A tunnel security management method and system based on big data

Abstract

This invention provides a tunnel safety management method and system based on big data, belonging to the field of tunnel safety management technology. This invention lays the foundation for precise management by dividing sections into multiple data segments and labeling them; it focuses on key information through multi-source data collection and key parameter screening; it performs big data analysis and processing through edge computing and cloud collaboration, balancing real-time and long-term analysis; it dynamically adjusts parameter warning thresholds to match actual operating conditions; and it employs a three-level response mechanism to respond promptly and effectively to different risks. Overall, it achieves precise, intelligent, and efficient tunnel safety management, effectively improving tunnel safety assurance levels and reducing the risk of accidents.

Description

Technical Field

[0001] This invention relates to the field of tunnel safety management technology, and in particular to a tunnel safety management method and system based on big data. Background Technology

[0002] With the continuous advancement of transportation infrastructure construction, tunnels, as key nodes in transportation networks such as highways and railways, are experiencing a sustained increase in both number and scale. However, tunnel safety management faces numerous complex and challenging issues, including: During tunnel operation, the impact of complex geological conditions cannot be ignored. Problems such as surrounding rock deformation, lining cracks, and surface settlement gradually emerge over time. If these issues are not detected and addressed promptly, they will seriously affect the structural safety and service life of the tunnel. Moreover, the geological conditions of tunnels in different geographical locations can vary greatly. Some areas have hard and stable rocks, while others have loose soil that is prone to settlement. Historical traffic flow and accident records also differ between different sections. Some sections may experience frequent accidents due to their location as transportation hubs, curves, or steep gradients. Due to the lack of detailed segmentation and labeling based on multi-dimensional data such as geographic information, geological survey reports, historical traffic flow statistics, and accident records, managers struggle to have a clear and accurate understanding of the characteristics and risk status of each section. This makes it difficult to formulate management strategies and safety measures that are tailored to the actual needs of different sections, resulting in a significant reduction in the effectiveness of safety management.

[0003] Tunnel safety monitoring involves various types of data. Traditional data acquisition methods are often fragmented, with different types of data collected by different sensors or systems, lacking unified coordination and management. The collected data also lacks effective filtering and integration, with a large amount of redundant and non-critical information mixed in, making it difficult to highlight key safety information. For example, within a large amount of normal equipment operation data, there may be early signs of impending equipment failure. However, due to the massive amount of data and the lack of effective filtering mechanisms, this critical information is easily overlooked, thus failing to detect potential safety hazards in a timely manner and increasing the risk of accidents.

[0004] Existing tunnel safety early warning systems typically use fixed warning thresholds, which are set during the initial stages of tunnel construction or based on past experience and are rarely adjusted once set. However, the actual operating conditions of a tunnel are constantly changing, influenced by various factors such as seasonal variations, increases and decreases in traffic flow, and equipment aging. Fixed warning thresholds may fail to detect abnormal equipment conditions in a timely manner, thus delaying maintenance and increasing the probability of safety accidents.

[0005] Therefore, it is necessary to provide a tunnel safety management method and system based on big data to solve the above-mentioned technical problems. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a tunnel safety management method and system based on big data to solve the problems of lack of precision in existing tunnel management technologies, low efficiency in data collection and processing, easy obscuration of key information, fixed early warning thresholds, and inability to adapt to actual operational changes.

[0007] This invention provides a tunnel safety management method based on big data, comprising the following steps: S1. The tunnel to be monitored is divided into sections and labeled according to geographical information, geological survey reports, historical traffic flow statistics and accident records. Each section is assigned a unique identifier. S2. Through the deployed sensor network, data on the equipment operating status and environmental parameters of each section of the tunnel during the current period are collected and compiled into a multi-source dataset. S3. Using the multi-source dataset collected in the previous tunnel monitoring cycle, the parameters that continuously deviate and whose deviation exceeds the deviation threshold are selected through a preset deviation threshold, and these parameters are organized into the key parameters in the current tunnel monitoring cycle. S4. Through locally deployed edge computing nodes, key parameters are preprocessed and organized in real time, and non-key parameters are uploaded to the cloud for phased preprocessing and organization. S5. Integrate environmental conditions, traffic flow, and equipment status of each section of the tunnel, and dynamically adjust the parameter warning thresholds of each section. S6. Compare the sorted key and non-key parameters with the dynamically adjusted parameter warning thresholds for each segment, and execute a three-level response, which includes a normal response, a first-level warning, and a second-level warning.

