Subway line environment risk monitoring method based on distributed optical fiber sensing device

Abstract

The application discloses a subway line environment risk monitoring method based on a distributed optical fiber sensing device, which comprises the following steps: installing the distributed optical fiber sensing device on the inner wall of a subway tunnel in the axial direction, constructing a data acquisition network covering a monitoring area, collecting sensing signals along the tunnel in real time, and forming an original data matrix; performing data cleaning on the original data matrix, and extracting signal data samples caused by external risk sources; constructing an intelligent case library based on network interaction and machine learning, wherein the intelligent case library stores feature data samples of multiple risk source signals; matching and comparing the signal data samples with the feature data samples, and identifying the types of the risk sources; and spatially positioning the identified risk sources to obtain the position information of the risk sources. The application has the advantages that the human intrusion and environmental mutation conditions around the operating subway can be monitored in real time, the categories, generation positions, disaster risk degrees and other information of the risk sources can be provided in real time, and the subway safety risks can be timely blocked and prevented.

Description

Technical Field

[0001] This invention relates to the technical field of environmental risk monitoring in subway tunnels, and in particular to a method for monitoring environmental risks along subway lines based on a distributed optical fiber sensing device. Background Technology

[0002] The areas surrounding operational subway tunnels are commonly subject to human intrusion such as drilling, piling, and trenching, as well as environmental hazards such as burst water pipes, gas pipelines, and collapsed sewer pipes. These factors pose potential catastrophic risks, impacting normal subway operations and even triggering major train safety accidents. Due to the long length of subway tunnels, high train frequency, and complex surrounding and surface environments, these human intrusions and underground pipeline accidents are unpredictable and uncertain. Their locations are often unpredictable and difficult to detect in a timely manner; their impact typically evolves over time, and if not detected and addressed promptly, they can easily escalate into major disasters. Therefore, timely and accurate perception and scientific assessment of the location and type of risks such as human intrusions and environmental hazards around subway tunnels, and timely control and mitigation of operational risks, are crucial technical aspects of all subway operation and management.

[0003] Because the risks of human intrusion and environmental disasters occur outside the tunnel, conventional inspection techniques and methods used inside the tunnel cannot detect or perceive the existence of these risks and their potential safety impacts. Secondly, the tunnels are long, and traditional acoustic, optical, and electronic sensor monitoring systems are too labor-intensive, costly, and limited in data acquisition methods to cover the entire subway line. Furthermore, the types of risk sources are numerous, and the vibration signals generated by different types of risk sources vary greatly in form, intensity, propagation mode, and attenuation characteristics. Coupled with the dynamic effects of subway train operation inside the tunnel, the coupling and superposition of multiple signals makes the perception, identification, and mathematical description of time-varying characteristics of monitoring signals extremely difficult. Traditional detection methods for operational inspections and pre-embedded sensors (displacement, acceleration, etc.) do not consider full coverage of the subway line, identification of multiple risk sources, data cleaning, or real-time rapid location of risk sources, thus failing to achieve full-area monitoring of the subway line, accurate location of risk sources, and timely management of the degree and evolution of risks.

[0004] In summary, there is an urgent need to develop a method for monitoring environmental risks along subway lines based on distributed optical fiber sensing devices. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of the prior art by providing a method for monitoring environmental risks along subway lines based on a distributed optical fiber sensing device. This method monitors human intrusion and sudden environmental changes around the operating subway in real time, and provides information such as the type of risk source, its location, and the degree of disaster risk, so as to promptly block and prevent subway safety risks.

[0006] The objective of this invention is achieved through the following technical solutions: A method for monitoring environmental risks along subway lines based on distributed optical fiber sensing devices, the method comprising the following steps: S1: Install distributed fiber optic sensing devices along the axial direction on the inner wall of the subway tunnel to build a data acquisition network covering the monitoring area, collect sensing signals along the tunnel in real time, and form a raw data matrix. S2: Perform data cleaning on the original data matrix and extract signal data samples caused by external risk sources; S3: Construct an intelligent case library based on network interaction and machine learning, wherein the intelligent case library stores feature data samples of signals from multiple risk sources; S4: Match and compare the signal data samples obtained in step S2 with the feature data samples in the intelligent case library to identify the type of risk source; S5: Spatial location of the identified risk sources to obtain their location information.

