avoiding blind spots. This significantly improves the comprehensiveness and timeliness of anomaly detection, particularly in humid or high-pressure tunnel environments.

[0036] 2. This invention utilizes multi-source data fusion technology within the analysis and processing unit to integrate deformation, stress, and temperature information to generate a comprehensive tunnel condition assessment. Compared to traditional systems that process single data, this significantly improves monitoring accuracy and reliability. This fusion mechanism provides a more comprehensive picture of the tunnel structure's health, providing more robust data support for risk assessment and maintenance decisions.

[0037] 3. By integrating a dynamic control module with a mobile inspection framework, this invention enables dynamic adjustment of monitoring range and path, overcoming the limitations of traditional systems with fixed monitoring frequencies and an inability to adapt to changing risks. The device optimizes inspection tasks based on real-time risk levels, prioritizing high-risk areas and reducing resource waste. Furthermore, it can rapidly adjust monitoring strategies in emergency situations, such as crack expansion or leakage, significantly improving response efficiency and monitoring flexibility.

[0038] Furthermore, the device provides intuitive tunnel status presentation and convenient operation through an interactive display panel and feedback optimization mechanism, overcoming the reliance on manual judgment and insufficient feedback inherent in traditional systems. Users can quickly identify risk areas and identify remedial measures through a visual interface. The system continuously optimizes monitoring parameters and remediation plans based on user input and historical data, ensuring the adaptability and effectiveness of monitoring strategies over the long term. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a system flow chart of the present invention;

[0040] Figure 2 A framework diagram for constructing a monitoring mixing device of the present invention;

[0041] Figure 3 This is a schematic diagram of the installation position of the monitoring mixing device of the present invention in a tunnel;

[0042] Figure 4 It is a system simulation diagram of the present invention;

[0043] Figure 5 It is an enlarged view of point A of the present invention.

[0044] Legend:

[0045] 1. Monitoring mixing device; 2. Deformation sensing platform; 3. Data efficient transmission module; 4. Inspection mobile frame; 5. Analysis and processing unit; 6. Dynamic control module; 7. Interactive display panel; 8. Waterproof communication interface. DETAILED DESCRIPTION

[0046] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention. Specific embodiment one:

[0048] like Figures 1 to 5As shown, a real-time automatic monitoring device and system for tunnel deformation is provided. This device collects tunnel deformation, stress and temperature data through a deformation sensing platform 2 and a patrol mobile frame 4, and transmits the data to an analysis and processing unit 5 via a data efficient transmission module 3 for feature extraction, data fusion and status analysis to generate a deformation status report. The risk assessment subsystem generates adjustment strategies and remedial measures based on the report, and transmits them to the dynamic control module 6 and the interactive display panel 7 respectively. The dynamic control module 6 drives the patrol mobile frame 4 to adjust the path, and the interactive display panel 7 displays the remedial measures to guide the staff to perform maintenance. The feedback optimization subsystem optimizes data collection and strategy generation based on user instructions and execution results. The following describes in detail the complete process from data collection, analysis, judgment to remedial measure generation, execution and feedback, clarifies the data flow between modules, hardware installation method and wireless transmission configuration, ensures full disclosure of technology, and especially details the relevant content of remedial measures.

[0049] Data collection process:

[0050] The data acquisition subsystem, located on Deformation Sensing Platform 2, is responsible for collecting deformation, stress, and temperature data from the tunnel's inner wall. Deformation Sensing Platform 2 consists of a stainless steel frame supporting phononic crystal sensors, flexible piezoelectric films, and thermally conductive fibers. Phononic crystal sensors, 5 cm in diameter, are embedded within the tunnel lining at a depth of 20 cm and arranged circumferentially every 2 meters, for example at angles of 0°, 90°, 180°, and 270°. They collect submillimeter deformation data with a resolution of 0.1 mm and are used to detect structural deformations such as microcracks. Flexible piezoelectric films, made of polyvinylidene fluoride, measure 30 cm x 20 cm and are 0.1 mm thick. They are bonded with epoxy resin and are 0.5 mm thick, covering the inner wall surface. They are laid axially every 3 meters to collect distributed stress data ranging from 0 to 10 MPa and identify areas of stress concentration. The thermal conductive fiber, made of carbon nanotube material, has a diameter of 2 mm and a length of 50 cm. It is embedded in the lining joints at a depth of 5 cm and distributed axially every 5 meters to collect deformation and temperature coupling data with a temperature range of 0-50 degrees Celsius to detect the risk of joint leakage.

[0051] The data acquisition module samples data at a 30-second interval, receiving raw data from three types of sensors. Before sampling, the module calibrates the sensors, checking, for example, that the zero-point drift of the phononic crystal sensor is less than 0.01 mm, that the output voltage of the piezoelectric film is stable between 0 and 5 volts, and that the resistance change of the thermal fiber is less than 0.1 ohm. During sampling, the module records deformation, stress, and temperature values, along with a timestamp and sensor number, for example, "2025-06-23 17:47:00, sensor ID 001, deformation 0.5 mm, stress 2.0 MPa, temperature 25°C." The data preprocessing module uses wavelet transform for denoising, selecting the Daubechies wavelet basis function, a decomposition layer of 4, and a threshold of 0.01. This method retains low-frequency components and removes high-frequency noise such as vibration or electromagnetic interference. After denoising, the data is standardized into a format containing deformation, stress, temperature, and location coordinates, for example, "deformation 0.5 mm, stress 2.0 MPa, temperature 25 degrees Celsius, location x = 100 meters, y = 2 meters." The data packet is approximately 1 kilobyte in size and is stored in a temporary buffer with a capacity of 10 megabytes.

[0052] The data transmission module sends standardized deformation data to the analysis and processing unit 5 via the efficient data transmission module 3. The wireless communication module of the efficient data transmission module 3 uses a low-power wide-area wireless network with a frequency of 868 MHz, a bandwidth of 125 kHz, the LoRa protocol, AES-128 encryption, a transmit power of 20 dBm, and a transmission range of 2 km. Before transmission, a CRC-16 checksum is added to the data packet to ensure integrity. Spread spectrum technology is used during transmission to mitigate interference and adapt to the humid environment of tunnels. If a transmission fails, such as if no confirmation signal is received three times in a row, the module automatically retransmits up to five times. If a transmission fails successfully, the data is stored in the analysis and processing unit 5's receive buffer, which has a capacity of 100 megabytes.

[0053] The auxiliary data collection subsystem, located within the inspection mobile frame 4, supplements data from the fixed sensors' blind spots. The inspection mobile frame 4 consists of a centimeter-class wall-climbing robot equipped with a laser rangefinder measuring 5 cm × 5 cm × 3 cm, fixed to the robot's top. This system collects dynamic deformation data with a resolution of 1 mm and a range of 0-5 meters. The robot moves axially along the inner wall at a speed of 0.1 m / s, covering a specified area (e.g., x = 0 to 1000 meters). The deformation measurement module samples data at a 30-second interval, recording distance values ​​and robot positions, for example, "distance 1.2 mm, x = 150 meters, y = 2 meters." The data processing module compresses data using the Zlib algorithm with a compression ratio of 5:1, resulting in a calibrated coordinate error of less than 0.1 meter. This generates compressed deformation data, consisting of deformation values ​​and coordinates, with a data packet size of approximately 200 bytes. Data is transmitted via waterproof communication interface 12 and corrosion-resistant wires to the short-range wireless communication transmitter in the data efficient transmission module 3. The transmitter operates at a frequency of 2.4 GHz, a bandwidth of 2 MHz, the ZigBee protocol, a transmission power of 0 dBm, a transmission range of 100 meters, and frequency hopping for interference mitigation. If a transmission fails, the module retransmits three times.

[0054] Data analysis process:

[0055] The deformation analysis subsystem is located in the analysis and processing unit 5, which receives standardized deformation data and compression deformation data, and performs feature extraction, data fusion and state analysis. The analysis and processing unit 5 is a high-performance embedded computer with a quad-core processor, 8GB of memory, and 1TB of storage, and is installed in the control room. The feature extraction module uses short-time Fourier transform technology, with a time window of 256 milliseconds and a step size of 128 milliseconds to analyze the time-frequency characteristics of the data and extract the deformation rate (mm / hour), stress distribution (MPa / m2), and temperature change (degrees Celsius / hour). The module filters outliers, for example, deformations greater than 5 mm are marked as invalid and request re-collection. The extracted features constitute a deformation feature set, such as "deformation rate 0.2 mm / hour, stress distribution 3.0 MPa / m2, temperature change 1.0 degrees Celsius / hour, x = 100 meters".

[0056] The data fusion module integrates the feature set and assigns weights based on sensor accuracy: 0.4 for phononic crystal sensors, 0.3 for flexible piezoelectric film, 0.2 for thermal fiber sensors, and 0.1 for laser rangefinders. The fusion process calculates a weighted average. For example, the fused deformation rate value is the sum of 0.4 times the phononic crystal value, 0.3 times the piezoelectric film value, 0.2 times the thermal fiber value, and 0.1 times the laser value. This generates a comprehensive deformation feature vector containing deformation, stress, and temperature values. Before fusion, the module checks data integrity and requests re-acquisition if any data is missing. The fused vector is stored in a 50-megabyte memory buffer.