[0008] Preferably, step S1 specifically includes: S101. Obtain the geographic information, geological survey report, historical traffic flow statistics and accident record archives of the tunnel to be monitored, forming four types of data; S102. Map the four types of data to the linear spatial coordinate system of the tunnel, using the station number as the reference axis, to form a multi-dimensional feature vector sequence along the tunnel length direction. S103. Using a sliding window to identify feature abrupt change points, determine the natural functional boundary points of the tunnel, and divide the tunnel into several sections based on the multi-dimensional feature vector sequence along the tunnel length direction. S104. Generate a globally unique segment identifier for each segment.

[0009] Preferably, step S2 specifically includes: S201. Deploy a sensor network encompassing the environment, equipment, and traffic within each section; S202. According to the preset monitoring cycle, the raw sensor data of each section is collected synchronously, and the raw data from different sensor types are aligned according to the section identifier, the timestamp is synchronized and the format is standardized, and integrated into a unified multi-source dataset.

[0010] Preferably, step S3 specifically includes: S301. Retrieve the multi-source dataset from the previous period and perform a time-series comparison with the data from the current period. Calculate the deviation of each parameter over multiple consecutive periods. The deviation includes the mean offset rate or variance growth. S302. Based on the preset deviation threshold, filter out the parameter items that meet the condition of continuous deviation and deviation exceeding the deviation threshold, mark them as key parameters, and classify the rest as non-key parameters according to their respective sections.

[0011] Preferably, step S4 specifically includes: S401. In the edge computing nodes deployed in each section, key parameters are preprocessed and organized in real time, including real-time cleaning, noise reduction, and feature extraction. S402. Non-critical parameters are uploaded to the cloud platform via the communication network, and the cloud performs batch phased processing and organization on a periodic basis, including aggregation and statistics. S403: Preprocessed key parameters are retained at the edge nodes for rapid response, while the results of processing non-key parameters are stored in the cloud database for long-term analysis.

[0012] Preferably, step S5 specifically includes: S501. Integrate the real-time environmental conditions, traffic flow, and equipment status of each section to construct a section operation status profile; S502. Based on the constructed section operation status profile, a machine learning model is used to dynamically adjust the warning thresholds of various parameters under each section. The warning thresholds include normal warning thresholds, first-level warning thresholds, and second-level warning thresholds. S503. The updated dynamic early warning thresholds are stored by segment identifier and synchronized to edge nodes and the cloud.

[0013] Preferably, step S6 specifically includes: S601. Compare the sorted key parameters and non-key parameters with the dynamically adjusted warning thresholds of their respective sections to obtain the comparison results. S602. In the comparison results, if all parameters are within the normal warning threshold range, a normal response is triggered; if any key parameter exceeds the first-level warning threshold but does not reach the second-level warning threshold, a first-level warning is triggered; if any key parameter reaches or exceeds the second-level warning threshold, a second-level warning is triggered, and a warning report is generated for notification.

[0014] A tunnel safety management system based on big data includes: The section identification module is used to divide the tunnel to be monitored into sections and label them according to geographical information, geological survey reports, historical traffic flow statistics and accident records. Each section is assigned a unique identifier. The multi-source acquisition module is used to collect data on the equipment operating status and environmental parameters of each section of the tunnel during the current period through the deployed sensor network, and organize them into a multi-source dataset. The parameter filtering module is used to use the multi-source dataset collected in the previous tunnel monitoring cycle to filter out parameters that continuously deviate and whose deviation exceeds the deviation threshold through a preset deviation threshold, and organize them into key parameters in the current tunnel monitoring cycle. The segmented processing module is used to preprocess and organize key parameters in real time through locally deployed edge computing nodes, and upload non-key parameters to the cloud for phased preprocessing and organization. The threshold adjustment module is used to dynamically adjust the parameter warning thresholds of each section by integrating the environmental conditions, traffic flow, and equipment status of each section of the tunnel. The early warning response module is used to compare the sorted key and non-key parameters with the dynamically adjusted early warning thresholds for each segment and execute a three-level response, which includes a normal response, a first-level early warning, and a second-level early warning.