[0007] Step S1 specifically includes: The subway network is constructed into multiple independent monitoring subdomains according to the operating lines and sections. Each monitoring subdomain is equipped with distributed optical fiber cables and connected to a distributed optical fiber demodulator. The distributed optical fiber cables and the distributed optical fiber demodulator constitute the distributed optical fiber sensing device. Each distributed optical fiber demodulator transmits the monitoring signals to the big data center and cloud computing platform through the 5G communication network to form a unified raw database.

[0008] Step S2 specifically includes: S21: Set a signal strength threshold, initially extract signals exceeding the threshold from the original data matrix, and construct an abnormal signal matrix; S22: Establish the characteristic function or characteristic spectrum of non-risk source interference signals in the tunnel, and filter or remove abnormal signal matrices based on the characteristic function or characteristic spectrum to obtain signal data samples caused by human intrusion or environmental disaster risk sources.

[0009] In step S3, the method for constructing the intelligent case library includes: S31: Share risk information characteristics through the internal network of the monitoring system and collect relevant engineering case data through the public Internet network; S32: Conduct typicality evaluation and cluster analysis on the collected case data; S33: Store the clustered data samples into the current data case library, and use machine learning algorithms to clean the samples, remove duplicate samples, and supplement and improve the intelligent case library.

[0010] In step S4, the matching comparison specifically involves matching the time history curve and spectral distribution of the signal data sample with the time history curve and spectral distribution function established by B-Spline interpolation stored in the intelligent case library.

[0011] Step S5 specifically includes: S51: Compare the signal strength of sensor nodes at different axial positions at the same time, and determine the tunnel section where the sensor node with the strongest signal strength is located as the axial position of the risk source. S52: Based on the principle of signal attenuation and the principle of minimum geometric path, the specific azimuth coordinates of the risk source within the tunnel cross section are solved by iterative calculation using the difference in signal propagation path length from the risk source to at least two sensor nodes in different orientations and the relationship of signal energy attenuation.

[0012] Step S52 specifically includes: Within the tunnel cross-section, 0≤ α For areas ≤180°, use Δ α =45° is divided into 4 subdomains: I, II, III, and IV. α The angle between the line connecting the risk source and the geometric centroid of the tunnel and the horizontal direction, and in α =45° and α At a azimuth of 135°, there are respectively... S 1= S 2 and S 2= S 3, of which S 1. S 2. S 3 represents the straight-line distance of signal propagation from the risk source to the three sensor nodes in different locations; Based on the line-of-sight distance of signal propagation from the risk source to the sensor node S 1. S 2. S 3. Determine the subdomain where the risk source is located according to the following rules. Φ : ; Based on the law of cosines, establish the propagation distance from the risk source to the sensor node and the azimuth angle of the risk source. α and the distance from the source of risk to the geometric center of the tunnel r Relationship: ; in, R The radius of the tunnel cross section; S 1 and S The ratio of 2 satisfies: ; Based on the signal energy attenuation relationship, let the energy of the risk source be... The energy received by the sensor nodes are respectively 1. 2. Energy decay rate η If the signal is constant and the same on all propagation paths, then the signal propagation distance and energy attenuation satisfy the following: ; The location of the risk source is determined using an iterative method: Based on the subdomain where the identified risk source is located Φ ,Sure α The initial value range and initial step size Δ α , will the initial α Substitute 0 into the following formula: ; Solving for the results r ( α 0); according to α i+1 = α i + Δ α Increasing α Value, recalculate r ( α i+1 ); Set convergence control parameters | |, when| r ( α i+1 ) - r ( α i )| ≤ | When the convergence requirement is met, output the iteration step. α i+1 and r ( α i+1 Use the location coordinates of the risk source as the coordinates; otherwise, continue iterating until the convergence condition is met.

