Road cavity abnormal signal feature extraction and analysis method based on distributed optical fibers

Abstract

The invention discloses a road cavity abnormal signal feature extraction and analysis method based on distributed optical fibers, and relates to the technical field of optical fiber sensing. The method comprises the following steps: constructing a sensing system by using existing distributed optical fibers buried along two sides of a road; exciting earth surface vibration waves by combining an active seismic source or a traffic load, and acquiring and storing dynamic disturbance signals in the optical fiber in real time through a data acquisition module; the collected original signals are subjected to preprocessing such as de-noising and normalization; extracting typical frequency domain and energy features from the preprocessed signals by using a time-frequency domain analysis method, and performing feature fusion in combination with correlation analysis of multichannel signals; by comparing signal features with a road structure model and combining an anomaly detection algorithm, road health state assessment and accurate positioning of cavity anomaly are realized, and finally early warning information is generated according to a positioning result and uploaded to an early warning management platform, thereby providing technical support for road maintenance and risk management and control.

Description

Technical Field

[0001] This invention belongs to the field of distributed optical fiber sensing technology, specifically relating to a method for extracting and analyzing abnormal signal features of road voids based on distributed optical fiber. Background Technology

[0002] With the accelerating pace of urbanization, the traffic flow and usage intensity of road infrastructure are increasing year by year, leading to a higher incidence of road structural defects. In particular, road cavities caused by foundation settlement, underground pipe leakage, and soil erosion have become a significant threat to road safety. Traditional road inspection methods, such as manual inspections and ground-penetrating radar detection, have limitations such as low detection frequency and limited coverage, making it difficult to detect and accurately locate potential underground cavities in a timely manner.

[0003] In recent years, distributed fiber optic sensing technology has been increasingly applied to the health monitoring of structures such as bridges and oil and gas pipelines due to its advantages of long-distance monitoring, resistance to electromagnetic interference, and high sensitivity. However, applying distributed fiber optic sensing technology to the detection of cavities in urban roads, and combining it with signal preprocessing and feature extraction to achieve real-time early warning and precise location of potential road cavity hazards, still faces technical challenges such as complex sensor signals, high environmental noise, and concealed anomaly features. Therefore, a method for feature extraction and analysis of abnormal road cavity signals based on distributed fiber optics is needed to improve the intelligence level and management efficiency of road structural hazard detection, and to provide strong technical support for the safe operation and scientific maintenance of urban roads. Summary of the Invention

[0004] This invention addresses the problems existing in the prior art by providing a method for extracting and analyzing abnormal road cavity signals based on distributed optical fiber. Through distributed optical fiber sensing technology and multi-feature signal analysis, it achieves abnormal detection and accurate location of road cavities.

[0005] To address the above technical problems, this invention provides the following technical solution: a method for extracting and analyzing abnormal road cavity signals based on distributed optical fibers, comprising the following steps:

[0006] S1. Based on a distributed optical fiber sensing system, the sensing technology used combines the dynamic disturbance in the optical fiber caused by the ground vibration wave to collect the original optical fiber vibration signal.

[0007] S2. Preprocess the original fiber vibration signal, including noise reduction, normalization, and filtering operations;

[0008] S3. Extract time-frequency domain features from the preprocessed fiber vibration signal and fuse the multi-channel signals to generate a comprehensive feature vector.

[0009] S4. By comparing signal features with road structure models and applying anomaly detection algorithms, road health assessment and cavity location are achieved.

[0010] Furthermore, the distributed optical fiber sensing system in step S1 above consists of existing distributed optical fibers buried longitudinally along the road, a data acquisition and demodulation instrument, etc.; the sensing technology includes Brillouin time-domain reflectometry, Rayleigh time-domain reflectometry, or phase-sensitive time-domain reflectometry.

[0011] Furthermore, in the aforementioned step S1, vibration signals inside the road structure are excited by active excitation, vehicle load, or environmental disturbance to form detectable dynamic disturbances, and the vibration signals along the optical fiber are continuously or periodically collected using a terminal demodulator.