[0015] Compared with related technologies, the tunnel safety management method and system based on big data provided by this invention has the following beneficial effects: This invention lays the foundation for precise management by dividing and labeling sections using multi-dimensional data; it focuses on key information through multi-source data collection and key parameter screening; it performs big data analysis and processing through edge computing and cloud collaboration, taking into account both real-time and long-term analysis; it dynamically adjusts parameter warning thresholds to match actual operating conditions; and it employs a three-level response mechanism to respond to different risks in a timely and effective manner. Overall, it achieves precise, intelligent, and efficient tunnel safety management, effectively improving the level of tunnel safety assurance and reducing the risk of accidents. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a tunnel safety management method based on big data according to the present invention. Figure 2This is a schematic diagram of the system modules of a tunnel safety management system based on big data according to the present invention. Detailed Implementation

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0018] Example 1 like Figure 1 As shown, a tunnel safety management method based on big data includes the following steps: S1. The tunnel to be monitored is divided into sections and labeled according to geographical information, geological survey reports, historical traffic flow statistics and accident records. Each section is assigned a unique identifier. S2. Through the deployed sensor network, data on the equipment operating status and environmental parameters of each section of the tunnel during the current period are collected and compiled into a multi-source dataset. S3. Using the multi-source dataset collected in the previous tunnel monitoring cycle, the parameters that continuously deviate and whose deviation exceeds the deviation threshold are selected through a preset deviation threshold, and these parameters are organized into the key parameters in the current tunnel monitoring cycle. S4. Through locally deployed edge computing nodes, key parameters are preprocessed and organized in real time, and non-key parameters are uploaded to the cloud for phased preprocessing and organization. S5. Integrate environmental conditions, traffic flow, and equipment status of each section of the tunnel, and dynamically adjust the parameter warning thresholds of each section. S6. Compare the sorted key and non-key parameters with the dynamically adjusted parameter warning thresholds for each segment, and execute a three-level response, which includes a normal response, a first-level warning, and a second-level warning.

[0019] In the specific implementation process, step S1 specifically includes: S101. Obtain the geographic information, geological survey report, historical traffic flow statistics and accident record archives of the tunnel to be monitored, forming four types of data.

[0020] Specifically, geographic information can be obtained from Geographic Information System (GIS) databases, map-making agencies, or relevant government departments, including the tunnel's geographical location and surrounding topography; geological survey reports are generated by professional geological survey teams after conducting surveys in the early stages of tunnel construction or during operation, covering information such as the geological structure and soil properties of the tunnel area; historical traffic flow statistics can be obtained from traffic management departments and tunnel operating units, recording the number of vehicles passing through the tunnel and the distribution of vehicle types in different time periods; and accident record files come from the daily records of the tunnel management department, containing information such as the time, location, type, and cause of various accidents that have occurred in the tunnel.

[0021] S102. Map the four types of data to the linear spatial coordinate system of the tunnel, using the station number as the reference axis, to form a multi-dimensional feature vector sequence along the tunnel length.

[0022] Specifically, the process begins by establishing a linear spatial coordinate system for the tunnel, using the station numbers as the reference axis. These station numbers are typically assigned sequentially at certain intervals, starting from the tunnel entrance. Then, the collected four types of data are mapped to corresponding station locations based on their correlation with the tunnel's location. For example, the topographic features surrounding the tunnel in the geographic information are mapped to station numbers according to their actual location; the location information of different geological layers in the geological survey report is also mapped to corresponding station numbers; and historical traffic flow data and accident records are mapped based on their specific locations (which can be determined within the tunnel's station range using relevant location information). Finally, the four types of data mapped to each station number are integrated to form a multi-dimensional feature vector sequence. Each vector contains information on geography, geology, traffic flow, and accidents at that station number.

[0023] S103. Using a sliding window to identify feature mutation points, determine the natural functional boundary points of the tunnel, and divide the tunnel into several sections based on the multi-dimensional feature vector sequence along the tunnel length direction.