[0013] The advantages of this invention are: 1. The system adopts a distributed fiber optic sensing device with a single cable length of ≥10km and a sensor spacing of ≤10cm, enabling long-distance, high-density, all-weather continuous monitoring of subway lines without blind spots. 2. Fiber optic sensors are naturally resistant to electromagnetic interference, small in size, corrosion-resistant, and easy to install in complex tunnel environments without disturbing the operating subway. 3. Based on machine learning and network interaction, an intelligent case library is built. By matching time history curves / spectrums with B-Spline functions, multiple risk sources such as drilling, piling, and pipe bursts are automatically identified. 4. By using threshold extraction and interference signal feature map filtering, interference such as train vibration and environmental noise can be effectively eliminated, and external intrusion signals can be accurately extracted, reducing the false alarm rate. 5. First, determine the axial section based on the point with the maximum signal strength, and then, based on the principle of signal attenuation and the principle of minimum geometric path, accurately solve the azimuth and distance of the risk source within the section through an iterative algorithm; 6. Independent monitoring subdomains are constructed in different regions and at multiple levels, and data is transmitted to the big data center via 5G to support centralized networked monitoring of multiple subway lines and long distances; 7. Collect cases through internal sharing and the Internet, and clean them through cluster analysis and machine learning to achieve continuous iteration of the case library and improve the recognition accuracy; 8. Create a real-time monitoring solution for human intrusion and environmental disasters along the subway line, effectively solving the industry problem of "difficulty in discovering, identifying, and locating risks". Attached Figure Description

[0014] Figure 1 This is a flowchart of the subway environmental risk monitoring method based on a distributed optical fiber sensing device according to the present invention. Figure 2 This is a diagram showing the composition of the distributed optical fiber sensing device of the present invention. Figure 3 This is a schematic diagram illustrating the principle of intelligent dynamic database fusion data sample cleaning in this invention. Figure 4 This is a schematic diagram illustrating the principle of determining the axial cross-sectional position of the risk source in this invention. Figure 5 This is a schematic diagram illustrating the principle of risk source location determination in this invention. Figure 6 This is a diagram showing the composition of the full-coverage real-time monitoring system of the present invention; Figure 7 This is a schematic diagram of the intelligent case library based on network interaction and machine learning of the present invention. Detailed Implementation

[0015] The features and other related features of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments, so as to facilitate understanding by those skilled in the art: Example: Figure 1 As shown, this embodiment relates to a method for monitoring environmental risks along subway lines based on a distributed optical fiber sensing device. The method mainly includes the following steps: S1: Install distributed fiber optic sensing devices along the axial direction on the inner wall of the subway tunnel to build a data acquisition network covering the monitoring area, collect sensing signals along the tunnel in real time, and form a raw data matrix.

[0016] In this embodiment, as Figure 2 and Figure 6As shown, a real-time monitoring system covering the entire area is constructed, and the subway network is divided into multiple independent monitoring subdomains according to operating lines and sections, such as the first one. i Each operating route includes n The monitoring segment, the first j Each operating route includes m Each monitoring segment is equipped with a distributed optical fiber cable. Each monitoring sub-domain has 1-3 distributed optical fiber cables connected to a distributed optical fiber demodulator. Multiple spaced sensor nodes are installed on the distributed optical fiber cables, each corresponding to a measurement point. The distributed optical fiber cables and the distributed optical fiber demodulator constitute a distributed optical fiber sensing device. Each distributed optical fiber demodulator transmits the monitoring signals to a big data center and cloud computing platform via a 5G communication network, forming a unified raw database. The distributed optical fiber sensing device employs a distributed grating vibration acoustic wave sensing system with anti-electromagnetic interference and high adaptability. Its sampling frequency is ≤100kHz (continuous or triggered acquisition mode), spatial resolution is ≤1.6m, response bandwidth is 0.01Hz~50kHz, and sensing distance is ≤100km.