[0057] The status analysis module assesses deformation trends using a time series analysis with a 24-hour sliding window and a 1-hour step size. It analyzes changes in eigenvectors, such as whether the deformation rate is continuously increasing. The module compares eigenvalues ​​with thresholds and labels deformations greater than 2 mm, stresses greater than 5 MPa, and temperatures greater than 40°C as high risk. The coordinates are recorded, for example, "x = 100 m, y = 2 m." Deformation status reports, including the coordinates, trend, and level of high-risk areas (for example, "x = 100 m, increasing trend, level 1"), are stored in a 1GB SQLite database and transmitted to the risk assessment subsystem via an internal bus at 100 Mbit / s. They are accompanied by a timestamp and checksum.

[0058] Risk judgment and strategy generation process:

[0059] The risk assessment subsystem is located in the analysis and processing unit 5, which receives deformation status reports, compression deformation data, and adjustment execution results of the dynamic adjustment subsystem, and performs task scheduling, deformation assessment, strategy generation, and remediation generation. The task scheduling module parses the deformation status report, assigns priority 1 based on deformation greater than 2 mm, and generates a task allocation table containing inspection coordinates and frequency, such as "x = 100 meters, inspection every 10 seconds". During scheduling, the module checks the number of inspection mobile frames 4, such as 5 robots, to ensure task balance. If multiple high-risk areas conflict, the module sorts them by level and prioritizes level 1 areas. The task allocation table is stored in memory with a capacity of 10 megabytes and transmitted to the strategy generation module.

[0060] The deformation assessment module uses finite element mesh modeling with a resolution of 0.5 meters, dividing the inner wall into a grid, calculating deformation values, and using compression deformation data to calibrate blind spots. The module uses a Kalman filter to predict one-hour trends with a 10-minute step size, generating a deformation distribution model and recording the deformation values ​​and rates of change at each grid point. Based on this model, the module generates a heat map with a resolution of 1920x1080, with red indicating high-risk areas. This map is stored in local storage with a capacity of 500 megabytes and transmitted to the strategy generation module and the user interaction subsystem.

[0061] The strategy generation module generates a control strategy based on the task allocation table and heat map. The module adjusts the monitoring range using fuzzy control. Inputs are a deformation rate of 0-5 mm / h and a stress of 0-10 MPa, and the output range is 0-10 meters. Fuzzy control uses the Mamdani model, with rules such as "If the deformation rate is high and the stress is high, then the range is expanded to 10 meters" and "If the deformation rate is low and the stress is low, then the range is reduced to 2 meters." Input values ​​are categorized as low, medium, and high, and output values ​​are categorized as small, medium, and large. Inference is used to calculate the precise range, for example, for a deformation rate of 2 mm / h and a stress of 5 MPa, the output range is 8 meters. The module optimizes the inspection path using linear programming, with the goal of minimizing path length and the constraint of covering high-risk areas. Path coordinates and attitude angles are generated, for example, "x = 100 meters, y = 2 meters, angle 30 degrees." The control strategy is stored in 5-megabyte memory and transmitted to the dynamic control module 6 via an internal bus.

[0062] The remediation generation module generates remedial measures based on the deformation distribution model and heat map, querying the SQLite reinforcement solution database, which contains 100 preset solutions, such as "deformation greater than 2 mm, steel plate reinforcement 10 cm thick, construction time 2 hours, priority 1" and "temperature greater than 40 degrees Celsius, spray 2 mm waterproof layer, construction time 1 hour, priority 2." The query process matches the deformation, stress, and temperature values ​​of high-risk areas. For example, if x = 100 meters, deformation 2.5 mm, and temperature 42 degrees Celsius, the solution "spray 2 mm waterproof layer" is matched. The module adjusts the priority based on the risk level, setting deformation greater than 3 mm or temperature greater than 45 degrees Celsius to priority 1, and generates remedial measures such as "x = 100 meters, y = 2 meters, spray 2 mm waterproof layer, construction time 1 hour, priority 1" or "x = 200 meters, y = 3 meters, steel plate reinforcement 10 cm thick, construction time 2 hours, priority 1." The measures include specific construction parameters, such as material type, thickness, construction location, and estimated time, along with safety warnings such as "Close tunnel traffic during construction." The module checks the feasibility of the measures, for example, ensuring sufficient reinforcement material inventory. If this is not feasible, it selects a second-best option, such as "Spray a 1mm waterproofing layer." The remedial measures are stored in a 10-megabyte database and transmitted via an internal bus to the interactive display panel 7. Before transmission, the module adds a checksum to ensure data accuracy.

[0063] The verification module evaluates the effectiveness of remedial measures and adjustment strategies, using historical deformation data and a simulated environment. For example, a simulated deformation of 3 mm verifies whether the deformation after reinforcement is less than 1 mm; a simulated path deviation of 0.5 meters verifies whether inspections cover high-risk areas. The verification process compares actual results with expected results. For example, if the remedial action requires a 50% reduction in deformation, and the actual reduction is 40%, the score is 80%. The verification report, including the score and feedback (e.g., "Remedial measures are 80% effective, and it is recommended to increase the reinforcement thickness"), is stored in a 5-megabyte database and fed back to the deformation analysis subsystem and the feedback optimization subsystem. If the score is less than 70%, the module requests that the measures be regenerated.

[0064] Dynamic adjustment process:

[0065] The dynamic adjustment subsystem, located in the dynamic control module 6, receives the adjustment policy and drives the patrol mobile frame 4 to adjust its path and posture. The dynamic control module 6 consists of a 32-bit microcontroller running at 120 MHz and drive circuitry, supporting four stepper motors. The command parsing module parses the adjustment policy, extracting path coordinates and posture angles, such as "x = 100 meters, y = 2 meters, angle 30 degrees," and converts them into control instructions, including a path offset of 0-0.5 meters and an angle of ±10 degrees. During parsing, the module checks the feasibility of the command, for example, whether the path exceeds the tunnel range (y = 0-5 meters). If not, a regeneration request is requested.

[0066] The drive control module activates the stepper motor with a resolution of 0.1 degrees, a torque of 2 Newton meters, and a speed of 0.1 meters per second to move the robot. Before driving, the module calibrates the robot's posture, ensuring, for example, that the angular deviation is less than 0.5 degrees. During driving, the module records the actual path, for example, "actual x = 100.2 meters, deviation 0.2 meters," and generates a control execution result. The collaborative scheduling module uses the NTP protocol with 1 millisecond accuracy to coordinate multiple groups of robots, for example, five, maintaining a spacing greater than 1 meter. Time slices are allocated, with each robot moving once every 10 seconds. The final control execution result, including all robot paths and deviations, is fed back to the risk assessment subsystem via the efficient data transmission module 3, which uses the LoRa protocol at a frequency of 868 MHz. Control commands are issued via the efficient data transmission module 3, and if a transmission fails, they are retransmitted three times.

[0067] Feedback optimization and user interaction process:

[0068] The feedback optimization subsystem, located in analysis and processing unit 5, receives user commands, verification reports, and adjustment execution results from the user interaction subsystem. The feedback analysis module conducts a comprehensive analysis and performance assessment. For example, if the adjustment deviation exceeds 0.5 meters or the remediation action score falls below 80%, it generates optimization instructions, such as "adjust the sampling frequency to 10 seconds" or "prioritize inspections to x = 100 meters." These instructions are stored in a 2-megabyte internal memory and transmitted to the data acquisition subsystem and risk assessment subsystem via an internal bus at a rate of 100 megabits per second. The module verifies command conflicts, for example, to ensure that frequency adjustments do not affect bandwidth.

[0069] The user interaction subsystem, located on interactive display panel 7, receives deformation status reports, heat maps, and remedial measures. Interactive display panel 7 is a 50 cm x 30 cm touchscreen display with a resolution of 1920 x 1080. The visualization module uses OpenGL to render the 3D deformation map, marking high-risk areas in red, such as x = 100 meters. The heat map displays deformation using a color gradient, along with coordinates and timestamps, such as "2025-06-23 17:47:00, x = 100 meters, deformation 2.5 mm." Remedial measures are displayed in text and graphics, such as "x = 100 meters, y = 2 meters, spray waterproofing layer 2 mm, construction time 1 hour, priority 1," with a schematic diagram showing the construction area. The interactive control module receives user commands, such as adjusting the inspection frequency to every 5 seconds or marking a measure "Not applicable, insufficient inventory" via the touchscreen. Staff confirm the execution status, such as "Waterproofing layer sprayed, completion time 2025-06-23 18:00," which is stored in the database and fed back to the feedback optimization subsystem. If the measures are not feasible, the staff inputs the reasons, such as “a thicker waterproof layer is needed”, which is transmitted to the feedback optimization subsystem to trigger the regeneration of measures.

[0070] Detailed process of remedial measures generation and execution:

[0071] The detailed operation of the remediation generation module ensures that the remedial measures are accurate and practical. The module receives the deformation distribution model and thermal map and analyzes the deformation, stress and temperature values ​​in the high-risk areas. During the analysis, the module sorts the priorities, for example, deformation greater than 3 mm or temperature greater than 45 degrees Celsius has priority 1, and deformation 1-3 mm or temperature 40-45 degrees Celsius has priority 2. After sorting, the module queries the reinforcement solution database, which contains 100 solutions. Each solution includes triggering conditions, construction methods, material specifications, construction time and priority. For example:

[0072] Trigger conditions: Deformation greater than 2 mm, stress greater than 5 MPa; Solution: Steel plate reinforcement, thickness 10 cm, construction time 2 hours, priority 1. Trigger conditions: Temperature greater than 40 degrees Celsius, deformation greater than 1 mm; Solution: Spray waterproofing layer, thickness 2 mm, construction time 1 hour, priority 2.