[0024] Specifically, by setting an appropriate window size and sliding step size, the system slides across a multidimensional feature vector sequence. During the sliding process, the changes in the data characteristics within the window are analyzed. When abrupt changes in certain features (such as geological type, traffic flow patterns, etc.) are detected, the location is recorded as a feature abrupt change point. These feature abrupt change points often reflect the natural functional boundaries of the tunnel, such as places where there are significant changes in geological structure or areas with significantly different traffic flow patterns. Based on these feature abrupt change points, the entire tunnel is divided into several relatively independent sections.

[0025] In this embodiment, a multidimensional feature vector sequence for a mountain highway tunnel is used, with a sliding window size of 200 meters and a sliding step size of 50 meters. During the sliding process, it was found that at K0+600, the geological type changed from sandstone to limestone, and the traffic flow pattern also changed. Previously, the traffic flow was relatively stable, but thereafter, the traffic flow fluctuated significantly during peak hours. This location was identified as a feature abrupt change point. Similarly, several feature abrupt change points were identified at other locations in the tunnel, such as at K0+1200, where changes in the surrounding terrain led to changes in traffic flow characteristics. Based on these feature abrupt change points, the tunnel was divided into several sections, such as from K0+000 to K0+600 and from K0+600 to K0+1200.

[0026] S104. Generate a globally unique segment identifier for each segment.

[0027] In the specific implementation process, step S2 specifically includes: S201. Deploy a sensor network that includes environmental, equipment, and traffic sensors within each section.

[0028] Specifically, environmental sensors include sensors for oxygen content, harmful gas concentrations, etc. Equipment sensors include sensors installed on lighting equipment, ventilation equipment, and fire-fighting equipment. For example, current and voltage sensors installed on lighting equipment monitor whether the current and voltage are normal during operation; wind speed and vibration sensors installed on ventilation equipment detect whether the wind speed meets the standard and whether the equipment is operating smoothly. Traffic sensors include vehicle detectors installed at the entrance to count the number and type of vehicles entering the tunnel; and radar speed detectors installed inside the section to monitor vehicle speed.

[0029] S202. According to the preset monitoring cycle, the raw sensor data of each section is collected synchronously, and the raw data from different sensor types are aligned according to the section identifier, the timestamp is synchronized and the format is standardized, and integrated into a unified multi-source dataset.

[0030] In the specific implementation process, step S3 specifically includes: S301. Retrieve the multi-source dataset from the previous period and perform a time-series comparison with the data from the current period. Calculate the deviation of each parameter over multiple consecutive periods, where the deviation includes the mean offset rate or variance growth.

[0031] Specifically, the multi-source dataset collected from the previous tunnel monitoring cycle is accurately retrieved from the stored database. This dataset contains information on the operating status of equipment and environmental parameters in various tunnel sections. The data from the previous and current cycles are arranged and compared chronologically. For each parameter, its corresponding time value within the two cycles is identified. For example, for the temperature parameter of a certain tunnel section, the temperature values ​​at the same time in the previous and current cycles are compared. For each parameter, the average value over multiple consecutive cycles in the previous and current cycles is calculated. For example, to calculate the mean deviation rate of a parameter over 5 consecutive cycles, the average value A of the parameter over the 5 cycles in the previous cycle is calculated first, and then the average value B of the parameter over the 5 cycles in the current cycle is calculated. The mean deviation rate is then calculated using the formula: mean deviation rate = (BA) / A × 100%. Variance reflects the dispersion of the data. Let C be the variance of the previous cycle and D be the variance of the current cycle. The formula for calculating variance growth is: variance growth = (DC) / C × 100%.

[0032] S302. Based on the preset deviation threshold, filter out the parameter items that meet the condition of continuous deviation and deviation exceeding the deviation threshold, mark them as key parameters, and classify the rest as non-key parameters according to their respective sections.