[0012] Furthermore, in the aforementioned step S2: denoising the original optical fiber vibration signal includes the following sub-steps:

[0013] S2.1. Construct a reference noise spectrum based on the statistical characteristics of noise, and dynamically subtract the noise component from the amplitude spectrum of the original optical fiber vibration signal. According to the additive assumption, the relationship between the denoised signal, the noise, and the noisy signal is as follows:

[0014] y(n) = x(n) + d(n)

[0015] Where y(n) is the acquired noisy signal, x(n) is the denoised signal, and d(n) is the noise;

[0016] S2.2, Convert to frequency domain:

[0017] Y(ω)=X(ω)+D(ω)

[0018]

[0019] Where Y(ω) represents the spectrum of the noisy signal y(n), X(ω) represents the spectrum of the denoised signal x(n), D(ω) represents the spectrum of the noise d(n), and |Y(ω)| is the amplitude value. For signal phase;

[0020] S2.3 The denoised signal is obtained by subtracting noise from the noisy signal, as shown in the following formula:

[0021]

[0022] in, This represents an estimated value of the spectrum of the denoised model. S2.4 represents the estimated value of the noise spectrum. If the value is negative, perform a half-wave rectification on the denoised signal, as shown in the following formula:

[0023]

[0024] Furthermore, in step S2 above, a sliding window energy normalization process is used, and the window length L is adaptively adjusted according to the transient characteristics of the signal, as shown in the following formula:

[0025]

[0026] Where x(t) is the amplitude of the original signal at time t, and ∈ is a small quantity to prevent division by zero.

[0027] Furthermore, the aforementioned step S3 includes the following sub-steps:

[0028] S3.1 Perform feature analysis on the preprocessed fiber vibration signal in the frequency domain to extract typical frequency distribution features for identifying abnormal vibration modes.

[0029] S3.2 Perform statistical calculations on the signal energy and analyze the energy change trends in different time periods or different channels;

[0030] S3.3 By calculating the minimum cross-correlation coefficient between each channel, the propagation range and consistency of abnormal signals are determined, thereby enhancing the detection and location capabilities of road cavity anomalies.

[0031] Furthermore, in step S3.2 above, the short-time energy of the signal is calculated to characterize the energy distribution characteristics of the signal, as shown in the following formula:

[0032]

[0033] Where W is the window length, x 2 (k) represents the square of the signal amplitude at time k, where k represents the time sampling point and t represents the current time sampling point.

[0034] Furthermore, in step S3.3 above, the minimum cross-correlation coefficient ρ between each channel is calculated. xy =min(|R xy (τ)|), used to determine the accuracy of abnormal signal location, is as follows: ρ xy =min(|R xy (τ)|), where, x(t) and y(t) are two channel signals, and τ is the time delay.

[0035] Furthermore, in the aforementioned step S4, the road anomaly location is based on the extracted abnormal signal features, combined with the spatial distribution information of the optical fiber laid along the road, to spatially map the abnormal signal, locate and mark the void or structural defect area, and provide accurate location information for subsequent verification and maintenance.

[0036] Furthermore, the aforementioned method for extracting and analyzing abnormal road cavity signals based on distributed optical fiber also includes centralized management and dynamic updating of abnormal signals and positioning results collected by the distributed optical fiber sensing system. By setting road anomaly risk levels and threshold standards, it enables automatic comparison and graded early warning of detected abnormal signals, and displays the distribution of road anomalies, risk points and corresponding geographical coordinates in real time through a visual interface.

[0037] Compared with the prior art, the beneficial technical effects of the present invention using the above technical solution are as follows:

[0038] 1. This invention uses existing communication optical fibers that are continuously deployed along the road structure to achieve accurate perception of anomalies such as tiny voids and settlement along the road, avoiding the coverage blind spots of traditional point sensors.

[0039] 2. This invention employs an improved spectral subtraction denoising method, which adaptively modifies the dynamic threshold, effectively extracting weak hole signals under background noise, enhancing the characteristics of the effective signal, and improving resolution.

[0040] 3. This invention combines the analysis of the time-frequency domain characteristics, energy changes, and multi-channel signal correlation of vibration signals to extract abnormal signal features from multiple perspectives, effectively distinguishing disturbances caused by cavities, realizing cross-verification of cavity anomalies, and improving detection accuracy and anti-interference ability.

