Online health monitoring method for prefabricated track filling structure

Abstract

The invention discloses an online health monitoring method for a prefabricated track filling structure, and relates to the technical field of track traffic engineering, and the method comprises the steps: S1, an optical fiber pre-embedding process; s2, signal processing; and S3, data processing: establishing a mapping relation between signal features and damage by utilizing machine learning. According to the on-line health monitoring method for the prefabricated track filling structure, a distributed optical fiber sensing network is pre-embedded and is arranged in a steel fiber fine aggregate concrete filling layer in a grid or spiral shape in the longitudinal direction and the transverse direction, and a key stress area is covered; a densely distributed strain demodulator and a distributed sound wave sensing system are combined to synchronously acquire strain and vibration data, and frequency domain analysis is utilized to eliminate steel fiber scattering interference; and establishing a mapping relation among the strain peak characteristics, the stress wave frequency spectrum change and the damage type through machine learning, and realizing crack width calculation and crack depth judgment.

Description

Technical Field

[0001] This invention belongs to the field of rail transit engineering technology, specifically relating to a real-time, non-destructive health monitoring system and method for steel fiber reinforced concrete filling layers in prefabricated track structures, used to assess the health status of the structure and ensure track safety and durability. Background Technology

[0002] In precast track structures, the steel fiber reinforced concrete filling layer is a key load-bearing component. However, it is prone to hidden damage such as microcracks and voids due to long-term exposure to train dynamic loads and temperature changes. Traditional detection methods such as manual inspection and ultrasonic testing cannot achieve real-time monitoring and are not sensitive to early damage. Existing fiber optic monitoring technologies are mostly used for large-volume concrete structures and are not optimized for the confined space of precast tracks and the characteristics of steel fiber reinforced concrete, resulting in the following shortcomings of existing technologies: Sensor deployment is difficult: the prefabricated track structure is compact, and the coverage of traditional point sensors (such as strain gauges) is limited; Steel fiber interference: The distribution of steel fibers may cause signal distortion in electromagnetic sensors; Insufficient long-term stability: Existing monitoring systems are ill-suited to the long-term vibration environment of rail transit.

[0003] Therefore, it is necessary to propose an online health monitoring method for prefabricated track filling structures to solve the above problems. Summary of the Invention

[0004] The main objective of this invention is to provide an online health monitoring method for prefabricated track filling structures, which can effectively solve the problems in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An online health monitoring method for a prefabricated track filling structure includes the following steps: S1: Fiber optic pre-embedding process. The fiber optic pre-embedding process is implemented in the prefabricated track filling layer to ensure stable coverage of key stress areas by the sensor network. S2: Signal processing, through a dual-signal demodulation system to achieve real-time data acquisition and analysis, uses a dense distributed strain demodulator to collect strain data and detect whether the ultimate tensile strength of concrete exceeds the limit and peak growth; the distributed acoustic wave sensing system generates stress waves by striking with a vibrating hammer, compares spectral differences, combines frequency domain analysis to eliminate steel fiber interference, and simultaneously fuses strain and vibration data to preliminarily identify cracks or voids. The process is automated, with short response time, ensuring high accuracy and anti-interference. S3: Data processing, which uses machine learning to establish a mapping relationship between signal features and damage.

[0006] Preferably, step S1 specifically includes the following steps: S101: Construction preparation and track slab positioning. Before the construction of the precast track bed, site survey and material preparation must be completed. First, according to the track design drawings, the key stress area of ​​the filling layer is determined. The construction team uses GPS and laser measuring tools to accurately mark the installation position of the track slab, ensuring that the error is controlled within ±2mm. Before installation, check the flatness of the track slab surface and remove debris and oil stains to prevent affecting the adhesion of optical fibers. S102: Fiber optic cable threading and initial fixing. After the track slab is installed in the designated position, the fiber optic cable laying begins. This method uses distributed fiber optic sensing cables, including strain sensing cables and vibration sensing cables, both of which are flexible designs. Specific steps: Cable threading: Insert the optical fiber cable through the pre-drilled hole under the track slab. During the operation, use a traction rope to assist in avoiding bending or damage to the optical fiber. The length of the optical cable is customized according to the size of the track slab. Fixing device: Two steel bar supports with a diameter of 8mm are installed at two-thirds of the height of the filling layer, with a spacing of 50cm. The optical fiber is tied to the supports with iron wire to ensure that the optical cable is straight and not loose. Layout design: The optical cables are arranged in a grid or spiral pattern along the longitudinal and transverse directions of the track. The longitudinal layout covers the length of the track, and the transverse layout connects adjacent track plates to form a dense grid with a spacing of 10cm, covering key stress areas. S103: Anti-offset and reserved treatment, used to prevent fiber optic offset caused by the impact during concrete pouring, and implements multiple protection measures; S104: Concrete pouring and integration. After the optical fiber is fixed, steel fiber fine stone concrete is poured. The concrete weight ratio is cement:sand:steel fiber = 1:2:0.1. The steel fiber is 30mm long and 0.5mm in diameter to enhance the crack resistance of the filling layer. S105: Optical cable access and series connection. After the concrete has cured, it is connected to the signal demodulation system. The strain sensing optical cable adopts the end-to-end series connection method, and the vibration sensing optical cable is laid in a straight line.