[0033] Specifically, based on the actual needs of tunnel safety management and historical experience data, reasonable deviation thresholds are preset for each parameter. In this embodiment, for the temperature parameter, the threshold for mean deviation rate is set to 15%, and the threshold for variance growth is set to 30%; for the wind speed parameter, the threshold for mean deviation rate is set to 10%, and the threshold for variance growth is set to 20%, etc. The mean deviation rate and variance growth of each parameter calculated in step S301 are compared with the preset deviation thresholds. If a parameter continuously deviates over multiple consecutive periods, and its mean deviation rate or variance growth exceeds the corresponding preset threshold, then the parameter is filtered out. For example, the mean deviation rate of the humidity parameter in a certain section is 12%, 16%, and 18% over three consecutive periods, respectively. The preset mean deviation rate threshold is 15%. Since the mean deviation rate of the latter two periods exceeds the threshold and the deviation is continuous, the humidity parameter is filtered out as a key parameter.

[0034] The selected parameters are marked as key parameters, and the remaining parameters that do not meet the criteria are marked as non-key parameters. This can be achieved by adding a label field to the dataset. Based on the tunnel segment to which each parameter belongs, key and non-key parameters are categorized separately, with key parameters from the same segment grouped together and non-key parameters grouped together.

[0035] In the specific implementation process, step S4 specifically includes: S401. In the edge computing nodes deployed in the nearest segment, the key parameters are preprocessed and organized in real time, including real-time cleaning, noise reduction and feature extraction.

[0036] Specifically, edge computing nodes are directly connected to the sensor network deployed in various sections of the tunnel, enabling them to receive key parameter data collected by the sensors in real time. Through these edge computing nodes, the received key parameter data undergoes preliminary checks, removing obviously erroneous or invalid data. Existing denoising algorithms, such as moving average filtering, are then used to smooth the key parameter data. Finally, feature information is extracted from the cleaned and denoised key parameter data. For example, for equipment operating status data, features such as equipment runtime and number of starts can be extracted; for humidity data in environmental parameters, features such as the rate of change of humidity and humidity peak values ​​can be extracted.

[0037] S402. Non-critical parameters are uploaded to the cloud platform via the communication network, and the cloud performs batch phased processing and organization on a periodic basis, including aggregation and statistics.

[0038] Specifically, through the cloud platform, non-critical parameter data stored in the temporary buffer is processed in batches according to a preset time period, such as hourly or daily: first, the data is aggregated and statistically analyzed, such as calculating the average temperature, maximum humidity, and total operating time of the equipment in a certain segment within a certain time period.

[0039] S403: Preprocessed key parameters are retained at the edge nodes for rapid response, while the results of processing non-key parameters are stored in the cloud database for long-term analysis.

[0040] Specifically, key parameter data, after real-time preprocessing at edge computing nodes, is directly stored on the local storage devices of the edge nodes. Non-key parameter data, after batch processing and aggregation statistics, is stored in a cloud database. The database is organized according to specific classification and indexing rules, facilitating long-term data querying and analysis by administrators.

[0041] In the specific implementation process, step S5 specifically includes: S501. Integrate the real-time environmental conditions, traffic flow, and equipment status of each section to construct a profile of the section's operational status.

[0042] Specifically, environmental conditions, traffic flow, and equipment status information are obtained from multi-source datasets. Data fusion technology is used to comprehensively analyze the pre-processed data and construct a profile of the section's operational status. For example, existing methods such as weighted averaging and fuzzy comprehensive evaluation are employed to assign corresponding weights to different parameters based on their importance to tunnel operational safety, calculating a comprehensive operational status index for each section. For instance, in ventilation equipment sections, parameters such as temperature, humidity, harmful gas concentration, and ventilation equipment wind speed are comprehensively considered, and a weighted calculation is used to obtain the ventilation operational status index for that section, intuitively reflecting the operational status of the ventilation system in that section.

[0043] S502. Based on the constructed section operation status profile, a machine learning model is used to dynamically adjust the warning thresholds of various parameters under each section. The warning thresholds include normal warning thresholds, first-level warning thresholds, and second-level warning thresholds.

[0044] Specifically, a suitable machine learning model is selected, such as a decision tree, support vector machine, or neural network. In this embodiment, a neural network model is used for training. First, historical data is collected, including parameter values ​​for different sections under different operational states and corresponding warning levels (normal, level 1 warning, level 2 warning). This data is used as a training set to train the selected neural network model, enabling the model to predict the warning thresholds for various parameters under different operational states. The constructed section operational state profile is then input into the trained neural network model, and the output is the predicted warning thresholds for various parameters under each section.