[0017] S2: Clean the original data matrix and extract signal data samples caused by external risk sources.

[0018] In this embodiment, as Figure 3 As shown, step S2 specifically includes: S21: Number the survey line I Measurement point number within the measurement line (sensor node on the distributed optical fiber cable) J Construct the initial data matrix D [ I , J ],in I =1, 2, ..., N , J =1, 2, ..., M Set a threshold for signal anomalies, and use the signal strength monitoring that is greater than or equal to the threshold to initially extract potential risk source signals, and establish an anomaly signal matrix.

[0019] S22: Considering that abnormal signals may contain non-external intrusion signals such as signal noise and train operation vibration, establish the characteristic function or characteristic spectrum of the relevant signals, perform data cleaning on the abnormal signal matrix based on the signal characteristics, and obtain signal data samples caused by human intrusion or environmental disaster risk sources.

[0020] External risk sources include human intrusion risk sources and environmental disaster risk sources. Human intrusion risk sources include drilling, piling, and trench excavation, while environmental disaster risk sources include water pipe bursts, gas pipe bursts, and sewer pipe collapses.

[0021] S3: Construct an intelligent case library based on network interaction and machine learning. The intelligent case library stores feature data samples of signals from multiple risk sources.

[0022] In this embodiment, as Figure 7 As shown, the construction methods for the intelligent case library include: S31: Monitor risk information characteristics shared through the internal network of the monitoring system and collect relevant engineering case data through the public Internet network.

[0023] S32: Conduct typicality evaluation and cluster analysis on the collected case data.

[0024] S33: Store the clustered data samples into the current data case library, and use machine learning algorithms to clean the samples, remove duplicate samples, and supplement and improve the intelligent case library.

[0025] S4: Match and compare the signal data samples obtained in step S2 with the feature data samples in the intelligent case library to identify the type of risk source.

[0026] In this embodiment, the matching comparison specifically involves matching the time history curve and spectral distribution of the signal data sample with the time history curve and spectral distribution function established by B-Spline interpolation stored in the intelligent case library.

[0027] S5: Spatial location of the identified risk sources to obtain their location information.

[0028] In this embodiment, step S5 specifically includes: S51: Axial section positioning: Compare the signal strength of sensor nodes at different axial positions at the same time, and determine the tunnel section where the sensor node with the strongest signal strength is located as the axial position of the risk source.

[0029] S52: In-section orientation location: Based on the signal attenuation principle and the minimum geometric path principle, the specific orientation coordinates of the risk source within the tunnel cross section are solved by iterative calculation using the difference in signal propagation path length from the risk source to at least two different orientation sensor nodes and the relationship of signal energy attenuation.

[0030] like Figure 4 As shown, step S51 specifically includes: Let the risk source be the first i The propagation distance of each sensor node is L i Based on the principle of signal attenuation, signal strength L iThe signal intensity decreases as the signal intensity increases. By comparing the signal strength received by different sensor nodes at the same time, the tunnel section corresponding to the node with the highest signal intensity is determined as the axial position of the risk source.

