Road and bridge deck snow melting and deicing monitoring method and system based on DAS

Abstract

The invention discloses a DAS-based road and bridge surface snow melting and deicing monitoring method and system, and the method comprises the steps: laying a high-sensitivity sensing optical cable along a road and bridge surface through a distributed optical fiber sound wave sensing technology, and collecting a vibration signal generated by the interaction of tires and a road surface in a vehicle driving process in real time; carrying out noise reduction processing on the collected original time domain signal, and extracting an effective vibration response component; the method comprises the following steps: performing spectral analysis on vibration signals in an unfrozen state by adopting fast Fourier transform and short-time Fourier transform to obtain a spectral component set which is used for establishing a spectral feature reference library; according to the invention, continuous and real-time monitoring of the icing state of the road and bridge surface is realized through distributed optical fiber sound wave sensing. Compared with a traditional point type temperature sensor, the road and bridge surface icing state identification device has the advantages of being wide in monitoring range, high in spatial resolution, high in anti-interference capacity and the like, can achieve high-time-efficiency identification of the road and bridge surface icing state, and provides reliable data support for automatic control of a road and bridge surface snow melting and deicing system.

Description

Technical Field

[0001] This invention relates to the field of road and bridge engineering pavement safety monitoring and snow melting and de-icing control technology, and more specifically, to a road and bridge pavement snow melting and de-icing monitoring method and system based on DAS. Background Technology

[0002] Existing snow melting and de-icing systems for road and bridge surfaces mainly rely on meteorological parameters (such as temperature, humidity, and precipitation) or surface temperature sensors for ice monitoring and early warning. However, such systems have the following main technical problems: (1) Strong detection lag. Monitoring methods based on temperature or humidity can only reflect macroscopic meteorological changes and cannot reflect the microscopic icing state of the road surface in real time. There are often situations where thin ice has formed on the road surface but the sensor system has not yet triggered an alarm, resulting in untimely de-icing response.

[0003] (2) Insufficient spatial resolution. Traditional point sensors are sparsely deployed and can only monitor limited locations. They cannot achieve large-scale, continuous condition identification of road and bridge surfaces, which can easily lead to monitoring blind spots.

[0004] (3) Weak anti-interference ability. Environmental factors such as vehicle load vibration, wind noise and equipment mechanical vibration can easily interfere with temperature sensors, leading to false alarms or missed alarms.

[0005] (4) Monitoring results are difficult to quantify and evaluate. Most existing technologies rely on empirical thresholds to judge freezing risk and lack a quantitative indicator system based on the evolution of signal characteristics, making it difficult to achieve automated and precise early warning and control.

[0006] In recent years, Distributed Acoustic Sensing (DAS) technology has utilized optical fibers as a continuously distributed sensing medium. By demodulating the phase or intensity changes of backscattered light in the fiber, it achieves highly sensitive detection of vibrations, acoustic waves, and structural responses along the pipeline. This technology boasts advantages such as high spatial resolution, resistance to electromagnetic interference, and long-distance monitoring, and has been widely applied in fields such as oil and gas pipeline safety monitoring, seismic detection, and structural health monitoring, providing new sensing methods and technological foundations for road and bridge condition monitoring. Summary of the Invention

[0007] This invention provides a method and system for monitoring snow melting and de-icing on road and bridge surfaces based on DAS. It constructs a real-time monitoring and active de-icing linkage system for the icing state of road and bridge surfaces, which can realize intelligent processing of the entire process from signal acquisition, feature extraction, feature verification to risk warning, so as to solve the technical problems of existing road and bridge surface de-icing systems such as slow response, low monitoring resolution and poor anti-interference.

[0008] According to one aspect of the present invention, a method for monitoring snow melting and de-icing on road and bridge surfaces based on DAS is provided, comprising the following steps: Step 1: Real-time acquisition of vibration data generated by the interaction between the tires and the road surface during vehicle operation using distributed fiber optic acoustic sensing technology, followed by noise reduction processing and extraction of vibration response components. Step 2: Perform spectral analysis on the vibration response components in the unfrozen state using Fast Fourier Transform and Short-Time Fourier Transform to obtain the spectral component set. Step 3: Extract three characteristic parameters from the spectral component set in Step 2: low-frequency energy ratio, spectral centroid frequency, and frequency mutation rate, and establish a static benchmark library and a dynamic benchmark library. Step 4: The vibration signal is collected in real time. Short-time Fourier transform sliding analysis is performed on the collected vibration signal to dynamically calculate the dynamic characteristic parameters. Step 5: The dynamic feature parameter results are combined with the Fast Fourier Transform to verify the static features; Step 6: Calculate the Mahalanobis distance between the real-time signal feature vector and the reference library feature vector in Step 3 to quantitatively determine the degree of deviation of the spectral energy distribution.