[0007] Preferably, step S103 specifically includes: Location optimization: The fiber optic cable should be laid at least 50cm away from the concrete pouring opening. The pouring opening is usually located at the edge of the track slab, where the impact force is large. The area near the fiber optic cable should be wrapped with cushioning material to reduce the impact of vibration. Reserved length: 10cm of optical fiber is reserved outside each track slab for easy maintenance later. The reserved end is sealed with a waterproof cap to prevent moisture intrusion. If the optical fiber breaks during monitoring, a backup optical cable can be quickly connected through the reserved part without interrupting track operation. Quality inspection: After deployment, use an optical time domain reflectometer to test the fiber continuity and loss to ensure signal integrity. The loss value should be controlled below 0.5dB / km. If it fails to meet the requirements, it should be redeployed.

[0008] Preferably, the pouring step of S104 is as follows: Layered pouring: The process is carried out in two layers. The first layer is 10cm thick. After covering the fiber optic network, it is lightly vibrated to remove air bubbles. The second layer is filled to the design height. The vibration intensity is controlled at a medium level to avoid direct impact on the fiber optics. The frequency of the vibrator is set to 50Hz. Temperature control: The pouring environment temperature is maintained between 5-30℃ to prevent thermal expansion and contraction from affecting the optical fiber. Shading measures are adopted in summer and antifreeze is added in winter. The initial setting time of concrete is about 2 hours. During this period, the optical fiber signal is monitored in real time to ensure no displacement. Integration and verification: 24 hours after pouring, conduct preliminary signal testing and read strain baseline data through demodulator. If the signal is abnormal, adjust the fiber position or add more pouring.

[0009] Preferably, step S2 specifically includes the following steps: S201: Data Acquisition and Synchronization Start-up. The acquisition process takes place during train operation hours, enabling 24 / 7 online monitoring. Start-up steps: System activation: Send a command via the cloud server to start the dense distributed strain demodulator and the distributed acoustic wave sensing system. The demodulator sampling frequency is set to 100Hz to cover the entire track length; the distributed acoustic wave sensing system has a sampling rate of 1kHz to capture high-frequency vibrations. Strain data acquisition: The strain sensing optical cable monitors the micro-strain changes of the filling layer in real time. The data is transmitted to the demodulator through the optical cable. During acquisition, the key stress area is focused and scanned once every 10 seconds to generate a strain distribution map. Vibration data acquisition: A vibrating hammer is used to strike different positions on the track slab to generate stress waves. The striking point is selected in a highly sensitive area. The vibration sensing optical cable captures the stress wave propagation data and transmits it to the distributed acoustic wave sensing system. S202: Demodulation analysis and frequency domain processing. The demodulation stage processes the original signal, eliminates steel fiber interference, and extracts effective features, specifically including: Strain signal demodulation: The dense distributed strain demodulator uses UWFBG technology to demodulate strain data. The process includes: converting optical signals into electrical signals, performing frequency domain analysis through fast Fourier transform to eliminate steel fiber scattering interference, and using a frequency band filtering algorithm to focus on 1-10kHz to avoid the steel fiber resonance region. The demodulated data includes a strain value sequence, which is displayed in real time on the monitoring interface. Vibration signal demodulation: The distributed acoustic wave sensing system processes vibration data, performs spectrum analysis, and uses fast Fourier transform to convert the time-domain signal into the frequency domain to extract the dominant frequency features; S203: Preliminary damage identification and verification, demodulation of data, and real-time damage warning and verification: Strain over-limit warning: Determines whether the measured strain exceeds the ultimate tensile strength of concrete. The algorithm compares the current strain with the threshold in real time. If the strain exceeds the limit and a significant peak appears, an early warning is triggered. Peak growth is monitored through time series analysis. Vibration verification: In the warning area, additional excitation hammers were used to strike the area, and vibration data was collected through a distributed acoustic wave sensing system. The stress wave spectrum at different times was compared: when there were no cracks, the stress wave propagation pattern was uniform; when there were cracks, the propagation pattern was significantly different. Comprehensive diagnosis: Combining strain and vibration results, the damage type is initially identified. If the strain peak is accompanied by a shift in the dominant frequency, it is judged to be a crack; if the vibration data shows energy attenuation, it is a void. The diagnosis results are output to the alarm system to trigger maintenance instructions.