[0073] Trigger conditions: deformation greater than 3 mm, joint leakage; solution: epoxy resin injection, volume 500 ml, application time 1.5 hours, priority 1. When querying, the module matches the area parameters. For example, if x = 100 meters, deformation 2.5 mm, and temperature 42 degrees Celsius, it matches "spray waterproofing 2 mm." If multiple solutions are available, such as steel plate reinforcement and waterproofing, the module selects the one with the highest priority, or selects based on the severity of the heat map, for example, steel plate reinforcement is prioritized in red areas.

[0074] When generating measures, the module specifies construction details: Location: Precise coordinates, such as "x = 100 meters, y = 2 meters, top of the inner lining wall." Material: Specified type and specifications, such as "Polyurethane waterproofing coating, 2 mm thickness, covering an area of ​​1 square meter." Construction time: Estimated duration, such as "1 hour, starting at 6:00 PM on June 23, 2025." Priority: Based on risk level, such as "Priority 1, execute immediately." Safety instructions: For example, "Close the tunnel to traffic during construction to ensure ventilation."

[0075] After generating a measure, the module checks its feasibility, for example by querying the inventory database to ensure that the inventory of waterproofing paint is greater than 10 liters. If this is not feasible, the module selects a second-best option, such as "spraying a 1 mm waterproofing layer." The remedial measures are stored in a 10-megabyte SQLite database with a timestamp, such as "2025-06-23 17:47:00," and transmitted to the interactive display panel 7 via the internal bus at a rate of 100 megabits per second, along with a checksum.

[0076] The interactive display panel 7 receives the remedial measures and displays text and graphics. The text describes the construction details, such as "x = 100 meters, y = 2 meters, spray waterproof layer 2 mm, completed in 1 hour." The graphic shows the location of the construction area in the three-dimensional map, marked in blue, with coordinates and priority. The staff views the measures through the touch screen, confirms execution or enters feedback, such as "Insufficient inventory, need to order paint." After confirmation, the execution status is stored in the database, such as "x = 100 meters, spraying completed, 2025-06-23 18:00", and fed back to the feedback optimization subsystem through the internal bus. If the feedback is not feasible, the module records the reason, such as "A 3 mm thick waterproof layer is required", triggering the remedial generation module to re-query the database and select an alternative solution, such as "Inject 500 ml of epoxy resin." The regenerated measures are transmitted to the interactive display panel 7 again, and the cycle continues until the execution is confirmed.

[0077] The verification module continuously monitors the effectiveness of remedial measures, using historical data and simulation environments. For example, if a deformation of 3 mm is simulated at x = 100 meters, data is collected after reinforcement to verify that the deformation is less than 1 mm. If the effect is insufficient, for example, the deformation remains at 1.5 mm, the module generates a report recommending "increasing the steel plate thickness to 15 cm" and assigning a score of 75%. This report is then fed back to the remedial generation module. The verification report is stored in a 5-megabyte database and transmitted to the deformation analysis subsystem to optimize subsequent analysis thresholds, for example, adjusting the deformation threshold from 2 mm to 1.5 mm.

[0078] Attachment Figure 4 The data flow diagram of the device system of the present invention is shown below:

[0079] The deformation sensing platform 2 generates standardized deformation data, and the inspection mobile frame 4 generates compressed deformation data, which are transmitted to the deformation analysis subsystem through the LoRa or ZigBee protocol of the data efficient transmission module 3.

[0080] The deformation analysis subsystem generates a deformation status report, which is transmitted to the risk assessment subsystem via the internal bus. The risk assessment subsystem generates an adjustment strategy and transmits it to the dynamic control module 6. Remedial measures and thermal maps are transmitted to the interactive display panel 7. The verification report is transmitted to the feedback optimization subsystem. The dynamic control module 6 generates adjustment execution results, which are fed back to the risk assessment subsystem via the efficient data transmission module 3. The interactive display panel 7 generates user instructions and transmits them to the feedback optimization subsystem. The feedback optimization subsystem generates optimization instructions and sends them to the data acquisition subsystem and the risk assessment subsystem. Specific embodiment two:

[0082] like Figures 1 to 5 As shown, the detailed hardware composition, hardware description and operating environment of each module in the real-time automatic monitoring device for tunnel deformation and the system of the present invention are as follows:

[0083] The data acquisition subsystem, located in Deformation Sensing Platform 2, includes a PC-S01 phononic crystal sensor, a PVDF-F02 flexible piezoelectric film, and an IP68 thermal conductivity fiber CNT-T03. The data acquisition and preprocessing module uses a 32-bit STM32F405 microcontroller with 4MB of flash memory for temporary data storage. Wavelet transform denoising uses a Daubechies basis with a 4-layer threshold of 0.01 and a processing speed of 1000 times per second. It generates 1KB JSON data packets, caches 10MB of data, and transmits them to Data Efficiency Transmission Module 3 at a 1Mbps rate. Running FreeRTOS and developed using STM32CubeIDE with HAL library support, the system achieves processing latency of less than 10ms, a standby power consumption of 0.5W, and a total sampling and preprocessing time of less than 20ms.

[0084] The auxiliary data acquisition subsystem, located in the inspection mobile frame 4, includes a LIDAR-LM01 IP67 laser rangefinder and a CR-B02 wall-climbing robot. The deformation measurement and data processing module uses an ESP32-S3 Wi-Fi module with ZigBee support, a 30-second sampling period, 2MB of storage, and Zlib compression (5:1500 times per second) with a 1MB buffer. Data is transmitted at 1Mbps via M12 waterproof PVC cables, with ZigBee latency under 20ms. Developed in C using ESP-IDF and supported by a Wi-Fi driver and compression library, the module achieves compression and calibration times under 15ms, a data loss rate under 0.1%, and coordinate errors under 0.1m.

[0085] Deformation Analysis Subsystem: Located in Analysis Processing Unit 5, NVIDIA Jetson Xavier NX SQLite 1GB. The feature extraction module uses the GPU to run short-time Fourier transforms (SFTs) with a 256ms window and a 128ms step size, running at 1000 sets per second. The data fusion module uses the CPU to run 500 sets per second, with weights of phonon 0.4, piezoelectric 0.3, thermal conductivity 0.2, and laser 0.1, using a 50MB buffer. The state analysis module uses 100 sets per second, running a 24h sliding window with a 1h step size, threshold deformation 2mm, stress 5MPa, and temperature 40°C. This system runs Ubuntu 20.04 with the JetPack SDK, CUDA 11.4 acceleration, and SQLite API support. Feature extraction takes less than 5ms, and fusion and analysis take less than 1s.

[0086] The risk assessment subsystem shares a Jetson Xavier NX processor. The task scheduling module runs 1000 tasks per second on the CPU. The deformation assessment module runs a 0.5m grid and a 10,000-grid heat map in 500ms on the GPU. The strategy generation module uses 50 fuzzy control groups per second, a Mamdani model with a monitoring range of 0-10m, and linear programming optimization paths. The remediation generation module uses SQLite queries for 100 solutions, 1000 times per second, and 10ms per second. The verification module uses both the CPU and GPU for 1000 simulations per second. Running Ubuntu 20.04 and using the JetPack SDK, CUDA-accelerated modeling, and optimized SQLite indexes, scheduling takes less than 1ms, modeling and strategy generation takes less than 1s, and remediation generation takes less than 10ms.

[0087] The dynamic regulation subsystem, located in dynamic control module 6, uses the NXP i.MX RT1062 DRV8833 to drive the CAN bus. Command resolution is 5ms, drive control is 0.1ms per second, and response time is 1ms. Coordinated scheduling and NTP synchronization with 1ms accuracy and 10s time slice functionality are implemented through the MCUXpresso SDK. Developed in C, it supports PWM and CAN. Parsing and execution take less than 10ms, and path deviation is less than 0.2ms.

[0088] Feedback Optimization Subsystem: Sharing a Jetson Xavier NX, the feedback analysis module CPU uses 100 SQLite 2MB blocks at 10ms per second, transmitting optimization instructions 1000 times per second via Gigabit Ethernet. Running Ubuntu 20.04 with SQLite API support, analysis and generation take less than 15ms, with an optimization accuracy of 95%.

[0089] User Interaction Subsystem: Located on Interactive Display Panel 7, ELO 3243L. The visualization module runs OpenGL 4.0 at 60fps in less than 50ms. The interactive control module uses capacitive touch and runs on embedded Linux, developed with Qt 5.15 and supporting QOpenGLWidgetQTouchEvent. Rendering and interaction take less than 50ms, with 98% input accuracy.

[0090] Monitoring hybrid devices and data transmission: Deformation sensing platform 2 utilizes a stainless steel frame secured with expansion bolts and coated with a waterproof coating. The 3AL-DT01 data transmission module features an IP68-rated LoRa SX1276 ZigBee CC2530 STM32L432M12 interface and runs Zephyr RTOS. It supports the LoRaWAN ZigBee protocol stack, achieving transmission times of less than 50ms and a 99.8% communication success rate.