[0009] Based on the above scheme, in step 1, the vibration data generated by the interaction between the tires and the road surface during vehicle operation is collected in real time using distributed fiber optic acoustic sensing technology, including: Step 11: The sensing optical cable is laid in an S-shaped reciprocating pattern within the cement leveling layer of the road and bridge surface. Step 12: The sensing optical cable and the heat exchange pipe of the snow melting and de-icing system are arranged in parallel to cover the entire width of the road and bridge surface. Step 13: Send an optical pulse into the sensing optical cable and receive the Rayleigh scattering echo signal. Obtain the continuous vibration response along the length of the optical cable based on the amount of phase change of the scattered light. Step 14: Establish the original time-domain signal dataset corresponding to time and strain rate based on the vibration response.

[0010] Based on the above scheme, in step 11, the vibration data is denoised and the vibration response components are extracted, which specifically includes: Step 111: The original time-domain signal is subjected to multi-scale wavelet decomposition transformation using the db wavelet basis to suppress environmental random noise and enhance the vibration signal response characteristics generated by the interaction between the tire and the road surface. Step 112: Determine the maximum decomposition scale n based on the number of signal sampling points N. max， To distinguish the characteristic components of different frequency bands; ; Step 113: Decompose the original time-domain signal into n levels, and use wavelet denoising algorithm to reconstruct the signal to provide high signal-to-noise ratio input data for subsequent spectrum analysis.

[0011] Based on the above scheme, step 2, which is the preferred option, specifically includes: Step 21: Perform a fast Fourier transform on the collected vibration signal in the unfrozen state to obtain the amplitude spectrum and power spectrum distribution of the signal in the full frequency range; Step 22: Perform short-time Fourier transform on the same signal to divide the signal into multiple overlapping time windows. Perform Fourier transform on each time window to obtain instantaneous spectrum information that changes with time. Perform energy normalization and amplitude-frequency feature extraction on the spectrum components of different time windows to form a set of spectrum components containing energy distribution of multiple time windows and multiple frequency points. Step 23: The set of spectral components reflects the spectral energy distribution characteristics and time-varying patterns of the road and bridge surface under vehicle load in the uniced state.

[0012] Based on the above scheme, step 3 preferably includes the following: Step 31: Define the low-frequency band as 0-500Hz and the mid-high frequency band as 500Hz-3kHz; calculate the energy distribution curves of different frequency ranges based on the spectral component set, and extract three characteristic parameters: low-frequency band energy proportion, spectral centroid frequency, and frequency change rate. Step 32: Establish a static benchmark library by using the statistical mean and standard deviation of the feature parameters obtained from the Fast Fourier Transform, and establish a dynamic benchmark library by using the mean and standard deviation of the sliding window feature parameters obtained from the Short Time Fourier Transform.

[0013] Based on the above scheme, the preferred method is as follows: the formulas for calculating the low-frequency energy ratio, the spectral centroid frequency, and the frequency abrupt change rate in step 31 are as follows: Low-frequency energy ratio (R) = Low-frequency energy (L) / Mid-to-high frequency energy (MH); Where: L is the sum of the squares of the amplitudes of all frequency points in the low-frequency band, and MH is the sum of the squares of the amplitudes of all frequency points in the mid-to-high-frequency band; Spectral centroid frequency : ; in: For frequency, The amplitude (energy) at frequency f is the square of the amplitude. This represents the highest analysis frequency. Frequency mutation rate : ; Where: Δt is the time difference.

[0014] Based on the above scheme, step 4 specifically includes: Step 41: The real-time acquired vibration signal is segmented according to a set sliding time window, and an overlap rate of 50% to 75% is set between adjacent time windows to improve time resolution. Step 42: Perform a short-time Fourier transform on the signal within each sliding time window to obtain the instantaneous spectral distribution within each time window; Step 43: Based on the instantaneous spectrum distribution, calculate the real-time values ​​of the energy proportion of the mid-to-high frequency band, the frequency of the spectrum centroid, and the frequency change rate to form a dynamic feature parameter set containing time series information; Step 44: Suppress instantaneous noise fluctuations using a time series smoothing algorithm or a weighted moving average method to obtain the continuous variation curve of the characteristic parameters.