[0045] S503. The updated dynamic early warning thresholds are stored by segment identifier and synchronized to edge nodes and the cloud.

[0046] In the specific implementation process, step S6 specifically includes: S601. Compare the sorted key parameters and non-key parameters with the dynamically adjusted warning thresholds of their respective sections to obtain the comparison results.

[0047] Specifically, in this embodiment, a tunnel is divided into two sections, A and B. In section A, after dynamic adjustment, the normal warning threshold for temperature is 20-25℃, the first-level warning threshold is 25-30℃, and the second-level warning threshold is above 30℃; the normal warning threshold for humidity is 40%-60%, the first-level warning threshold is 60%-70%, and the second-level warning threshold is above 70%. At a certain moment, the temperature in section A is 28℃ and the humidity is 65%. Comparing the temperature of 28℃ with the warning threshold for temperature in section A, it is found to be within the first-level warning threshold range; comparing the humidity of 65% with the warning threshold for humidity in section A, it is also within the first-level warning threshold range.

[0048] S602. In the comparison results, if all parameters are within the normal warning threshold range, a normal response is triggered; if any key parameter exceeds the first-level warning threshold but does not reach the second-level warning threshold, a first-level warning is triggered; if any key parameter reaches or exceeds the second-level warning threshold, a second-level warning is triggered, and a warning report is generated for notification.

[0049] Example 2 like Figure 2 As shown, a tunnel safety management system based on big data includes: The section identification module is used to divide the tunnel to be monitored into sections and label them according to geographical information, geological survey reports, historical traffic flow statistics and accident records. Each section is assigned a unique identifier. The multi-source acquisition module is used to collect data on the equipment operating status and environmental parameters of each section of the tunnel during the current period through the deployed sensor network, and organize them into a multi-source dataset. The parameter filtering module is used to use the multi-source dataset collected in the previous tunnel monitoring cycle to filter out parameters that continuously deviate and whose deviation exceeds the deviation threshold through a preset deviation threshold, and organize them into key parameters in the current tunnel monitoring cycle. The segmented processing module is used to preprocess and organize key parameters in real time through locally deployed edge computing nodes, and upload non-key parameters to the cloud for phased preprocessing and organization. The threshold adjustment module is used to dynamically adjust the parameter warning thresholds of each section by integrating the environmental conditions, traffic flow, and equipment status of each section of the tunnel. The early warning response module is used to compare the sorted key and non-key parameters with the dynamically adjusted early warning thresholds for each segment and execute a three-level response, which includes a normal response, a first-level early warning, and a second-level early warning.

[0050] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0051] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0052] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

Claims

1. A tunnel safety management method based on big data, characterized in that, Includes the following steps: S1. The tunnel to be monitored is divided into sections and labeled according to geographical information, geological survey reports, historical traffic flow statistics and accident records. Each section is assigned a unique identifier. S2. Through the deployed sensor network, data on the equipment operating status and environmental parameters of each section of the tunnel during the current period are collected and compiled into a multi-source dataset. S3. Using the multi-source dataset collected in the previous tunnel monitoring cycle, the parameters that continuously deviate and whose deviation exceeds the deviation threshold are selected through a preset deviation threshold, and these parameters are organized into the key parameters in the current tunnel monitoring cycle. S4. Through locally deployed edge computing nodes, key parameters are preprocessed and organized in real time, and non-key parameters are uploaded to the cloud for phased preprocessing and organization. S5. Integrate environmental conditions, traffic flow, and equipment status of each section of the tunnel, and dynamically adjust the parameter warning thresholds for each section. S6. Compare the sorted key and non-key parameters with the dynamically adjusted parameter warning thresholds for each segment, and execute a three-level response, which includes a normal response, a first-level warning, and a second-level warning.