[0091] Real-time and feasibility: Data acquisition takes less than 20ms, auxiliary data acquisition less than 30ms, deformation analysis less than 1s, risk assessment less than 1s, dynamic adjustment less than 10ms, user interaction less than 50ms, and data transmission less than 50ms. Jetson Xavier NX, STM32 ESP32, LoRa ZigBee, Qt OpenGL, FreeRTOS Zephyr, and IP67 and IP68 protection ensure real-time monitoring of a 1000m tunnel, with deformation detection accuracy of 97.5% and path deviation less than 0.2m. Specific embodiment three:

[0093] like Figures 1 to 5 As shown, the key algorithms mentioned in Example 1 are analyzed in detail below:

[0094] The data acquisition subsystem, located on Deformation Perception Platform 2, utilizes wavelet transform denoising and data standardization algorithms. This aims to address the problem of raw sensor data being contaminated by noise from vibration, electromagnetic interference, and other factors in tunnel environments, ensuring the accuracy of deformation, stress, and temperature data and providing reliable input for subsequent analysis. The core logic uses wavelet transforms to decompose data into high- and low-frequency components, filtering out high-frequency noise while retaining the low-frequency signals that reflect the actual deformation. This process also unifies multi-sensor data into a standard format for easier transmission and processing. The specific steps include: first, starting a 30-second periodic sampling to obtain the original data of the phononic crystal sensor, flexible piezoelectric film, and thermal conductive fiber, such as deformation of 0.52 mm, stress of 2.1 MPa, and temperature of 25.2 degrees Celsius; then checking the data integrity. If the missing value exceeds 10% or the deformation exceeds 5 mm, mark it as invalid and request resampling; then apply wavelet transform, select Daubechies wavelet basis function, decompose the number of layers to 4, set the denoising threshold to 0.01, set the signal less than the threshold in the high-frequency component to zero, retain the low-frequency component, and reconstruct data such as deformation of 0.50 mm, stress of 2.0 MPa, and temperature of 25.0 degrees Celsius; finally, standardize the data, add a timestamp such as 18:46:00 on June 23, 2025, a sensor number such as ID001, and position coordinates such as x = 100 meters, y = 2 meters, generate a data packet of about 1 kilobyte, store it in a 10-megabyte cache area, and transmit it to the data efficient transmission module 3 first. The algorithm takes about 10 milliseconds to process each set of data and 1 millisecond to normalize, meeting a 30-second sampling period. Fixed decomposition layers and thresholds reduce computational complexity. Tests have shown that 90% of noise can be filtered out, and data accuracy reaches 99%, ensuring real-time and reliability.

[0095] The auxiliary data collection subsystem, located in the inspection mobile frame 4, utilizes Zlib data compression and coordinate calibration algorithms to address the challenges of large laser rangefinder data volumes, limited transmission bandwidth, and coordinate deviations caused by robot motion, thereby improving data transmission efficiency and positioning accuracy. The core logic performs lossless compression on the laser rangefinder's deformation data to reduce the data packet size. It also utilizes track encoders and an inertial measurement unit to calibrate coordinates, eliminating motion errors and generating accurate compressed deformation data. The specific steps are as follows: Sampling is performed at a 30-second interval to obtain deformation values ​​and preliminary coordinates, such as a distance of 1.22 mm, x = 150.1 meters, and y = 2.1 meters. Data validity is checked; if the distance exceeds 0-5 meters, the data is invalidated and resampled. Distance traveled is obtained from the track encoder with an accuracy of 0.01 meters, and angular deviation is obtained from the inertial measurement unit with an accuracy of 0.1 degrees. The coordinates are corrected to x = 150.0 meters, y = 2.0 meters. The Zlib compression algorithm is applied, setting compression level 5 and a compression ratio of 5:1, generating a data packet of approximately 200 bytes. The decompressed data is verified to be within 0.01 mm of the original data, otherwise it is recompressed. The data is stored in a 1-megabyte buffer and transmitted to the data efficient transmission module 3 via the ZigBee protocol. The total compression and calibration time is approximately 15 milliseconds, meeting real-time transmission requirements. Tests have shown a data loss rate of less than 0.1% and a coordinate error of less than 0.1 meter, making it suitable for dynamic monitoring of a 1000-meter tunnel.

[0096] The deformation analysis subsystem is located in the analysis and processing unit 5, and includes short-time Fourier transform feature extraction, weighted average data fusion and time series trend analysis algorithms. It solves the problems of heterogeneous multi-sensor data, difficulty in comprehensively evaluating deformation trends, and untimely identification of high-risk areas, ensuring accurate early warning. The core logic is to decompose data into time-frequency components, extract dynamic features, fuse multi-sensor data to improve reliability, and mark high-risk areas through long-term trend analysis. The specific steps include: receiving standardized deformation data and compression deformation data, such as deformation of 0.50 mm, stress of 2.0 MPa, temperature of 25.0 degrees Celsius, x = 100 meters; applying short-time Fourier transform with a window of 256 milliseconds and a step size of 128 milliseconds to extract deformation rate such as 0.2 mm / hour, stress distribution such as 3.0 MPa / square meter, and temperature change such as 1.0 degrees Celsius / hour; filtering abnormal features, such as requesting re-collection if the deformation rate is greater than 5 mm / hour; assigning weights (0.4 for phononic crystals, 0. 3. Thermal fiber 0.2, laser 0.1), calculate the weighted average, for example, at a deformation rate of 0.18 mm / hour, and generate a comprehensive feature vector. Time series analysis is performed using a 24-hour sliding window and a 1-hour step size, comparing the features with thresholds (deformation 2 mm, stress 5 MPa, temperature 40°C). If the deformation is 2.5 mm, mark x = 100 meters as high risk, record the trend (if it is increasing), and generate a deformation status report containing coordinates, trends, and levels (e.g., level 1). The report is stored in a SQLite database and transmitted to the risk assessment subsystem. Feature extraction takes 5 milliseconds, fusion takes 2 milliseconds, and time series analysis takes 500 milliseconds, meeting real-time requirements. Optimizing the window and weights reduces complexity. Tests show that the accuracy of identifying high-risk areas is 95%, with a false negative rate of less than 1%.

[0097] The risk assessment subsystem is located in the analysis and processing unit 5. It adopts priority task scheduling, finite element mesh modeling, fuzzy control strategy generation, database matching remediation generation and historical data verification algorithms to solve the problems of uneven distribution of inspection tasks in high-risk areas, inaccurate deformation distribution prediction, untimely or inapplicable remedial measures, and provide efficient decision support. The core logic is to schedule tasks according to risk levels, model and predict deformation, optimize inspections with fuzzy control, generate measures with database matching, and verify the effects. The specific steps are: parse the deformation status report, extract high-risk areas such as x = 100 meters, deformation 2.5 mm, level 1; assign tasks according to level, inspect level 1 areas every 10 seconds, and generate a task table; if 5 robots cover 10 areas, high levels are prioritized and low levels are delayed; divide the grid with a resolution of 0.5 meters, calculate the deformation, and calibrate with laser data; Kalman filter predicts 1-hour trends with a step size of 10 minutes, generates a heat map, and marks high risks in red; fuzzy control input deformation rate 0-5 mm / hour, stress 0-10 megahertz The system is divided into three levels: low, medium, and high. The monitoring range is 0-10 meters. For example, a rule such as "high deformation rate and high stress, range 10 meters" generates an output of 8 meters. Linear programming optimizes the path, minimizing distance and generating strategies such as x = 100 meters, y = 2 meters, and an angle of 30 degrees. A database query matches a deformation of 2.5 mm and a temperature of 42°C, generating measures such as spraying a 2 mm waterproofing layer with construction details. A deformation of 3 mm is simulated, and after verification, the reinforcement deformation is less than 1 mm. A report is generated, such as 80% effectiveness. Strategies, measures, and reports are stored and transmitted. Scheduling takes 1 millisecond, modeling takes 500 milliseconds, strategy and remedy generation each takes 10 milliseconds, and verification takes 1 second, meeting real-time requirements. Testing shows a 90% remedy applicability rate and 100% path coverage.

[0098] The dynamic adjustment subsystem, located in dynamic control module 6, uses command parsing and path execution, along with multi-robot coordination algorithms, to address inaccurate path execution and the risk of multi-robot collisions caused by the tunnel's complex geometry, ensuring efficient and safe inspections. Its core logic parses the path strategy into motor commands, precisely controls robot movement, and allocates time slices to synchronize multiple robots. The specific steps are: Parse the strategy, extracting x = 100 meters, y = 2 meters, and a 30-degree angle; verify that the path is within the range of x = 0-1000 meters, y = 0-5 meters, otherwise request a new strategy; convert it into motor commands, such as forward 100 steps and rotate 300 steps; execute it through stepper motors, and record the actual path, such as x = 100.2 meters with a 0.2-meter deviation; allocate time slices to the five robots, such as robot 1 moving at 0 seconds and robot 2 at 10 seconds, maintaining a 1-meter spacing; synchronize using the NTP protocol with an error of less than 1 millisecond; and record the execution results and transmit them to the risk assessment subsystem. Parsing and execution each take 5 milliseconds, and coordination takes 1 millisecond, meeting real-time adjustments. Fixed time slices reduce complexity. Tests show that path deviation is less than 0.2 meters and the collision avoidance rate is 100%.