[0015] Based on the above scheme, step 5 specifically includes: Step 51: Perform sliding comparison analysis on the dynamic characteristic parameter results obtained in the real-time monitoring stage. When the dynamic characteristics show instantaneous abnormal fluctuations, select the original vibration signal of the corresponding time window to perform fast Fourier transform to obtain the global spectrum distribution. Step 52: Compare the fast Fourier transform spectral characteristics of the time window with the average spectral characteristics in the static reference library, and calculate the deviation of the energy proportion of the mid-to-high frequency band and the frequency of the spectral centroid. When the characteristics obtained by the short-time Fourier transform are consistent with the characteristics obtained by the fast Fourier transform within the set tolerance range, the dynamic characteristic change is determined to be the true structural response. If the difference exceeds the tolerance threshold, it is determined to be local noise interference and is removed.

[0016] Based on the above scheme, step 6 preferably includes the following: Step 61: Combine the three characteristic parameters calculated in the real-time monitoring phase—the proportion of energy in the mid-to-high frequency band, the frequency of the spectral centroid, and the frequency change rate—into a real-time signal feature vector. Step 62: Extract the corresponding reference eigenvector and covariance matrix from the dynamic reference library; calculate the Mahalanobis distance value based on the difference between the real-time signal eigenvector and the reference eigenvector. : ; ; in: For real-time signal feature vectors, The mean of the baseline eigenvectors, The covariance matrix of the baseline eigenvectors, Set a threshold for Mahalanobis distance. The number of feature vectors, It is a chi-square distribution of the squared Mahalanobis distance.

[0017] This invention provides a DAS-based monitoring system for snow melting and de-icing on road and bridge surfaces, comprising: The data acquisition module is used to lay high-sensitivity distributed sensing optical cables along the road and bridge surface, and to use distributed optical fiber acoustic wave sensing technology to collect vibration signals generated by the interaction between the tires and the road surface during vehicle operation in real time, forming a raw time domain signal dataset. The signal preprocessing module is used to perform wavelet noise reduction on the original time-domain signal dataset to reduce environmental noise and structural noise interference, and extract effective vibration response components containing vehicle load response characteristics. The time-frequency analysis module is used to perform spectral analysis on a large number of signals in the unfrozen state using fast Fourier transform and short-time Fourier transform, and obtain a set of spectral components that include energy distribution characteristics, time-varying characteristics and spectral stability. The feature extraction module is used to extract feature parameters such as the low-frequency energy ratio, the frequency of the spectral centroid, and the frequency change rate based on the set of spectral components, and to form a multi-dimensional feature vector characterizing the vibration characteristics of the road and bridge surface. The benchmark library creation module is used to create static and dynamic benchmark libraries based on characteristic parameters under different environmental conditions, in order to describe the standard vibration characteristics of road and bridge surfaces in the non-icing state. The real-time monitoring module is used to perform short-time Fourier transform sliding analysis on the vibration signals acquired in real time during system operation, dynamically calculate real-time characteristic parameters, and generate corresponding real-time feature vectors. The static verification module is used to verify and compare the real-time feature parameter results with the static spectrum results obtained by the fast Fourier transform, so as to suppress local noise interference and improve the reliability and accuracy of the monitoring features. The decision-making and early warning module is used to calculate the Mahalanobis distance between the real-time feature vector and the feature vector of the benchmark library to quantitatively determine the degree of deviation of the spectrum energy distribution. When the Mahalanobis distance exceeds the set threshold and continues for more than three time windows, the road and bridge surface freezing risk warning is automatically triggered, and the snow melting and de-icing system is activated to perform de-icing operations.

[0018] This invention discloses a method and system for monitoring snow melting and de-icing on road and bridge surfaces based on distributed optical fiber acoustic sensing (DAS) technology. High-sensitivity sensing optical cables are continuously laid along the road and bridge surface to achieve real-time acquisition and continuous monitoring of vibration signals from the interaction between tires and the road surface during vehicle movement. Unlike traditional point-based temperature or visual sensing methods, the DAS system features full-line distribution and continuous coverage, enabling real-time perception of the icing state of the road and bridge surface at any time and any location of the optical cable. This provides a continuous, high-frequency, and highly timely data foundation for early identification of icing.