[0010] Preferably, step S3 specifically includes the following steps: S301: Data preprocessing and feature extraction, used to preprocess noise and extract key features, specifically including: Strain data cleaning: A moving average filter with a window size of 5 points is applied to the discrete strain data to eliminate random noise. Feature extraction includes peak locations, denoted here as... a 1 Starting point and a 2 Termination point, strain amplitude ε(x) The growth rate used to calculate the time derivative; Vibration data conversion: Extracting the dominant frequency from frequency domain data The peak detection algorithm identifies the main frequency value, and its features include: main frequency offset Δ. The spectrum energy distribution is then input into a machine learning model; S302: Crack width calculation, discrete strain data integration to quantify crack width, the formula is: ; in a 1 and a 2 The peak positions are specifically the starting and ending points of strain exceeding the benchmark value, which is set at 90% of the ultimate tensile strength of concrete. ε(x) The strain distribution at different locations on the track slab is represented by numerical integration of discrete point data. Formula for identifying peak regions: When ε(x) > Baseline value, marker a 1 and a 2 ; Numerical Integration: Calculating integral values ​​using the Python SciPy library The unit is mm. Calibration and verification: Compared with visual inspection, the error is controlled within <5%; S303: Crack depth determination, stress wave dominant frequency shift, assessment of crack depth, formula for the dominant frequency of normal concrete vibration: ; in For wave speed, The width of the track slab; The formula for the frequency shift when a crack occurs is: ; in The horizontal distance between the striking surface and the crack; The judgment method is as follows: Calculation benchmark : Main frequency without damage; Measurement offset When there is a crack, f1 is obtained through spectral analysis; Compare differences: Δ If Δ If the crack depth is greater than a preset threshold, then... It can be approximated as Combine a pre-trained support vector machine machine learning model to establish a Δ and The mapping is used to improve accuracy. S304: Application of machine learning models. Machine learning establishes the mapping relationship between strain and vibration characteristics and damage, specifically including: Model training: The classification model is trained using historical data. Input features such as peak amplitude and main frequency shift are used to output the damage type. The model uses SVM. Online data is input into the model to generate diagnostic reports. The model parameters are updated monthly to adapt to environmental changes.

[0011] Compared with the prior art, the present invention provides an online health monitoring method for prefabricated track filling structures, which has the following beneficial effects: 1. The online health monitoring method for the precast track filling structure utilizes a pre-embedded distributed optical fiber sensor network, arranged in a grid or spiral pattern along the longitudinal and transverse directions within the steel fiber fine stone concrete filling layer, covering key stress areas. It combines a densely distributed strain demodulator with a distributed acoustic wave sensing system to synchronously collect strain and vibration data, using frequency domain analysis to eliminate steel fiber scattering interference. Furthermore, it establishes a mapping relationship between strain peak characteristics, stress wave spectrum changes, and damage types through machine learning, enabling crack width calculation and crack depth assessment.

[0012] 2. The online health monitoring method for the precast track filling structure adopts a flexible fiber optic mesh layout, which is adapted to the narrow space of the precast track. It has an anti-interference signal processing mechanism to overcome the electromagnetic interference of steel fibers. Through a multi-parameter fusion diagnostic model, it integrates strain over-limit early warning and vibration wave propagation analysis to accurately identify early damage such as cracks and voids. This method significantly improves the detection rate of hidden defects, realizes real-time, non-destructive, and full-area health monitoring of the precast track filling structure, extends the service life of the track, and reduces maintenance costs.