[0099] The feedback optimization subsystem, located in analysis and processing unit 5, employs feedback analysis and optimized command generation algorithms to address issues such as unoptimized sampling frequency or inspection priority, as well as slow user feedback response, thereby improving system performance. Its core logic analyzes user commands, verifies reports and execution results, identifies performance gaps, and adjusts parameters. The specific steps are: It collects commands such as "inspection frequency 5 seconds," reports such as "80% effectiveness," and results such as "deviation 0.5 meters." It analyzes deviations exceeding 0.5 meters or effectiveness below 85%, flagging them as poor performance. It generates commands such as "sampling frequency 10 seconds, priority x = 100 meters." It verifies that commands are conflict-free, such as if the frequency does not exceed the bandwidth. It then transmits the commands to the data acquisition and risk assessment subsystem and stores them in memory. Analysis takes 10 milliseconds, and generation takes 5 milliseconds, meeting real-time optimization requirements. Fixed thresholds simplify decision-making. Testing shows that optimization improves accuracy by 10% and command success rate by 95%.

[0100] The user interaction subsystem, located on the interactive display panel 7, uses 3D visualization and interactive control algorithms to address the challenges of difficult-to-understand deformation data and slow user response to remedial instructions, thereby improving decision-making efficiency. Its core logic renders a color-coded 3D deformation map and interprets touch input to adjust parameters or confirm actions. The specific steps are: receiving reports and heatmaps, such as x = 100 meters, 2.5 mm deformation, and a red zone; rendering the 3D map using OpenGL, with red indicating high risk, along with coordinates; displaying remedial actions, such as "Spray a 2 mm waterproofing layer, priority 1," with an accompanying diagram; capturing touches, such as selecting x = 100 meters and sliding to adjust the frequency by 5 seconds; verifying that the parameters are within the 1-60 second range, prompting a reselection; generating commands, such as "Frequency 5 seconds" or "Spraying completed," which are then transmitted to the feedback optimization subsystem for storage and confirmation. Rendering takes 50 milliseconds, and input processing takes 10 milliseconds, meeting real-time display requirements. Optimized rendering achieves 60 frames per second. Testing demonstrates 98% input accuracy and stable interactive operation. Specific embodiment four:

[0102] like Figures 1 to 5 As shown, the following provides specific use cases:

[0103] Case 1: Urban subway tunnel lining crack monitoring: A 500-meter urban subway tunnel with a 0.5-meter concrete lining, 85% humidity, and frequent traffic vibrations was deployed. Monitoring began at 12:00 PM on May 23, 2024, to detect deformation, stress, and temperature to prevent structural deterioration. A deformation sensing platform was deployed every 5 meters, with 100 monitoring points equipped with phononic crystal sensors, flexible piezoelectric films, and thermal conductivity fibers. Five LoRa / ZigBee nodes were located every 10 meters in the arch. Three wall-climbing robots equipped with laser rangefinders covered an x-range of 0-500 meters and a y-range of 0-5 meters. An analysis and processing unit, dynamic control module, and a 50 cm x 30 cm touchscreen were placed in the control room. At x = 150 meters and y = 180°, deformation of 2.3 mm, stress of 4.5 MPa, and temperature of 28°C were measured. Data was denoised using wavelet transform (4 layers, threshold 0.01), with a 1KB data packet transmitted via LoRa in 30 milliseconds. A robotic laser measured deformation of 2.4 mm, with a 200B data packet transmitted via ZigBee in 20 milliseconds. A short-time Fourier transform (window 256 milliseconds, step size 128 milliseconds) was used to extract a deformation rate of 0.2 mm / hour, a stress distribution of 4.5 MPa / square meter, and a temperature change of 0.5°C / hour. A weighted fusion (phonon 0.4, piezoelectric 0.3, thermal conductivity 0.2, and laser 0.1) was used to generate a feature vector. A 24-hour sliding window analysis flagged the area x = 150 meters as high risk (level 1, increasing trend). Inspections were scheduled every 10 seconds at x = 145-155 meters. A Kalman filter was used to predict a one-hour trend. The heat map marked x = 150 meters in red. Fuzzy control suggested an 8-meter monitoring range. The optimized path was x = 145-155 meters and y = 175-185 degrees. Remedial measures included spraying a 2mm polyurethane waterproofing layer, applying for one hour, with priority 1. The robot moved along the optimized path (with a deviation of 0.1 meters and a time of 10 seconds). The touchscreen displayed a 3D deformation map, red high-risk areas, and remediation details. The operator confirmed "Sprayed, May 23, 2024, 1:30 PM" with 90% effectiveness. The sampling frequency was adjusted to 15 seconds. Result: Cracks were precisely located, quickly repaired, and prevented from spreading.

[0104] Case 2: Leakage Monitoring of Joints in Mountain Railway Tunnels: A 1000-meter railway tunnel with a 0.6-meter concrete lining, 15°C temperature, and 90% humidity present a high risk of leakage. Monitoring commenced at 2:00 PM on June 23, 2024, to monitor temperature, deformation, and stress to ensure waterproof performance. A deformation sensing platform was installed every 5 meters, with 200 monitoring points embedded in the lining and joints. Ten LoRa / ZigBee nodes were installed every 10 meters, located in the arch. Four wall-climbing robots were equipped with laser rangefinders, covering x = 0-1000 meters and y = 0-5 meters. The analysis and processing unit, dynamic control module, and touch screen were located in the control room. At x = 600 meters and y = 0°, the measured temperature was 40°C, deformation was 1.9 mm, and stress was 3.0 MPa. After denoising, the data was transmitted via ZigBee in 25 milliseconds. The robot's laser measurement of deformation was 1.85 mm, with a transmission time of 15 milliseconds. Feature extraction revealed a deformation rate of 0.15 mm / hour, a stress distribution of 3.0 MPa / m2, and a temperature change of 1.0°C / hour. These features were fused to generate a feature vector, and the analysis marked x = 600 meters as high risk (level 2, stable trend). Inspections were scheduled for x = 595-605 meters, with each inspection occurring every 15 seconds. The heat map marked x = 600 meters orange, with a monitoring range of 5 meters and a path of x = 595-605 meters. The remedial measure was to inject 500 ml of epoxy resin, with a construction time of 1.5 hours and a priority of 2. Two robots conducted a collaborative inspection (with a spacing of 1 meter, a deviation of 0.15 meters, and a time duration of 12 seconds). The touchscreen displayed the heat map and remedial measures, and the staff confirmed "Injection completed, June 23, 2024, 16:00", with an effectiveness of 85%. The sampling frequency was adjusted to 10 seconds. Results: Leaks were accurately located, reinforcements were implemented in a timely manner, and waterproofing performance was guaranteed.

[0105] Case 3: Monitoring High-Stress Areas in an Undersea Tunnel: An 800-meter undersea tunnel with a 0.7-meter concrete lining, water pressure of 0.5 MPa, and a temperature of 20°C were monitored. Monitoring began at 4:00 PM on October 23, 2024, to detect stress, deformation, and temperature, allowing for rapid reinforcement and preventing structural failure. A deformation sensing platform was installed every 5 meters, with 160 monitoring points embedded in the inner wall. Eight LoRa nodes were located at the top, one every 10 meters. Three wall-climbing robots were equipped with laser rangefinders, covering an area of ​​0 to 800 meters (x = 0-800 meters) and 0 to 5 meters (y = 0-5 meters). The analysis and processing unit, dynamic control module, and touch screen were located in an onshore control room. At x = 300 meters and y = 270°, stress of 6.2 MPa, deformation of 2.8 mm, and temperature of 30°C were measured. Data was transmitted via LoRa in 35 milliseconds. The robot's laser measurement of deformation was 2.9 mm, with a transmission time of 18 milliseconds. Feature extraction revealed a deformation rate of 0.3 mm / hour, a stress distribution of 6.2 MPa / m2, and a temperature change of 0.2°C / hour. These features were fused to generate a feature vector, and the analysis marked x = 300 meters as high risk (Level 1, sharp increase). Inspections were scheduled every 10 seconds from x = 290-305 meters, with the heat map marking x = 300 meters red. The monitoring range was 10 meters, and the path x = 290-310 meters. The remedial measure was to reinforce a 10-cm thick steel plate. The construction took two hours, with a priority of 1. Three robots conducted a collaborative inspection (with a deviation of 0.12 meters and a time of 15 seconds). The touchscreen displayed the red high-risk area and the remedial measures. The staff confirmed, "Steel plate reinforcement completed, October 23, 2024, 18:30," with an effectiveness of 92%. The deformation threshold was adjusted to 2.5 mm. The result: Rapid response to high stress risks ensured structural stability.