[0019] Meanwhile, this invention performs spectral analysis on the signal using Fast Fourier Transform and Short-Time Fourier Transform, extracting key characteristic parameters such as the proportion of energy in the mid-to-high frequency band, the frequency of the spectral centroid, and the frequency abrupt change rate. It then establishes static and dynamic benchmark libraries to achieve quantitative expression of these characteristic indicators. Furthermore, the system employs the Mahalanobis distance algorithm to calculate the difference between the real-time signal feature vector and the benchmark feature vector, enabling quantitative determination of icing risk and overcoming the uncertainty problems of traditional empirical judgment and threshold identification. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a schematic flowchart of the DAS-based monitoring method for snow melting and de-icing on road and bridge surfaces according to the present invention. Figure 2 This is a schematic diagram of the optical cable laying for the DAS-based road and bridge surface snow melting and de-icing monitoring system of the present invention. Detailed Implementation

[0021] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0022] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of a descriptive feature, integral, step, operation, element, and / or component, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or sets.

[0023] To keep the drawings concise, only the parts relevant to the invention are shown schematically in each figure, and they do not represent the actual structure of the product. Furthermore, for ease of understanding, in some figures, only one of components with the same structure or function is shown schematically, or only one is labeled. In this document, "one" can mean not only "only one" but also "more than one".

[0024] It should also be further understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0025] In the embodiments shown in the accompanying drawings, the directional indications (such as up, down, left, right, front, and back) used to explain the structure and movement of the various components of the invention are relative rather than absolute. These descriptions are appropriate when these components are in the positions shown in the drawings. If the descriptions of the positions of these components change, these directional indications also change accordingly.

[0026] Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the specific implementation methods of the present invention will be described below with reference to the accompanying drawings. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings and other implementation methods can be obtained based on these drawings without any creative effort.

[0028] Please see Figure 1 and combined Figure 2 As shown, a method for monitoring snow melting and de-icing on road and bridge surfaces based on DAS according to the present invention includes the following steps: Step 1: Real-time acquisition of vibration data generated by the interaction between the tires and the road surface during vehicle operation using distributed fiber optic acoustic sensing technology, followed by noise reduction processing and extraction of vibration response components. Specifically, firstly, the sensing optical cable is laid in an S-shaped reciprocating pattern within the cement leveling layer of the road and bridge surface; secondly, the sensing optical cable is arranged parallel to the heat exchange pipes of the snow melting and de-icing system, uniformly covering the entire width of the road and bridge surface to achieve full coverage monitoring of the entire road and bridge surface area; light pulses are sent into the optical cable and Rayleigh scattering echo signals are received, and the continuous vibration response along the length of the optical cable is obtained based on the phase change of the scattered light; and a raw time-domain signal dataset corresponding to time and strain rate is established based on the vibration response.

[0029] In step 1, the vibration data is denoised and the vibration response components are extracted, which includes the following: Step 111: The original time-domain signal is subjected to multi-scale wavelet decomposition transformation using the db wavelet basis to suppress environmental random noise and enhance the vibration signal response characteristics generated by the interaction between the tire and the road surface. Step 112: Determine the maximum decomposition scale nmax based on the number of signal sampling points N, in order to distinguish the characteristic components of different frequency bands; ; Step 113: Decompose the original time-domain signal into n levels, and use wavelet denoising algorithm to reconstruct the signal to provide high signal-to-noise ratio input data for subsequent spectrum analysis.

[0030] Step 2: Perform spectral analysis on the vibration response components in the unfrozen state using Fast Fourier Transform and Short-Time Fourier Transform to obtain the spectral component set. The specific implementation method is as follows. Step 21: Perform a fast Fourier transform on the collected vibration signal in the unfrozen state to obtain the amplitude spectrum and power spectrum distribution of the signal in the full frequency range, so as to reflect the overall spectral characteristics of the signal under steady-state conditions. Step 22: Perform short-time Fourier transform on the same signal to divide the signal into multiple overlapping time windows. Perform Fourier transform on each time window to obtain instantaneous spectrum information that changes with time. Perform energy normalization and amplitude-frequency feature extraction on the spectrum components of different time windows to form a set of spectrum components containing energy distribution of multiple time windows and multiple frequency points. Step 23: The spectral component set comprehensively reflects the spectral energy distribution characteristics and time-varying patterns of the road and bridge surface under vehicle loads in the uniced state, providing a frequency domain data foundation for subsequent feature parameter extraction and benchmark library construction.