[0013] 3. The online health monitoring method for this precast track filling structure involves pre-embedding distributed optical fiber sensing cables within the steel fiber reinforced concrete filling layer, arranged in a grid or spiral pattern along the longitudinal and transverse directions of the track to cover key stress areas. Two densely distributed strain sensing cables are connected in series along the track slab to a densely distributed strain demodulator. The distributed acoustic wave sensing system uses a vibrating hammer to excite stress waves at different locations on the track slab. Utilizing the significant difference in the propagation patterns of stress waves in concrete with and without crack peaks, the system performs spectral analysis of the vibration signals to make judgments. Machine learning is used to establish a mapping relationship between signal characteristics and damage types, enabling early warning. This method achieves real-time, non-destructive, and comprehensive monitoring of the precast track filling structure, improving the detection rate of hidden defects, extending the service life of the track, and reducing maintenance costs. Attached Figure Description

[0014] Figure 1 This is a fiber optic cable layout diagram of the present invention; Figure 2 This is a hardware deployment diagram of the present invention; Figure 3 This is a flowchart of the crack calculation process of the present invention. Detailed Implementation

[0015] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments. Example 1:

[0016] like Figure 1-3 As shown, an online health monitoring method for a prefabricated track filling structure includes the following steps: S1: Fiber optic pre-embedding process, which is implemented in the prefabricated track filling layer to ensure stable coverage of the sensor network in key stress areas. S101: Construction preparation and track slab positioning. Before the construction of the precast track bed, site survey and material preparation must be completed. First, according to the track design drawings, the key stress areas of the filling layer are determined, such as the track slab connection and high load areas. These areas are easily affected by train dynamic load and temperature changes, and are high-risk areas for cracks and delamination. The construction team uses GPS and laser measuring tools to accurately mark the installation position of the track slab, ensuring that the error is controlled within ±2mm. The track slab is made of precast concrete components, and the size is usually 6m×2.5m×0.3m to adapt to the standard track width. Before installation, check the flatness of the track slab surface and remove debris and oil stains to prevent affecting the adhesion of optical fibers. S102: Fiber optic cable threading and initial fixing. After the track slab is installed in the designated position, the fiber optic cable laying begins. This method uses distributed fiber optic sensing cables, including strain sensing cables and vibration sensing cables, both of which are flexible designs. Specific steps: Cable threading operation: Insert the optical fiber cable through the pre-drilled hole under the track slab. During the operation, use a traction rope to assist in avoiding bending or damage to the optical fiber. The length of the optical cable is customized according to the size of the track slab. For example, an 8-meter optical cable is required for a 6-meter track slab. Fixing device: Two steel bar supports with a diameter of 8mm are erected at two-thirds of the height of the filling layer, with a spacing of 50cm. The optical fiber is tied to the supports with iron wire to ensure that the optical cable is straight and not loose. When tying, a cross method is used, with a tying point every 20cm to prevent displacement during concrete pouring. Layout design: The optical cable is laid out in a grid or spiral pattern along the longitudinal and transverse directions of the track. The longitudinal layout covers the length of the track, and the transverse layout connects adjacent track plates to form a dense grid with a spacing of 10cm, covering the key stress area. S103: Anti-offset and pre-installation treatment, used to prevent fiber optic misalignment caused by impact during concrete pouring, implementing multiple protective measures, specifically including: Location optimization: The fiber optic cable should be laid at least 50cm away from the concrete pouring opening. The pouring opening is usually located at the edge of the track slab, where the impact force is large. The area near the fiber optic cable should be wrapped with cushioning material to reduce the impact of vibration. Reserved length: 10cm of optical fiber is reserved outside each track slab for easy maintenance later. The reserved end is sealed with a waterproof cap to prevent moisture intrusion. If the optical fiber breaks during monitoring, a backup optical cable can be quickly connected through the reserved part without interrupting track operation. Quality inspection: After deployment, use an optical time domain reflectometer to test the fiber continuity and loss to ensure signal integrity. The loss value should be controlled below 0.5dB / km. If it fails to meet the requirements, it should be redeployed. S104: Concrete pouring and integration. After the optical fiber is fixed, steel fiber fine aggregate concrete is poured. The concrete weight ratio is cement:sand:steel fiber = 1:2:0.1, with steel fibers 30mm in length and 0.5mm in diameter. The pouring steps to enhance the crack resistance of the filling layer are as follows: Layered pouring: The process is carried out in two layers. The first layer is 10cm thick. After covering the fiber optic network, it is lightly vibrated to remove air bubbles. The second layer is filled to the design height. The vibration intensity is controlled at a medium level to avoid direct impact on the fiber optics. The frequency of the vibrator is set to 50Hz. Temperature control: The pouring environment temperature is maintained between 5-30℃ to prevent thermal expansion and contraction from affecting the optical fiber. Shading measures are adopted in summer and antifreeze is added in winter. The initial setting time of concrete is about 2 hours. During this period, the optical fiber signal is monitored in real time to ensure no displacement. Integration and verification: 24 hours after pouring, conduct preliminary signal testing and read strain baseline data through a demodulator. If the signal is abnormal, adjust the fiber position or add more pouring. S105: Optical cable access and series connection. After the concrete has cured, the signal demodulation system is connected. The strain sensing optical cable adopts a series connection method, and the vibration sensing optical cable is laid in a straight line. Specifically: Series connection: Two densely distributed strain sensing optical cables are laid along the track slab, each with a length matching the track slab. The two ends are connected in series through optical fiber connectors to the densely distributed strain demodulator. The series design enhances signal continuity and reduces data loss. Straight-line deployment of vibration optical cables: Vibration sensing optical cables are deployed independently, with a straight path covering high-vibration areas (such as track joints), and connected to a distributed acoustic wave sensing system. When connecting, FC / APC connectors are used to ensure low reflection loss. System initialization: Start the demodulator and calibrate the reference values ​​(such as zero-point strain), which takes about 30 minutes. The calibration data is stored on the local server to lay the foundation for subsequent monitoring. Example 2:

[0017] like Figure 1-3 As shown, an online health monitoring method for precast track filling structures is described. S2: Signal processing. Real-time data acquisition and analysis are achieved through a dual-signal demodulation system. Strain data is collected using a dense distributed strain demodulator to detect whether the ultimate tensile strength of the concrete exceeds the limit and to detect peak growth. A distributed acoustic wave sensing system generates stress waves by striking with a vibrating hammer, compares spectral differences, and combines frequency domain analysis to eliminate steel fiber interference. Strain and vibration data are simultaneously fused to preliminarily identify cracks or voids. The process is automated, with a short response time, ensuring high accuracy and anti-interference capabilities. Specifically, the method includes the following steps: S201: Data Acquisition and Synchronization Start-up. The acquisition process takes place during train operation hours, enabling 24 / 7 online monitoring. Start-up steps: System activation: Send a command via the cloud server to start the dense distributed strain demodulator and the distributed acoustic wave sensing system. The demodulator sampling frequency is set to 100Hz to cover the entire track length; the distributed acoustic wave sensing system has a sampling rate of 1kHz to capture high-frequency vibrations. Strain data acquisition: The strain sensing optical cable monitors the micro-strain changes of the filling layer in real time. The data is transmitted to the demodulator through the optical cable. During acquisition, the key stress area is focused and scanned once every 10 seconds to generate a strain distribution map. Background noise is suppressed by hardware filter and the initial signal-to-noise ratio is ≥60dB. Vibration data acquisition: Stress waves are generated by striking different locations on the track slab with a vibrating hammer. Highly sensitive areas, such as the center of the slab or joints, are selected as the striking points. The force is standardized to 50N to avoid overload. The vibration sensing optical cable captures the stress wave propagation data and transmits it to the distributed acoustic wave sensing system. The acquisition interval is 5 minutes to reduce the system load. The synchronization mechanism ensures that the timestamps of strain and vibration data are aligned with an error of <1ms, which facilitates subsequent fusion analysis. S202: Demodulation analysis and frequency domain processing. The demodulation stage processes the original signal, eliminates steel fiber interference, and extracts effective features, specifically including: Strain signal demodulation: The dense distributed strain demodulator uses UWFBG technology to demodulate strain data. The process includes: converting optical signals into electrical signals, performing frequency domain analysis through fast Fourier transform to eliminate steel fiber scattering interference, and using a frequency band filtering algorithm to focus on 1-10kHz to avoid the steel fiber resonance region. The demodulated data includes a strain value sequence, which is displayed in real time on the monitoring interface. Vibration signal demodulation: The distributed acoustic wave sensing system processes vibration data, performs spectrum analysis, and uses fast Fourier transform to convert the time-domain signal into the frequency domain to extract the dominant frequency features; S203: Preliminary damage identification and verification, demodulation of data, and real-time damage warning and verification: Strain over-limit warning: Determines whether the measured strain exceeds the ultimate tensile strength of concrete. The algorithm compares the current strain with the threshold in real time. If the strain exceeds the limit and a significant peak appears, an early warning is triggered. Peak growth is monitored through time series analysis. Vibration verification: In the warning area, additional excitation hammers were used to strike the area, and vibration data was collected through a distributed acoustic wave sensing system. The stress wave spectrum at different times was compared: when there were no cracks, the stress wave propagation pattern was uniform; when there were cracks, the propagation pattern was significantly different. Comprehensive diagnosis: Combining strain and vibration results, the damage type is initially identified. If the strain peak is accompanied by a shift in the dominant frequency, it is judged to be a crack; if the vibration data shows energy attenuation, it is a void. The diagnosis results are output to the alarm system to trigger maintenance instructions. Example 3:

[0018] like Figure 1-3As shown, an online health monitoring method for a prefabricated track filling structure, S3: data processing, which establishes a mapping relationship between signal features and damage by utilizing machine learning, specifically includes the following steps: S301: Data preprocessing and feature extraction, used to preprocess noise and extract key features, specifically including: Strain data cleaning: A moving average filter with a window size of 5 points is applied to the discrete strain data to eliminate random noise. Feature extraction includes peak locations, denoted here as... a 1 Starting point and a 2 Termination point, strain amplitude ε(x) The growth rate used to calculate the time derivative; Vibration data conversion: Extracting the dominant frequency from frequency domain data The peak detection algorithm identifies the main frequency value, and its features include: main frequency offset Δ. The spectrum energy distribution is then input into a machine learning model; S302: Crack width calculation, discrete strain data integration to quantify crack width, the formula is: ; in a 1 and a 2 The peak positions are specifically the starting and ending points of strain exceeding the benchmark value, which is set at 90% of the ultimate tensile strength of concrete. ε(x) The strain distribution at different locations on the track slab is represented by numerical integration of discrete point data. Formula for identifying peak regions: When ε(x) > Baseline value, marker a 1 and a 2 ; Numerical Integration: Calculating integral values ​​using the Python SciPy library The unit is mm; Calibration and verification: Compared with visual inspection, the error is controlled within <5%; S303: Crack depth determination, stress wave dominant frequency shift, assessment of crack depth, formula for the dominant frequency of normal concrete vibration: ; in For wave speed, The width of the track slab; The formula for the frequency shift when a crack occurs is: ; in The horizontal distance between the striking surface and the crack; The judgment method is as follows: Calculation benchmark : Main frequency without damage; Measurement offset When there is a crack, f1 is obtained through spectral analysis; Compare differences: Δ If Δ If the crack depth is greater than a preset threshold, then... It can be approximated as Combine a pre-trained support vector machine machine learning model to establish a Δ and The mapping is used to improve accuracy. S304: Application of machine learning models. Machine learning establishes the mapping relationship between strain and vibration characteristics and damage, specifically including: Model training: The classification model is trained using historical data. Input features such as peak amplitude and main frequency shift are used to output the damage type. The model uses SVM. Online data is input into the model to generate diagnostic reports. The model parameters are updated monthly to adapt to environmental changes.

[0019] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. An online health monitoring method for a prefabricated track filling structure, characterized in that: The following steps are included: S1: Fiber optic pre-embedding process. The fiber optic pre-embedding process is implemented in the prefabricated track filling layer to ensure stable coverage of key stress areas by the sensor network. S2: Signal processing, through a dual-signal demodulation system to achieve real-time data acquisition and analysis, uses a dense distributed strain demodulator to collect strain data and detect whether the ultimate tensile strength of concrete exceeds the limit and peak growth; the distributed acoustic wave sensing system generates stress waves by striking with a vibrating hammer, compares spectral differences, combines frequency domain analysis to eliminate steel fiber interference, and simultaneously fuses strain and vibration data to preliminarily identify cracks or voids. The process is automated, with short response time, ensuring high accuracy and anti-interference. S3: Data processing, which uses machine learning to establish a mapping relationship between signal features and damage.