[0106] The following provides complete experimental data, generated through laboratory simulations and actual hardware testing. Based on the aforementioned case, the test environment was a simulated tunnel (1000 meters long, concrete lining thickness 0.5-0.7 meters, 80% humidity, and temperature 15-40°C). The test simulated cracks, leakage, and high-stress scenarios using hardware such as the Deformation Sensing Platform 2 and the Inspection Mobile Frame 4. Detection accuracy, response time, remediation effectiveness, and path deviation were recorded. The following is a table of experimental data:

[0107]

[0108] The table reflects the performance of the devices in the three case studies. The Deformation Detection Value column displays the measurement results of the Deformation Sensing Platform 2 and the Inspection Mobile Frame 4. For example, in Case 1, the phononic crystal sensor measured 2.3 mm, while the laser rangefinder verified it to be 2.4 mm, a difference of 0.1 mm, indicating high sensor consistency. Detection accuracy is calculated by comparing with known deformation calibration values. Case 3 has an accuracy of 99.0%, reflecting reliability in high-stress scenarios. Response time, the total time from data acquisition to remedial action generation, is 0.05 seconds in Case 1 and slightly higher in Case 2 due to the collaborative nature of leak verification requiring multiple robots. Case 3 has a response time of 0.06 seconds, both meeting real-time requirements. The Remedial Action column corresponds to the output of the risk assessment subsystem. In Case 1, the epoxy resin effectiveness is 85%, indicating an 85% reduction in deformation. In Case 3, the steel plate reinforcement effectiveness is 95%, representing the best results. Path deviation reflects the execution accuracy of the inspection mobile framework 4. Case 1 exhibited a deviation of 0.1 meter, Case 2 exhibited a deviation of 0.15 meter, and Case 3 exhibited a deviation of 0.12 meter, all within 0.2 meters. This demonstrates the accuracy of the dynamic adjustment subsystem's control. These data demonstrate that the device can rapidly detect high-risk areas, generate effective remedial measures, and maintain high-precision inspections in various scenarios, meeting the diverse needs of tunnel safety monitoring.

[0109] The test platform uses the hardware specified in the above embodiment, including a deformation sensing platform 2, a patrol mobile frame 4 (laser rangefinder L IDAR-LM01), an analysis and processing unit 5 (NVIDIA Jetson Xavier NX), a dynamic control module 6 (NXP Pi.MXRT1062), and an ELO3243L touch screen. A simulated tunnel is set with known deformation points (1.8-3 mm), leakage points (temperature 40 degrees Celsius), and high stress points (6.2 MPa). The detection values ​​are verified using high-precision instruments (laser displacement meter, stress tester). The response time is recorded in the system log, and the effectiveness of the remediation is evaluated by the percentage reduction in deformation after construction. The path deviation is measured by the robot encoder. The experiment is repeated 10 times, and the average value is taken. The standard deviation is less than 5% to ensure data stability. Specific embodiment five:

[0111] like Figures 1 to 5 As shown, the following is a supplement to the above embodiment:

[0112] To ensure the long-term accuracy and stability of the phononic crystal sensors, flexible piezoelectric films, and thermal conductive fibers in the deformation sensing platform 2, this device is designed with automatic calibration and regular maintenance mechanisms to address sensor drift or aging that may be caused by high humidity, high pressure, or vibration environments.

[0113] Calibration process:

[0114] Calibration frequency: Automatic calibration is performed every 30 days, or instant calibration is triggered when abnormal data is detected, that is, deformation value exceeds 5 mm, stress exceeds 10 MPa, and temperature exceeds 50°C.

[0115] Calibration method:

[0116] Phononic crystal sensors: Zero drift is calibrated using a built-in reference signal. During calibration, the data acquisition module sends a test signal and records the deviation of the sensor output from the reference value. If the deviation is greater than 0.01 mm, the sensor gain is adjusted. This process takes approximately 5 seconds. The calibration tool uses embedded firmware, specifically the STM32F405, with HAL library support, and stores calibration parameters in 4MB of flash memory.

[0117] Flexible piezoelectric film: Apply a known stress and calibrate the output voltage (0-5V). If the voltage deviation is >0.1V, adjust the amplifier circuit parameters. This takes approximately 3 seconds. Calibration is performed via the SPI interface, and the parameters are recorded in flash memory.

[0118] Thermal Conductivity Fiber: Calibrate resistance changes using a constant temperature calibrator (range 0-50°C, accuracy 0.1°C). If the resistance deviation is > 0.1 ohm, the calibration curve is updated, which takes approximately 4 seconds. Calibration is performed via the UART interface and stored in flash memory.

[0119] Calibration environment requirements: temperature 15-40°C, humidity <90%, no strong electromagnetic interference. Calibration data is stored in a SQLite database, including timestamp, sensor ID, and deviation value, for analysis by the feedback optimization subsystem.

[0120] Maintenance process:

[0121] Regular Inspection: Have personnel inspect the physical condition of the sensor every six months. If the crystal structure of a phononic crystal sensor is damaged, replace it with a new sensor, model PC-S01. If the epoxy resin on the flexible piezoelectric film peels off, re-bond or replace it, model PVDF-F02. If the carbon nanotubes in the thermal conductive fiber are broken, replace them with new fibers, model CNT-T03.

[0122] Aging Treatment: After two years of operation, if the sensor's accuracy decreases by more than 5%, the system prompts you to replace it through the interactive display panel 7, displaying "Sensor ID001 accuracy insufficient, replacement recommended." Replacement takes about 30 minutes and uses standard tools (screwdriver, adhesive).

[0123] Maintenance records: Maintenance data is stored in the database for feedback optimization subsystem to adjust thresholds (such as reducing the deformation threshold from 2 mm to 1.8 mm).

[0124] Abnormal situation handling mechanism:

[0125] To cope with abnormal situations such as sensor failure, robot path blockage or communication interruption, this device is designed with a multi-level error detection and recovery mechanism to ensure the robustness of the system in complex tunnel environments.

[0126] Sensor troubleshooting:

[0127] Detection: The data acquisition module checks the sensor output every 30 seconds. If five consecutive samples are invalid (for example, the phononic crystal sensor output is constant at 0 mm, the piezoelectric film voltage exceeds 5 volts, or the thermal fiber resistance exceeds 10 ohms), it is marked as a fault and the ID and timestamp are recorded.

[0128] Recovery: The system switches to a nearby sensor (such as a phononic crystal sensor within 2 meters) to collect additional data, or dispatches a patrol mobile frame 4 to cover the fault area (within a range of 5 meters). The recollected data is marked as temporary data, with a lower priority than normal data, and is transmitted to the deformation analysis subsystem.

[0129] Notice: Interactive display panel 7 displays "Sensor ID001 fault, x=100 meters, please check". The staff can confirm the repair or ignore it through the touch screen, which takes less than 10 seconds.

[0130] Robot path blockage processing:

[0131] Detection: The coordinated dispatching module of the patrol mobile frame 4 detects movement obstructions through the track encoder and laser rangefinder. If obstructed, the coordinates (e.g., x=150 meters, y=2 meters) are recorded and marked as blocked.

[0132] Notice: Interactive display panel 7 displays "Robot ID002 is blocked, x=150 meters, it is recommended to check for obstacles." The staff can manually adjust or confirm the automatic detour.

[0133] Communication interruption processing:

[0134] Detection: The data efficient transmission module 3 monitors the LoRa or ZigBee signal. If no confirmation signal is received three times in a row (timeout of 500 milliseconds), it is marked as interrupted and the node ID and time are recorded.

[0135] Recovery: Switch to the backup protocol (e.g., if LoRa is interrupted, switch to ZigBee, bandwidth 2 MHz, distance 100 meters). If it still fails, store the data in the local cache (10MB) and retransmit it after the signal is restored. The maximum cache size is 24 hours.

[0136] Notice: Interactive display panel 7 displays "Node ID003 communication interrupted, x=200 meters, please check the signal". The staff can check the node or switch to wired transmission (M12 interface).

[0137] Abnormal data processing:

[0138] If the deformation analysis subsystem detects data anomalies, it triggers the data acquisition subsystem to resample at a 10-second interval for three consecutive times. If the anomaly persists, it is marked as potentially high-risk and prioritized for inspection by the mobile framework 4, which takes less than 30 seconds.

[0139] Abnormal data is stored in the database with an “abnormal” label for analysis by the feedback optimization subsystem to adjust the threshold or sampling frequency.

[0140] Multi-robot collaboration constraints:

[0141] To ensure efficient coordination of the inspection mobile frame 4 in complex tunnel geometry, the dynamic adjustment subsystem is designed with detailed constraints and conflict avoidance algorithms to support the collaborative work of multiple robots in narrow or curvature-changing areas.

[0142] Constraints:

[0143] Tunnel geometry: Supports tunnel widths of 2-10 meters, curvature radius greater than 5 meters, and slopes less than 30 degrees. Robot spacing is maintained at >1 meter to prevent collisions, using real-time monitoring using track encoders and laser rangefinders.

[0144] Task Allocation: The task scheduling module assigns tasks based on deformation status reports, prioritizing high-risk areas. Each robot has a coverage range of 50-200 meters, an inspection speed of 0.1 meters per second, and a single task takes less than 20 minutes.

[0145] Environmental obstacles: Consider obstacles in the tunnel (such as cables and pipes). A laser rangefinder is used to detect the obstacle distance (0-5 meters). If the distance is less than 0.5 meters, path replanning is triggered.

[0146] Communication synchronization: Multi-robot time is synchronized using the NTP protocol (1 millisecond accuracy), ensuring that task time slices (every 10 seconds) do not overlap. Coordination commands are transmitted via the CAN bus (1MB / s) with a latency of <5 milliseconds.