[0031] Step 3: Extract three characteristic parameters from the spectral component set in Step 2: low-frequency energy ratio, spectral centroid frequency, and frequency mutation rate, and establish a static benchmark library and a dynamic benchmark library. Step 3 includes the following details: Step 31: Define the low-frequency band as 0-500Hz and the mid-high frequency band as 500Hz-3kHz; calculate the energy distribution curves of different frequency ranges based on the spectral component set, and extract three characteristic parameters: low-frequency band energy proportion, spectral centroid frequency, and frequency abrupt change rate; among them, the low-frequency band energy proportion is used to characterize the energy proportion of the signal in the mid-high frequency band, the spectral centroid frequency is used to reflect the shift of the overall frequency distribution of the signal, and the frequency abrupt change rate is used to measure the rate of change of the frequency centroid.

[0032] Step 32: Establish a static benchmark library by statistically analyzing the mean and standard deviation of the feature parameters obtained from the Fast Fourier Transform, and establish a dynamic benchmark library by analyzing the mean and standard deviation of the sliding window feature parameters obtained from the Short Time Fourier Transform. The static benchmark library reflects the global spectral feature distribution under the non-icing state, while the dynamic benchmark library reflects the time-varying features under the non-icing state, providing a benchmark basis for feature matching and deviation determination in the real-time monitoring stage.

[0033] The formulas for calculating the low-frequency energy proportion, the spectral centroid frequency, and the frequency abrupt change rate are as follows: Low-frequency energy ratio (R) = Low-frequency energy (L) / Mid-to-high frequency energy (MH); Where: L is the sum of the squares of the amplitudes of all frequency points in the low-frequency band, and MH is the sum of the squares of the amplitudes of all frequency points in the mid-to-high-frequency band; Spectral centroid frequency : ; in: For frequency, The amplitude (energy) at frequency f is the square of the amplitude. This represents the highest analysis frequency. Frequency mutation rate : ; Where: Δt is the time difference.

[0034] Step 4: The vibration signal is collected in real time. Short-time Fourier transform sliding analysis is performed on the collected vibration signal to dynamically calculate the dynamic characteristic parameters. The specific steps are as follows: Step 41, the vibration signal acquired in real time is segmented according to the set sliding time window, and the overlap rate between adjacent time windows is set to 50% to 75% to improve the time resolution. Step 42: Perform a short-time Fourier transform on the signal within each sliding time window to obtain the instantaneous spectral distribution within each time window; Step 43: Based on the instantaneous spectrum distribution, calculate the real-time values ​​of the energy proportion of the mid-to-high frequency band, the frequency of the spectrum centroid, and the frequency change rate to form a dynamic feature parameter set containing time series information; Step 44: Suppress instantaneous noise fluctuations using a time series smoothing algorithm or a weighted moving average method to obtain the continuous variation curve of the characteristic parameters.

[0035] Among them, dynamic characteristic parameters are used to reflect the spectral response changes of road and bridge structures under vehicle loads and environmental changes in real time, providing real-time data support for subsequent anomaly identification and risk warning based on characteristic deviations.

[0036] Step 5: Combine the dynamic feature parameter results with Fast Fourier Transform to perform static feature verification. The specific steps are as follows; Step 51: Perform sliding comparison analysis on the dynamic characteristic parameter results obtained in the real-time monitoring stage. When the dynamic characteristics show instantaneous abnormal fluctuations, select the original vibration signal of the corresponding time window to perform fast Fourier transform to obtain the global spectrum distribution. Step 52: Compare the Fast Fourier Transform (FFT) spectral characteristics of the time window with the average spectral characteristics in the static benchmark library to calculate the deviation of the mid-to-high frequency energy ratio and the spectral centroid frequency. When the characteristics obtained from the Short-Time Fourier Transform (SFT) and the characteristics obtained from the Fast Fourier Transform (FFT) are consistent within a set tolerance range, the dynamic characteristic change is determined to be a true structural response. If the difference exceeds the tolerance threshold, it is determined to be local noise interference and is removed. The tolerance range can be obtained through system adaptive learning or experimental calibration. Through the above static characteristic verification, the accuracy correction of dynamic monitoring results and local noise suppression are achieved, ensuring the authenticity and stability of the spectral change trend.

[0037] Step 6: Calculate the Mahalanobis distance between the real-time signal feature vector and the reference library feature vector in Step 3 to quantitatively determine the degree of deviation in the spectral energy distribution. Detailed steps are as follows: Step 61: Combine the three characteristic parameters calculated in the real-time monitoring phase—the proportion of energy in the mid-to-high frequency band, the frequency of the spectral centroid, and the frequency change rate—into a real-time signal feature vector. Step 62: Extract the corresponding reference eigenvector and covariance matrix from the dynamic reference library; calculate the Mahalanobis distance value based on the difference between the real-time signal eigenvector and the reference eigenvector. : ; ; in: For real-time signal feature vectors, The mean of the baseline eigenvectors, The covariance matrix of the baseline eigenvectors, Set a threshold for Mahalanobis distance. The number of feature vectors, It is a chi-square distribution of the squared Mahalanobis distance.