2. The tunnel safety management method based on big data according to claim 1, characterized in that, The specific steps in S1 include: S101. Obtain the geographic information, geological survey report, historical traffic flow statistics and accident record archives of the tunnel to be monitored, forming four types of data; S102. Map the four types of data to the linear spatial coordinate system of the tunnel, using the station number as the reference axis, to form a multi-dimensional feature vector sequence along the tunnel length direction. S103. Using a sliding window to identify feature abrupt change points, determine the natural functional boundary points of the tunnel, and divide the tunnel into several sections based on the multi-dimensional feature vector sequence along the tunnel length direction. S104. Generate a globally unique segment identifier for each segment.

3. The tunnel safety management method based on big data according to claim 1, characterized in that, The specific steps in S2 include: S201. Deploy a sensor network encompassing the environment, equipment, and traffic within each section; S202. According to the preset monitoring cycle, the raw sensor data of each section is collected synchronously, and the raw data from different sensor types are aligned according to the section identifier, the timestamp is synchronized and the format is standardized, and integrated into a unified multi-source dataset.

4. The tunnel safety management method based on big data according to claim 1, characterized in that, The specific steps in S3 include: S301. Retrieve the multi-source dataset from the previous period and perform a time-series comparison with the data from the current period. Calculate the deviation of each parameter over multiple consecutive periods. The deviation includes the mean offset rate or variance growth. S302. Based on the preset deviation threshold, filter out the parameter items that meet the condition of continuous deviation and deviation exceeding the deviation threshold, mark them as key parameters, and classify the rest as non-key parameters according to their respective sections.

5. The tunnel safety management method based on big data according to claim 1, characterized in that, The specific steps in S4 include: S401. In the edge computing nodes deployed in each section, key parameters are preprocessed and organized in real time, including real-time cleaning, noise reduction, and feature extraction. S402. Non-critical parameters are uploaded to the cloud platform via the communication network, and the cloud performs batch phased processing and organization on a periodic basis, including aggregation and statistics. S403: Preprocessed key parameters are retained at the edge nodes for rapid response, while the results of processing non-key parameters are stored in the cloud database for long-term analysis.

6. The tunnel safety management method based on big data according to claim 1, characterized in that, The specific steps in S5 include: S501. Integrate the real-time environmental conditions, traffic flow, and equipment status of each section to construct a section operation status profile; S502. Based on the constructed section operation status profile, a machine learning model is used to dynamically adjust the warning thresholds of various parameters under each section. The warning thresholds include normal warning thresholds, first-level warning thresholds, and second-level warning thresholds. S503. The updated dynamic early warning thresholds are stored by segment identifier and synchronized to edge nodes and the cloud.

7. The tunnel safety management method based on big data according to claim 1, characterized in that, The specific steps of S6 include: S601. Compare the sorted key parameters and non-key parameters with the dynamically adjusted warning thresholds of their respective sections to obtain the comparison results. S602. In the comparison results, if all parameters are within the normal warning threshold range, a normal response is triggered; if any key parameter exceeds the first-level warning threshold but does not reach the second-level warning threshold, a first-level warning is triggered; if any key parameter reaches or exceeds the second-level warning threshold, a second-level warning is triggered, and a warning report is generated for notification.

8. A tunnel safety management system based on big data, employing a tunnel safety management method based on big data as described in any one of claims 1-7, characterized in that, The management system includes: The section identification module is used to divide the tunnel to be monitored into sections and label them according to geographical information, geological survey reports, historical traffic flow statistics and accident records. Each section is assigned a unique identifier. The multi-source acquisition module is used to collect data on the equipment operating status and environmental parameters of each section of the tunnel during the current period through the deployed sensor network, and organize them into a multi-source dataset. The parameter filtering module is used to use the multi-source dataset collected in the previous tunnel monitoring cycle to filter out parameters that continuously deviate and whose deviation exceeds the deviation threshold through a preset deviation threshold, and organize them into key parameters in the current tunnel monitoring cycle. The segmented processing module is used to preprocess and organize key parameters in real time through locally deployed edge computing nodes, and upload non-key parameters to the cloud for phased preprocessing and organization. The threshold adjustment module is used to dynamically adjust the parameter warning thresholds of each section by integrating the environmental conditions, traffic flow, and equipment status of each section of the tunnel. The early warning response module is used to compare the sorted key and non-key parameters with the dynamically adjusted early warning thresholds for each segment and execute a three-level response, which includes a normal response, a first-level early warning, and a second-level early warning.

Images