[0031] like Figure 5 As shown, step S52 specifically includes: Within the tunnel cross-section, 0≤ α For areas ≤180°, use Δ α =45° is divided into 4 subdomains: I, II, III, and IV. α The angle between the line connecting the risk source and the geometric centroid of the tunnel and the horizontal direction, and in α =45° and α At a azimuth of 135°, there are respectively... S 1= S 2 and S 2= S 3, of which S 1. S 2. S 3 represents the straight-line distance of signal propagation from the risk source to the three sensor nodes in different locations; Based on the line-of-sight distance of signal propagation from the risk source to the sensor node S 1. S 2. S 3. Determine the subdomain where the risk source is located according to the following rules. Φ : ; Based on the law of cosines, establish the propagation distance from the risk source to the sensor node and the azimuth angle of the risk source. α and the distance from the source of risk to the geometric center of the tunnel r Relationship: ; in, R The radius of the tunnel cross section; S 1 and S The ratio of 2 satisfies: ; Based on the signal energy attenuation relationship, let the energy of the risk source be... The energy received by the sensor nodes are respectively 1. 2. Energy decay rate η If the signal is constant and the same on all propagation paths, then the signal propagation distance and energy attenuation satisfy the following: ; The location of the risk source is determined using an iterative method: Based on the subdomain where the identified risk source is located Φ ,Sure α The initial value range and initial step size Δ α , will the initial α Substitute 0 into the following formula: ; Solving for the results r ( α 0); according to α i+1 = α i + Δ α Increasing α Value, recalculate r ( α i+1 ); Set convergence control parameters | |, when| r ( α i+1 ) - r ( α i )| ≤ | When the convergence requirement is met, output the iteration step. α i+1 and r ( α i+1 Use the location coordinates of the risk source as the coordinates; otherwise, continue iterating until the convergence condition is met.

[0032] The beneficial technical effects of this embodiment are as follows: 1. The system adopts a distributed fiber optic sensing device with a single cable length of ≥10km and a sensor spacing of ≤10cm, enabling long-distance, high-density, all-weather continuous monitoring of subway lines without blind spots. 2. Fiber optic sensors are naturally resistant to electromagnetic interference, small in size, corrosion-resistant, and easy to install in complex tunnel environments without disturbing the operating subway. 3. Based on machine learning and network interaction, an intelligent case library is built. By matching time history curves / spectrums with B-Spline functions, multiple risk sources such as drilling, piling, and pipe bursts are automatically identified. 4. By using threshold extraction and interference signal feature map filtering, interference such as train vibration and environmental noise can be effectively eliminated, and external intrusion signals can be accurately extracted, reducing the false alarm rate. 5. First, determine the axial section based on the point with the maximum signal strength, and then, based on the principle of signal attenuation and the principle of minimum geometric path, accurately solve the azimuth and distance of the risk source within the section through an iterative algorithm; 6. Independent monitoring subdomains are constructed in different regions and at multiple levels, and data is transmitted to the big data center via 5G to support centralized networked monitoring of multiple subway lines and long distances; 7. Collect cases through internal sharing and the Internet, and clean them through cluster analysis and machine learning to achieve continuous iteration of the case library and improve the recognition accuracy; 8. Create a real-time monitoring solution for human intrusion and environmental disasters along the subway line, effectively solving the industry problem of "difficulty in discovering, identifying, and locating risks".

Claims

1. A method for monitoring environmental risks along subway lines based on distributed optical fiber sensing devices, characterized in that... The method includes the following steps: S1: Install distributed fiber optic sensing devices along the axial direction on the inner wall of the subway tunnel to build a data acquisition network covering the monitoring area, collect sensing signals along the tunnel in real time, and form a raw data matrix. S2: Perform data cleaning on the original data matrix and extract signal data samples caused by external risk sources; S3: Construct an intelligent case library based on network interaction and machine learning, wherein the intelligent case library stores feature data samples of signals from multiple risk sources; S4: Match and compare the signal data samples obtained in step S2 with the feature data samples in the intelligent case library to identify the type of risk source; S5: Spatial location of the identified risk sources to obtain their location information.

2. The method for monitoring environmental risks along subway lines based on a distributed optical fiber sensing device as described in claim 1, characterized in that... Step S1 specifically includes: The subway network is constructed into multiple independent monitoring subdomains according to the operating lines and sections. Each monitoring subdomain is equipped with distributed optical fiber cables and connected to a distributed optical fiber demodulator. The distributed optical fiber cables and the distributed optical fiber demodulator constitute the distributed optical fiber sensing device. Each distributed optical fiber demodulator transmits the monitoring signals to the big data center and cloud computing platform through the 5G communication network to form a unified raw database.