[0147] Collision avoidance algorithm:

[0148] Algorithm Description: The collaborative scheduling module uses a distributed AI algorithm with a grid resolution of 0.5 meters. The goal is to minimize the total path length, with constraints such as collision avoidance and coverage of high-risk areas. The algorithm inputs include the robot's current position, target coordinates, and tunnel geometry (width, curvature).

[0149] step:

[0150] Initialization: Each robot is assigned a unique ID and records the starting coordinates (x = 100 meters, y = 2 meters) and the target area (x = 145-155 meters).

[0151] Collision detection: If the distance between the intersection points of the two robots' paths is less than 1 meter, it is marked as a collision and the time slice (such as 0 seconds, 10 seconds) is recorded.

[0152] Conflict resolution: Adjust the path or time slice of the lower-priority robot (based on task level), for example, by delaying it by 10 seconds or offsetting it by 0.5 meters. Replanning takes < 100 milliseconds.

[0153] Execution and Verification: Path instructions are issued, and the robot will feedback the actual path after execution. A deviation of >0.2 meters triggers replanning.

[0154] Performance: Tests show that five robots conducted collaborative inspections in a 1,000-meter tunnel with a 100% collision avoidance rate, a path deviation of <0.2 meters, and 100% mission coverage.

[0155] Feasibility verification logic of remedial measures:

[0156] To ensure the practicality and effectiveness of remedial measures, detailed verification logic and database update mechanism are designed for the remediation generation module and verification module of the risk assessment subsystem.

[0157] Validation logic:

[0158] Input: deformation distribution model (grid resolution 0.5 meters), thermal map (deformation > 2 mm is marked in red), historical data (deformation, stress, and temperature records for the past 30 days), and simulated environment (deformation 1-5 mm, stress 0-10 MPa, and temperature 0-50°C).

[0159] step:

[0160] Matching solutions: The remediation generation module queries an SQLite database (100 solutions) and matches measures based on deformation, stress, and temperature. For example, a deformation of 2.5 mm and a temperature of 42°C matches "spraying a 2 mm waterproof layer, priority 2."

[0161] Feasibility check: Query the inventory database to confirm material availability (e.g. waterproof coating > 10 liters). If insufficient, select the next best option (e.g. 1 mm waterproof layer), taking < 5 milliseconds.

[0162] Simulation Verification: The verification module simulates the construction results and uses finite element analysis to predict the deformation after reinforcement (e.g., reduction to <1 mm). Input simulation parameters: deformation 1-5 mm, stress 0-10 MPa, temperature 0-50°C, and a time of 1 second.

[0163] Scoring: Compare the actual and expected results. For example, if the required deformation reduction is 50% and the actual reduction is 40%, the score is 80%. If the score is less than 70%, choose a suboptimal solution or adjust the parameters.

[0164] Feedback: Generate a verification report, store it in the database, and feed it back to the deformation analysis subsystem to adjust the threshold.

[0165] Iterative process: If verification fails (score < 70%), the module re-queries the database, lowers the priority by one level, or adjusts the construction parameters. Iterations can be performed up to 3 times, taking < 3 seconds.

[0166] Database update mechanism:

[0167] Update frequency: The database is updated every 3 months or when new materials / solutions are introduced, and the staff inputs new solutions through the interactive display panel 7.

[0168] Entry criteria: New proposals must include triggering conditions, construction methods, material specifications, construction time, and priority. For example, "deformation > 3 mm, stress > 5 MPa, steel fiber reinforced concrete, thickness 20 cm, priority 1."

[0169] Verification: Before entering the new plan, it is verified through finite element simulation to confirm that the deformation reduction is >50% and the construction time is <4 hours. It is stored in the SQLite database and the capacity is increased by 10MB.

[0170] Backup: The database is backed up daily to a 1TB solid-state drive to prevent data loss, and recovery takes less than 1 minute.

[0171] User interaction error handling mechanism:

[0172] In order to improve the robustness of the user interaction subsystem, the interactive display panel 7 is designed with an error input detection and correction mechanism to ensure that the staff enters valid instructions.

[0173] Error Detection:

[0174] Input range check: The interactive control module verifies user input parameters. For example, the inspection frequency range is 1-60 seconds, and the waterproof layer thickness is 0.5-5 mm. If the input frequency is 0.5 seconds or the thickness is 10 mm, it will be marked as invalid. This takes less than 1 millisecond.

[0175] Conflict detection: Checks for conflicts between input and system status, such as a 5-second inspection frequency but insufficient bandwidth (>2MB / s), or construction parameters exceeding inventory.

[0176] Repeated input: If the same invalid command (such as "frequency 0.5 seconds") is entered three times in a row, it will be marked as a repeated error and the timestamp and user ID will be recorded.

[0177] Corrective Mechanism:

[0178] Tips and Suggestions: Interactive Display Panel 7 displays error prompts, such as "Frequency 0.5 seconds is invalid, please enter 1-60 seconds" or "Insufficient waterproof coating, 1 mm thickness is recommended." The prompts contain suggested values ​​based on database recommendations.

[0179] Automatic correction: If the user does not make corrections, the system automatically selects the most recent valid value. For example, a frequency of 0.5 seconds is corrected to 1 second, and a thickness of 10 mm is corrected to 5 mm, which takes less than 2 milliseconds.

[0180] Recording and feedback: Incorrect input is stored in a database (2MB capacity) with a timestamp and correction results, and fed back to the feedback optimization subsystem to optimize the interaction logic.

[0181] Performance: Tests show that error detection accuracy is 98%, correction takes less than 5 milliseconds, and user interaction response time is less than 50 milliseconds, ensuring rapid decision-making. The interactive interface supports multilingual prompts to meet the needs of different staff members.

[0182] Extreme environment adaptability parameters:

[0183] To support tunnel types not specified in the claims and more extreme environments, this device specifies the extreme working conditions of the hardware and system to ensure its applicability in different geological conditions and environments.

[0184] Environmental parameter range:

[0185] Temperature: -20°C to 60°C, covering both extreme cold (such as high-latitude railway tunnels) and high temperatures (such as desert highway tunnels). The phononic crystal sensor (PC-S01) has a temperature resistance of -10°C to 60°C, and the thermal fiber (CNT-T03) has a temperature resistance of 0°C to 50°C. The heating module automatically activates at low temperatures (power consumption 2 watts, heating to 0°C takes 5 minutes).

[0186] Humidity: 0-95%, suitable for submarine tunnels (humidity > 90%) and arid tunnels (humidity < 10%). All hardware (sensors, robots, and data transmission modules) are coated with an anti-corrosion coating (polyurethane, 0.1 mm thick) with an IP67 / IP68 protection grade to prevent moisture erosion.

[0187] Water pressure: 0-1 MPa, adapted to the high water pressure environment of submarine tunnels. The stainless steel frame (compressive strength 200 MPa) and expansion bolts (tensile force 10 kN) of the deformation sensing platform (2) ensure structural stability.

[0188] Geological conditions: Supports soft soil, rock, and mixed strata. Sensor embedment depth (20 cm for phononic crystals, 5 cm for thermal fiber) adapts to different lining materials (concrete, reinforced concrete).

[0189] Tunnel length: supports 100-5000 meters. The data efficient transmission module 3 covers 2 kilometers through the LoRa protocol. If necessary, add relay nodes (1 every 2 kilometers, model AL-DT01).

[0190] Robot adaptability:

[0191] Wall Climbing Capability: The inspection mobile frame 4 supports slopes < 45 degrees, curvature radii > 3 meters, and magnetic attraction > 20N, suitable for reinforced concrete or steel linings. The rubber non-slip tracks (5 cm wide, with a wear-resistant coating) support rough surfaces (friction coefficient > 0.5).

[0192] Battery Life: Each robot (CR-B02) is equipped with a 5000mAh, 12V lithium-ion battery with a battery life of 8 hours and a range of 1000 meters. A charging station is located in the control room and charging time is 2 hours.

[0193] Performance Verification:

[0194] Test environment: simulated tunnel (length 2000 meters, humidity 95%, temperature -10℃, water pressure 0.8 MPa), deployed with 10 deformation sensing platforms, 5 robots, and 10 data efficient transmission modules.

[0195] Results: Deformation detection accuracy was 97.5% (calibrated value 2 mm, measured value 2.05 mm), response time was 0.06 seconds, remediation effectiveness was 90% (deformation reduced to 0.9 mm), path deviation was 0.15 meters, and communication success rate was 99.8% (LoRa signal attenuation <1%).

[0196] Conclusion: The device operates stably in extreme environments and meets the needs of submarine, cold or long-distance tunnels.

[0197] Attachment Figure 4 This is a thermal simulation diagram of the deformation distribution of the system used in this invention. The horizontal axis represents tunnel location (meters), ranging from 0 to 1000 meters in 5-meter steps, corresponding to 200 monitoring points. The red asterisk ("snowflake") in the middle marks high-risk areas where deformation values ​​exceed a threshold (2 mm).

[0198] The vertical axis represents the monitoring layer. Only one level is displayed, which is simplified to one-dimensional monitoring and actually corresponds to the deformation distribution of the tunnel wall.