[0041] 4. This invention can realize the sensing function using existing communication optical fibers without the need for additional deployment, which reduces the deployment and maintenance costs and is suitable for large-scale promotion and application. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the overall process of the method described in this invention.

[0043] Figure 2 This is a schematic diagram illustrating the correlation characteristics of the method described in this invention. Detailed Implementation

[0044] To better understand the technical content of the present invention, specific embodiments are described below in conjunction with the accompanying drawings.

[0045] In this invention, various aspects of the invention are described with reference to the accompanying drawings, in which numerous illustrative embodiments are shown. Embodiments of the invention are not limited to those depicted in the drawings. It should be understood that the invention is implemented through any of the various concepts and embodiments described above, as well as the concepts and embodiments described in detail below, because the concepts and embodiments disclosed herein are not limited to any particular implementation. Furthermore, some aspects of the invention disclosed may be used alone or in any suitable combination with other aspects of the invention disclosed.

[0046] like Figure 1 As shown, this invention provides a method for feature extraction and analysis of road cavity anomaly signals based on distributed optical fiber, the steps of which are as follows:

[0047] S1. Based on a distributed optical fiber sensing system, the sensing technology used combines the dynamic disturbance in the optical fiber caused by ground vibration waves to collect the original optical fiber vibration signal.

[0048] In this embodiment, existing distributed optical fibers buried longitudinally along both sides or below the road are used and connected to a terminal demodulator to form a distributed optical fiber sensing network that can be monitored in real time. Optical time domain reflectance (OTDR) or optical coherent reflectance (Φ-OTDR) and other technologies are used to detect vibration disturbance signals of optical fibers along the road.

[0049] Ground vibrations induced by active excitation (such as hammering or seismic source vehicles) or traffic loads generate elastic waves that propagate along the road surface. These waves act on optical fibers buried along the road, creating detectable dynamic disturbances. The terminal demodulator collects the disturbance signals along the optical fiber in real time and uploads the collected data to a data acquisition server.

[0050] S2. Preprocess the original fiber optic vibration signal, including denoising, normalization, and filtering. Specifically: Preprocess the acquired fiber optic signal. First, use spectral subtraction adaptive denoising to estimate the background noise spectrum in real time and denoise the target signal, reducing the impact of environmental noise, traffic interference, and other non-target signals, thereby improving the signal-to-noise ratio of the target signal. The denoising of the original fiber optic vibration signal includes the following sub-steps:

[0051] S2.1. Construct a reference noise spectrum based on the statistical characteristics of noise, and dynamically subtract the noise component from the amplitude spectrum of the original optical fiber vibration signal. According to the additive assumption, the relationship between the denoised signal, the noise, and the noisy signal is as follows:

[0052] y(n) = x(n) + d(n)

[0053] Where y(n) is the acquired noisy signal, x(n) is the denoised signal, and d(n) is the noise;

[0054] S2.2, Convert to frequency domain:

[0055] Y(ω)=X(ω)+D(ω)

[0056]

[0057] Where Y(ω) represents the spectrum of the noisy signal y(n), X(ω) represents the spectrum of the denoised signal x(n), D(ω) represents the spectrum of the noise d(n), and |Y(ω)| is the amplitude value. For signal phase;

[0058] S2.3 The denoised signal is obtained by subtracting noise from the noisy signal, as shown in the following formula:

[0059]

[0060] in, This represents an estimated value of the spectrum of the denoised model. This represents an estimate of the noise spectrum.

[0061] S2.4, based on If the value is negative, perform a half-wave rectification on the denoised signal, as shown in the following formula:

[0062]

[0063] Secondly, the denoised signal is further normalized to unify the amplitude of the signal collected from different measurement points and time periods into a standardized range, eliminating amplitude fluctuation differences caused by factors such as excitation intensity and fiber optic burial depth, thereby ensuring the consistency of feature extraction and anomaly detection results of subsequent multi-channel data.

[0064]

[0065] Where x(t) is the amplitude of the original signal at time t, and ∈ is a small quantity to prevent division by zero.