2. The online health monitoring method for a prefabricated track filling structure according to claim 1, characterized in that: S1 specifically includes the following steps: S101: Construction preparation and track slab positioning. Before the construction of the precast track bed, site survey and material preparation must be completed. First, according to the track design drawings, the key stress area of ​​the filling layer is determined. The construction team uses GPS and laser measuring tools to accurately mark the installation position of the track slab, ensuring that the error is controlled within ±2mm. Before installation, check the flatness of the track slab surface and remove debris and oil stains to prevent affecting the adhesion of optical fibers. S102: Fiber optic cable threading and initial fixing. After the track slab is installed in the designated position, the fiber optic cable laying begins. This method uses distributed fiber optic sensing cables, including strain sensing cables and vibration sensing cables, both of which are flexible designs. Specific steps: Cable threading: Insert the optical fiber cable through the pre-drilled hole under the track slab. During the operation, use a traction rope to assist in avoiding bending or damage to the optical fiber. The length of the optical cable is customized according to the size of the track slab. Fixing device: Two steel bar supports with a diameter of 8mm are installed at two-thirds of the height of the filling layer, with a spacing of 50cm. The optical fiber is tied to the supports with iron wire to ensure that the optical cable is straight and not loose. Layout design: The optical cables are arranged in a grid or spiral pattern along the longitudinal and transverse directions of the track. The longitudinal layout covers the length of the track, and the transverse layout connects adjacent track plates to form a dense grid with a spacing of 10cm, covering key stress areas. S103: Anti-offset and reserved treatment, used to prevent fiber optic offset caused by the impact during concrete pouring, and implements multiple protection measures; S104: Concrete pouring and integration. After the optical fiber is fixed, steel fiber fine stone concrete is poured. The concrete weight ratio is cement:sand:steel fiber = 1:2:0.

1. The steel fiber is 30mm long and 0.5mm in diameter to enhance the crack resistance of the filling layer. S105: Optical cable access and series connection. After the concrete has cured, it is connected to the signal demodulation system. The strain sensing optical cable adopts the end-to-end series connection method, and the vibration sensing optical cable is laid in a straight line.

3. The online health monitoring method for a prefabricated track filling structure according to claim 2, characterized in that: Specifically, S103 includes: Location optimization: The fiber optic cable should be laid at least 50cm away from the concrete pouring opening. The pouring opening is usually located at the edge of the track slab, where the impact force is large. The area near the fiber optic cable should be wrapped with cushioning material to reduce the impact of vibration. Reserved length: 10cm of optical fiber is reserved outside each track slab for easy maintenance later. The reserved end is sealed with a waterproof cap to prevent moisture intrusion. If the optical fiber breaks during monitoring, a backup optical cable can be quickly connected through the reserved part without interrupting track operation. Quality inspection: After deployment, use an optical time domain reflectometer to test the fiber continuity and loss to ensure signal integrity. The loss value should be controlled below 0.5dB / km. If it fails to meet the requirements, it should be redeployed.

4. The online health monitoring method for a prefabricated track filling structure according to claim 2, characterized in that: The pouring steps for S104 are as follows: Layered pouring: The process is carried out in two layers. The first layer is 10cm thick. After covering the fiber optic network, it is lightly vibrated to remove air bubbles. The second layer is filled to the design height. The vibration intensity is controlled at a medium level to avoid direct impact on the fiber optics. The frequency of the vibrator is set to 50Hz. Temperature control: The pouring environment temperature is maintained between 5-30℃ to prevent thermal expansion and contraction from affecting the optical fiber. Shading measures are adopted in summer and antifreeze is added in winter. The initial setting time of concrete is about 2 hours. During this period, the optical fiber signal is monitored in real time to ensure no displacement. Integration and verification: 24 hours after pouring, conduct preliminary signal testing and read strain baseline data through demodulator. If the signal is abnormal, adjust the fiber position or add more pouring.