[0199] Content: The heatmap uses colors (from blue to red) to represent the deformation value (in millimeters) after fusion. Red areas indicate higher deformation and blue indicates lower deformation. Red asterisks mark high_risk_idx high-risk areas with deformation exceeding 2 mm.

[0200] Conclusion: The figure shows that deformations in some areas of the tunnel (e.g., around x = 200-300 meters) are significantly higher than 2 mm, indicating the presence of high-risk areas that may require further inspection or reinforcement. The overall deformation distribution is uneven, with some areas exhibiting lower deformations, indicating comprehensive monitoring coverage.

[0201] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further restrictions, an element defined by the statement "comprising a reference structure" does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0202] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A real-time automatic monitoring system for tunnel deformation, including a monitoring and processing system, characterized by: The monitoring and processing system includes a data acquisition subsystem, an auxiliary acquisition subsystem, a deformation analysis subsystem, a risk assessment subsystem, a feedback optimization subsystem, a dynamic adjustment subsystem and a user interaction subsystem; A tunnel deformation real-time automatic monitoring device corresponding to the tunnel deformation real-time automatic monitoring system comprises a monitoring hybrid device (1), a deformation sensing platform (2), a data efficient transmission module (3), an inspection mobile frame (4), an analysis and processing unit (5), a dynamic control module (6) and an interactive display panel (7), wherein a waterproof communication interface (8) is provided on one side of the data efficient transmission module (3); The data acquisition subsystem includes a data acquisition module, a data preprocessing module and a data transmission module; the auxiliary acquisition subsystem includes a deformation measurement module and a data processing module; the deformation analysis subsystem includes a feature extraction module, a data fusion module and a state analysis module; the risk assessment subsystem includes a task scheduling module, a deformation assessment module, a strategy generation module and a remedy generation module; the feedback optimization subsystem includes a feedback analysis module; the dynamic adjustment subsystem includes an instruction parsing module, a drive control module and a collaborative scheduling module; the user interaction subsystem includes a visualization module and an interactive control module; the monitoring and processing system collects deformation, stress and temperature data through the data acquisition subsystem and transmits it to the deformation analysis subsystem; the deformation analysis subsystem generates a deformation status report and pushes it to the risk assessment subsystem; the risk assessment subsystem generates an adjustment strategy and remedial measures, the adjustment strategy is transmitted to the dynamic adjustment subsystem, and the remedial measures are transmitted to the user interaction subsystem; the user interaction subsystem displays and transmits the remedial measures to the staff, and the feedback optimization subsystem generates optimization instructions based on user instructions and adjustment results.

2. The real-time automatic monitoring system for tunnel deformation according to claim 1, characterized in that: The monitoring hybrid device (1) is a comprehensive monitoring module integrating a deformation sensing platform (2), a data efficient transmission module (3) and an inspection mobile frame (4), adopts a modular design, and has a power supply voltage of 24VDC, and is used to realize real-time collection and dynamic inspection of deformation, stress and temperature data in the tunnel; the deformation sensing platform (2) is composed of a stainless steel frame and a sensor array, the stainless steel frame is a rectangular structure, the surface is coated with a waterproof coating, and is fixed to the inner wall of the tunnel lining by M10 expansion bolts to support the sensor array, and the operating temperature range is -20°C to 60°C, the sensor array includes phononic crystal sensors, flexible piezoelectric films, and thermal conductivity fibers. The phononic crystal sensors are pre-embedded inside the tunnel lining, arranged every 2 meters along the tunnel circumference, to collect submillimeter deformation data. The flexible piezoelectric film is made of polyvinylidene fluoride and covered with epoxy resin on the inner wall surface of the tunnel. It is laid along the tunnel axis every 3 meters to collect distributed stress data. The thermal conductivity fibers are made of carbon nanotubes and embedded in the joints of the tunnel lining, distributed every 5 meters along the tunnel axis, to collect deformation and temperature coupled data.

3. The real-time automatic monitoring system for tunnel deformation according to claim 1, characterized in that: The top of the deformation sensing platform (2) is connected to the data efficient transmission module (3) through a snap-fit; the data efficient transmission module (3) consists of a corrosion-resistant aluminum alloy shell and a wireless communication module, the corrosion-resistant aluminum alloy shell is arc-shaped, the surface is coated with an anti-corrosion coating, and is fixed to the top arch structure of the tunnel by M8 stainless steel bolts, and a waterproof communication interface is set on the side, connecting the deformation sensing platform (2) and the inspection mobile frame (4) through a corrosion-resistant wire, and the wireless communication module includes a low-power wide-area wireless network transmitter and a short-range wireless communication transmitter, which are used to transmit standardized deformation data and compressed deformation data to the analysis and processing unit (5).

4. The real-time automatic monitoring system for tunnel deformation according to claim 1, characterized in that: The inspection mobile frame (4) is composed of an aluminum alloy body, a rubber anti-skid track, a laser rangefinder, an electro-permanent magnetic suction cup and a climbing arm robot. The laser rangefinder is fixed to the top of the climbing arm robot by an M5 bolt to collect tunnel surface deformation data. The electro-permanent magnetic suction cup is a magnetic foot type magnetic adsorption, which uses an electromagnet to realize magnetic force switching through an electric control switch. The adsorption force is fed back in real time through a Hall sensor with a control accuracy of ±0.5N. The magnetic adsorption force is achieved through the magnetic interaction between the electromagnet and the reinforced concrete on the inner wall of the tunnel. It moves along the axial direction of the tunnel and is used to supplement the blind area of ​​the fixed sensor. The analysis and processing unit (5) is composed of a high-performance embedded computer, which is installed in the tunnel entrance control room via a steel fixed bracket, and is used to process multi-source data and generate deformation status reports and remedial measures; the dynamic control module (6) is composed of a microcontroller and a drive circuit, which is installed in the tunnel entrance control room via a steel fixed bracket, and is used to drive the dynamic adjustment of the inspection mobile frame (4); the interactive display panel (7) is composed of a touch screen and a display, which is fixed to the wall of the control room via a wall-mounted bracket, and is used to display three-dimensional deformation mapping and transmit remedial measures to staff.

5. The real-time automatic monitoring system for tunnel deformation according to claim 1, characterized in that: The data acquisition subsystem acquires data through the deformation sensing platform (2), specifically including: the data acquisition module receives the original data of the phononic crystal sensor, the flexible piezoelectric film and the thermal conductive fiber, and generates deformation, stress and temperature data; the data preprocessing module uses wavelet transform to remove noise, and the unified format is JSON to generate standardized deformation data; the data transmission module transmits the data to the deformation analysis subsystem through a low-power wide-area wireless network; the auxiliary acquisition subsystem generates compressed deformation data through the laser rangefinder of the inspection mobile frame (4), and transmits the compressed deformation data to the deformation analysis subsystem through a short-range wireless communication protocol.

6. The real-time automatic monitoring system for tunnel deformation according to claim 1, characterized in that: The deformation analysis subsystem processes multi-source data through an analysis and processing unit (5), specifically including: a feature extraction module receives standardized deformation data and compression deformation data, extracts deformation rate, stress distribution and temperature change characteristics using a time-frequency analysis method, and generates a deformation feature set; a data fusion module fuses the deformation feature set using a weighted average method to generate a comprehensive deformation feature vector; a state analysis module evaluates deformation trends using a time series analysis method, detects high-risk areas using a threshold comparison method, generates a deformation state report, stores it in a local database, and transmits it to the risk assessment subsystem.

7. The real-time automatic monitoring system for tunnel deformation according to claim 1, characterized in that: The risk assessment subsystem performs deformation assessment and task optimization through the analysis and processing unit (5), specifically including: the task scheduling module receives the deformation status report and adjustment feedback data, and uses the priority scheduling method to generate a task allocation table for the inspection mobile frame (4); the deformation assessment module identifies the deformation area through the geometric modeling method, tracks the deformation trend through the motion prediction method, constructs the deformation distribution model, and generates the deformation distribution heat map; the remedial generation module generates remedial measures based on the deformation distribution model and the heat map, combined with the preset reinforcement solution database, and transmits them to the user interaction subsystem.

8. The real-time automatic monitoring system for tunnel deformation according to claim 7, characterized in that: The risk assessment subsystem generates an adjustment strategy through the analysis and processing unit (5), specifically including: a strategy generation module receives a task allocation table and a deformation distribution heat map, adjusts the monitoring range through a fuzzy control method, optimizes the path and posture of the inspection mobile frame (4) through a linear programming method, generates an adjustment strategy, and transmits it to the dynamic adjustment subsystem; the remedial generation module generates remedial measures and transmits them to the staff through an interactive display panel (7); the verification module uses historical deformation data and simulated environment data to verify the strategy and remedial measures, generates a verification report, and feeds it back to the deformation analysis subsystem and the feedback optimization subsystem.

9. The real-time automatic monitoring system for tunnel deformation according to claim 1, characterized in that: The dynamic adjustment subsystem realizes dynamic adjustment of the inspection mobile frame (4) through the dynamic control module (6), specifically comprising: the instruction parsing module parsing the adjustment strategy into path offset and posture adjustment parameters; the drive control module driving the inspection mobile frame (4) through the stepper motor to generate an adjustment execution result; the collaborative scheduling module coordinates multiple groups of inspection mobile frames (4) through a time synchronization method to generate a final adjustment execution result, which is fed back to the risk assessment subsystem.

Images