[0066] S3. For the preprocessed fiber vibration signal, extract time-frequency domain features and fuse the multi-channel signals to generate a comprehensive feature vector. This includes the following sub-steps:

[0067] S3.1. For the preprocessed fiber optic signal, typical frequency distribution features are extracted using methods such as Fourier transform and power spectral density estimation. By analyzing the typical frequency distribution, normal traffic vibrations and abnormal cavity vibrations can be distinguished. S3.2. Statistical calculations are performed on the energy variation trend of the signal. A sliding window analysis is conducted using the short-time energy method to obtain the energy changes of each channel signal within different time periods. Key features such as abrupt energy changes and persistently high-energy sections are identified to reflect the degree of local anomalies or damage to the internal road structure. The short-time energy of the signal is calculated to characterize its energy distribution characteristics.

[0068]

[0069] Where W is the window length, x 2 (k) represents the square of the signal amplitude at time k, where k represents the time sampling point and t represents the current time sampling point.

[0070] S3.3 By calculating the minimum cross-correlation coefficient between each channel, the propagation range and consistency of abnormal signals are determined, thereby enhancing the detection and location capabilities of road cavity anomalies.

[0071] Cross-correlation analysis of multi-channel fiber optic vibration signals is performed to calculate the correlation coefficients between different channels, assessing the synchronicity and propagation direction of anomalous disturbances across multiple channels. A significant decrease in the correlation between signals from multiple channels within a certain region may indicate local structural loosening or voids, thus improving the reliability of anomaly location identification. The correlation between multi-channel signals is evaluated using a cross-correlation function.

[0072]

[0073] Here, x(t) and y(t) are two channel signals, and τ is the time delay. Calculate the minimum cross-correlation coefficient ρ between each channel. xy =min(|R xy (τ)|) is used to determine the accuracy of the location of abnormal signals. For example Figure 2 As shown, by calculating the cross-correlation coefficients between each channel, the spatial correlation distribution of the signal is analyzed; the channel with the smallest correlation coefficient is used as the criterion for abnormal channels, thereby achieving preliminary identification and location of abnormal signal positions.

[0074] S4. By comparing signal features with road structure models and applying anomaly detection algorithms, road health assessment and cavity location are achieved. Based on the extracted abnormal signal features, the model is compared with existing health status models. Combined with threshold analysis and anomaly detection algorithms, it is determined whether structural anomalies such as cavities or voids exist in the signal. Furthermore, based on the spatial location information of the distributed optical fibers buried longitudinally along the road, spatiotemporal mapping of the abnormal signals is performed to achieve precise location of abnormal events.

[0075] In this embodiment, a signal energy attenuation model with propagation distance is established based on the location information of existing optical fibers.

[0076] E(d)=E0e -αd +β

[0077] Road anomaly localization is achieved through least squares fitting.

[0078] Where E(d) represents the signal energy value at a distance d, E0 represents the initial energy, α is the attenuation coefficient, and β represents the ambient noise floor energy.

[0079] This invention also includes an early warning management platform for centralized storage, management, and dynamic updating of abnormal signals and location results collected by the distributed fiber optic sensing system. Based on pre-set road anomaly risk levels and threshold standards, it can automatically compare and intelligently classify detected abnormal signals to provide early warnings, and automatically generate alarm information and handling suggestions for different risk levels. Furthermore, through a visual interface, the platform can display the real-time distribution of anomalies along the entire road, key risk points, and their corresponding geographical coordinates. It also supports integration with external road maintenance and management systems, facilitating timely dispatch, verification, and handling by maintenance personnel, further enhancing the intelligence and efficiency of road health monitoring and operation and maintenance management.

[0080] According to another aspect of the present invention, the present invention provides a distributed optical fiber sensing demodulator, which is used to modulate and demodulate Rayleigh scattered light signals and obtain corresponding vibration signals for subsequent signal processing.

[0081] While the present invention has been described above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention shall be determined by the claims.