3. The method for monitoring environmental risks along subway lines based on a distributed optical fiber sensing device as described in claim 1, characterized in that... Step S2 specifically includes: S21: Set a signal strength threshold, initially extract signals exceeding the threshold from the original data matrix, and construct an abnormal signal matrix; S22: Establish the characteristic function or characteristic spectrum of non-risk source interference signals in the tunnel, and filter or remove abnormal signal matrices based on the characteristic function or characteristic spectrum to obtain signal data samples caused by human intrusion or environmental disaster risk sources.

4. The method for monitoring environmental risks along subway lines based on a distributed optical fiber sensing device as described in claim 1, characterized in that... In step S3, the method for constructing the intelligent case library includes: S31: Share risk information characteristics through the internal network of the monitoring system and collect relevant engineering case data through the public Internet network; S32: Conduct typicality evaluation and cluster analysis on the collected case data; S33: Store the clustered data samples into the current data case library, and use machine learning algorithms to clean the samples, remove duplicate samples, and supplement and improve the intelligent case library.

5. The method for monitoring environmental risks along subway lines based on a distributed optical fiber sensing device as described in claim 1, characterized in that... In step S4, the matching comparison specifically involves matching the time history curve and spectral distribution of the signal data sample with the time history curve and spectral distribution function established by B-Spline interpolation stored in the intelligent case library.

6. The method for monitoring environmental risks along subway lines based on a distributed optical fiber sensing device as described in claim 1, characterized in that... Step S5 specifically includes: S51: Compare the signal strength of sensor nodes at different axial positions at the same time, and determine the tunnel section where the sensor node with the strongest signal strength is located as the axial position of the risk source. S52: Based on the principle of signal attenuation and the principle of minimum geometric path, the specific azimuth coordinates of the risk source within the tunnel cross section are solved by iterative calculation using the difference in signal propagation path length from the risk source to at least two sensor nodes in different orientations and the relationship of signal energy attenuation.

7. A method for monitoring environmental risks along subway lines based on a distributed optical fiber sensing device as described in claim 5, characterized in that... Step S52 specifically includes: Within the tunnel cross-section, 0≤ α For areas ≤180°, use Δ α =45° is divided into 4 subdomains: I, II, III, and IV. α The angle between the line connecting the risk source and the geometric centroid of the tunnel and the horizontal direction, and in α =45° and α At a azimuth of 135°, there are respectively... S 1= S 2 and S 2= S 3, of which S 1. S 2. S 3 represents the straight-line distance of signal propagation from the risk source to the three sensor nodes in different locations; Based on the line-of-sight distance of signal propagation from the risk source to the sensor node S 1. S 2. S 3. Determine the subdomain where the risk source is located according to the following rules. Φ : ； Based on the law of cosines, establish the propagation distance from the risk source to the sensor node and the azimuth angle of the risk source. α and the distance from the source of risk to the geometric center of the tunnel r Relationship: ； in, R The radius of the tunnel cross section; S 1 and S The ratio of 2 satisfies: ； Based on the signal energy attenuation relationship, let the energy of the risk source be... The energy received by the sensor nodes are respectively 1.

2. Energy decay rate η If the signal is constant and the same on all propagation paths, then the signal propagation distance and energy attenuation satisfy the following: ； The location of the risk source is determined using an iterative method: Based on the subdomain where the identified risk source is located Φ ,Sure α The initial value range and initial step size Δ α , will the initial α Substitute 0 into the following formula: ； Solving for the results r ( α 0); according to α i+1 = α i + Δ α Increasing α Value, recalculate r ( α i+1 ); Set convergence control parameters | |, when| r ( α i+1 ) - r ( α i )| ≤ | When the convergence requirement is met, output the iteration step. α i+1 and r ( α i+1 Use the location coordinates of the risk source as the reference point; otherwise, continue iterating until the convergence condition is met.

Images