5. The online health monitoring method for a prefabricated track filling structure according to claim 1, characterized in that: S2 specifically includes the following steps: S201: Data Acquisition and Synchronization Start-up. The acquisition process takes place during train operation hours, enabling 24 / 7 online monitoring. Start-up steps: System activation: Send a command via the cloud server to start the dense distributed strain demodulator and the distributed acoustic wave sensing system. The demodulator sampling frequency is set to 100Hz to cover the entire track length; the distributed acoustic wave sensing system has a sampling rate of 1kHz to capture high-frequency vibrations. Strain data acquisition: The strain sensing optical cable monitors the micro-strain changes of the filling layer in real time. The data is transmitted to the demodulator through the optical cable. During acquisition, the key stress area is focused and scanned once every 10 seconds to generate a strain distribution map. Vibration data acquisition: A vibrating hammer is used to strike different positions on the track slab to generate stress waves. The striking point is selected in a highly sensitive area. The vibration sensing optical cable captures the stress wave propagation data and transmits it to the distributed acoustic wave sensing system. S202: Demodulation analysis and frequency domain processing. The demodulation stage processes the original signal, eliminates steel fiber interference, and extracts effective features, specifically including: Strain signal demodulation: The dense distributed strain demodulator uses UWFBG technology to demodulate strain data. The process includes: converting optical signals into electrical signals, performing frequency domain analysis through fast Fourier transform to eliminate steel fiber scattering interference, and using a frequency band filtering algorithm to focus on 1-10kHz to avoid the steel fiber resonance region. The demodulated data includes a strain value sequence, which is displayed in real time on the monitoring interface. Vibration signal demodulation: The distributed acoustic wave sensing system processes vibration data, performs spectrum analysis, and uses fast Fourier transform to convert the time-domain signal into the frequency domain to extract the dominant frequency features; S203: Preliminary damage identification and verification, demodulation of data, and real-time damage warning and verification: Strain over-limit warning: Determines whether the measured strain exceeds the ultimate tensile strength of concrete. The algorithm compares the current strain with the threshold in real time. If the strain exceeds the limit and a significant peak appears, an early warning is triggered. Peak growth is monitored through time series analysis. Vibration verification: In the warning area, additional excitation hammers were used to strike the area, and vibration data was collected through a distributed acoustic wave sensing system. The stress wave spectrum at different times was compared: when there were no cracks, the stress wave propagation pattern was uniform; when there were cracks, the propagation pattern was significantly different. Comprehensive diagnosis: Combining strain and vibration results, the damage type is initially identified. If the strain peak is accompanied by a shift in the dominant frequency, it is judged to be a crack; if the vibration data shows energy attenuation, it is a void. The diagnosis results are output to the alarm system to trigger maintenance instructions.

6. The online health monitoring method for a prefabricated track filling structure according to claim 1, characterized in that: S3 specifically includes the following steps: S301: Data preprocessing and feature extraction, used to preprocess noise and extract key features, specifically including: Strain data cleaning: A moving average filter with a window size of 5 points is applied to the discrete strain data to eliminate random noise. Feature extraction includes peak locations, denoted here as... a 1 Starting point and a 2 Termination point, strain amplitude ε(x), Used to calculate the growth rate of the time derivative; Vibration data conversion: Extracting the dominant frequency from frequency domain data The peak detection algorithm identifies the main frequency value, and its features include: main frequency offset Δ. The spectrum energy distribution is then input into a machine learning model; S302: Crack width calculation, discrete strain data integration to quantify crack width, the formula is: ； in a 1 and a 2 The peak positions are specifically the starting and ending points of strain exceeding the benchmark value, which is set at 90% of the ultimate tensile strength of concrete. ε(x) The strain distribution at different locations on the track slab is represented by numerical integration of discrete point data. Formula for identifying peak regions: When ε(x) > Baseline value, mark a 1 and a 2 ; Numerical Integration: Calculating integral values ​​using the Python SciPy library The unit is mm; Calibration and verification: Compared with visual inspection, the error is controlled within <5%; S303: Crack depth determination, stress wave dominant frequency shift, assessment of crack depth, formula for the dominant frequency of normal concrete vibration: ; in For wave speed, The width of the track slab; The formula for the frequency shift when a crack occurs is: ; in The horizontal distance between the striking surface and the crack; The judgment method is as follows: Calculation benchmark : Main frequency without damage; Measurement offset When there is a crack, f1 is obtained through spectral analysis; Compare differences: Δ If Δ If the crack depth is greater than a preset threshold, then... It can be approximated as Combine a pre-trained support vector machine machine learning model to establish a Δ and The mapping is used to improve accuracy. S304: Application of machine learning models. Machine learning establishes the mapping relationship between strain and vibration characteristics and damage, specifically including: Model training: The classification model is trained using historical data. Input features such as peak amplitude and main frequency shift are used to output the damage type. The model uses SVM. Online data is input into the model to generate diagnostic reports. The model parameters are updated monthly to adapt to environmental changes.

Images