Claims

1. A method for extracting and analyzing road hollow anomaly signal features based on distributed optical fiber, characterized in that, The method comprises the following steps: S1, based on the distributed optical fiber sensing system, using the sensing technology, combining the ground vibration wave to cause the dynamic disturbance in the optical fiber, collecting the original optical fiber vibration signal; S2, preprocessing the original optical fiber vibration signal, including denoising, normalization, filtering operation; S3, for the preprocessed optical fiber vibration signal, extracting the time-frequency domain feature, and fusing the multi-channel signal to generate a comprehensive feature vector; S4, by comparing the signal characteristics with the road structure model, and applying the anomaly detection algorithm, realizing the road health assessment and cavity positioning. 2.The method of claim 1, wherein, In step S1, the distributed optical fiber sensing system is composed of existing distributed optical fibers buried along the road, data acquisition and demodulation instruments, etc.; the sensing technology includes Brillouin optical time domain reflection, Rayleigh optical time domain reflection or phase-sensitive time domain reflection. 3.The method of claim 1, wherein, In step S1, the vibration signal inside the road structure is excited by active excitation, vehicle load or environmental disturbance to form a detectable dynamic disturbance, and the terminal demodulation instrument is used to continuously or timingly collect the vibration signal along the optical fiber. 4.The method of claim 1, wherein, In step S2, the original optical fiber vibration signal is denoised, including the following substeps: S2.1, based on the noise statistical characteristics, a reference noise spectrum is constructed, the noise component is dynamically subtracted from the original optical fiber vibration signal amplitude spectrum, and according to the additive assumption, the relationship among the denoised signal, the noise and the noisy signal is as follows: y(n)=x(n)+d(n) Wherein, y(n) is the collected noisy signal, x(n) is the denoised signal, and d(n) is the noise; S2.2, convert to frequency domain: Y(ω)=X(ω)+D(ω) Wherein, Y(ω) represents the spectrum of the noisy signal y(n), X(ω) represents the spectrum of the de-noised signal x(n), D(ω) represents the spectrum of the noise d(n), |Y(ω)| is the amplitude value, is the signal phase; S2.3, the denoised signal is obtained by subtracting the noise from the noisy signal, as follows: wherein denotes an estimate of the noise spectrum, denotes an estimate of the noise spectrum, S2.4, based on is negative, a half-wave rectification is performed on the de-noised signal, as follows: 5.The method of claim 1, wherein, In step S2, sliding window energy normalization processing is adopted, and the window length L is adaptively adjusted according to the signal transient characteristics, as follows: Wherein, x(t) is the amplitude of the original signal at time t, and ∈ is a small amount to prevent division by zero. 6.The method of claim 1, wherein, Step S3 includes the following substeps: S3.1, the preprocessed optical fiber vibration signal is analyzed in the frequency domain to extract the typical frequency distribution characteristics, which are used to identify abnormal vibration patterns; S3.2, the energy of the signal is calculated statistically to analyze the energy change trend in different time periods or different channels; S3.3, by calculating the minimum cross-correlation coefficient between channels, the propagation range and consistency of abnormal signals are judged to enhance the detection and positioning ability of road cavity anomalies. 7.The method of claim 6, wherein, In step S3.2, the short-time energy of the signal is calculated to represent the energy distribution characteristics of the signal, as follows: where W is the window length, x 2 (k) represents the signal amplitude squared at time k, k represents the time sample point, and t represents the current time sample point. 8.The method of claim 6, wherein, In step S3.3, the minimum cross-correlation coefficient p between the channels is calculated xy = min ( | R xy (τ) | ), for judging the accuracy of the abnormal signal position as follows: p xy = min ( | R xy (τ) | ), wherein x(t) and y(t) are two channel signals, and τ is a time delay. 9.The method of claim 6, wherein, In step S4, the road anomaly positioning is based on the extracted abnormal signal characteristics, combined with the spatial distribution information of the optical fiber along the road, the abnormal signal is spatially mapped, and the cavity or structural defect area is positioned and marked to provide accurate position information for subsequent verification and maintenance. 10.The method of claim 1, wherein It also includes centralized management and dynamic updating of the abnormal signals collected by the distributed optical fiber sensing system, positioning results, setting road anomaly risk level and threshold standard, realizing automatic comparison and hierarchical early warning of the detected abnormal signals, and real-time display of road anomaly distribution, risk points and corresponding geographic coordinates through the visual interface.

Images