[0038] Mahalanobis distance is used to comprehensively measure the overall deviation between multiple features, avoiding misjudgments caused by fluctuations in a single feature. When the Mahalanobis distance exceeds a set threshold, the system determines that the current spectral energy distribution has deviated abnormally, and records the time window number and deviation magnitude, providing a quantitative basis for subsequent continuous judgment and early warning triggering.

[0039] This invention can also set an alarm threshold; when the Mahalanobis distance value exceeds D... 阈值 If the freezing continues for more than three time windows, an automatic warning for the risk of road and bridge surface freezing will be triggered, and the snow melting and de-icing system will be activated in conjunction with it.

[0040] Furthermore, the present invention also provides a DAS-based monitoring system for snow melting and de-icing on road and bridge surfaces, comprising: The data acquisition module is used to lay high-sensitivity distributed sensing optical cables along the road and bridge surface, and to use distributed optical fiber acoustic wave sensing technology to collect vibration signals generated by the interaction between the tires and the road surface during vehicle operation in real time, forming a raw time domain signal dataset. The signal preprocessing module is used to perform wavelet noise reduction on the original time-domain signal dataset to reduce environmental noise and structural noise interference, and extract effective vibration response components containing vehicle load response characteristics. The time-frequency analysis module is used to perform spectral analysis on a large number of signals in the unfrozen state using fast Fourier transform and short-time Fourier transform, and obtain a set of spectral components that include energy distribution characteristics, time-varying characteristics and spectral stability. The feature extraction module is used to extract feature parameters such as the low-frequency energy ratio, the frequency of the spectral centroid, and the frequency change rate based on the set of spectral components, and to form a multi-dimensional feature vector characterizing the vibration characteristics of the road and bridge surface. The benchmark library creation module is used to create static and dynamic benchmark libraries based on characteristic parameters under different environmental conditions, in order to describe the standard vibration characteristics of road and bridge surfaces in the non-icing state. The real-time monitoring module is used to perform short-time Fourier transform sliding analysis on the vibration signals acquired in real time during system operation, dynamically calculate real-time characteristic parameters, and generate corresponding real-time feature vectors. The static verification module is used to verify and compare the real-time feature parameter results with the static spectrum results obtained by the fast Fourier transform, so as to suppress local noise interference and improve the reliability and accuracy of the monitoring features. The decision-making and early warning module is used to calculate the Mahalanobis distance between the real-time feature vector and the feature vector of the benchmark library to quantitatively determine the degree of deviation of the spectrum energy distribution. When the Mahalanobis distance exceeds the set threshold and continues for more than three time windows, the road and bridge surface freezing risk warning is automatically triggered, and the snow melting and de-icing system is activated to perform de-icing operations.

[0041] This invention, based on distributed fiber optic acoustic sensing technology, constructs a real-time monitoring and active de-icing linkage system for road and bridge surface icing, enabling intelligent processing throughout the entire process from signal acquisition, feature extraction, feature verification to risk warning. Compared with existing icing monitoring methods that rely on point-based temperature sensors or image recognition, this invention has the following significant advantages: (1) Real-time monitoring of icing status This invention employs distributed optical fiber acoustic sensing (DAS) technology, continuously laying high-sensitivity sensing optical cables along the road and bridge surface to achieve real-time acquisition and continuous monitoring of vibration signals from the interaction between tires and the road surface during vehicle movement. Unlike traditional point-based temperature or visual sensing methods, the DAS system features full-line distribution and continuous coverage, enabling real-time perception of the icing state of the road and bridge surface at any time and any location of the optical cable, providing a continuous, high-frequency, and highly timely data foundation for early identification of icing.

[0042] (2) Quantification of monitoring results This invention performs spectral analysis on signals using Fast Fourier Transform and Short-Time Fourier Transform, extracting key characteristic parameters such as the proportion of energy in the mid-to-high frequency band, the frequency of the spectral centroid, and the frequency abrupt change rate. It establishes static and dynamic benchmark libraries to achieve quantitative expression of these characteristic indicators. Furthermore, the system employs the Mahalanobis distance algorithm to calculate the difference between the real-time signal feature vector and the benchmark feature vector, enabling quantitative determination of icing risk and overcoming the uncertainty problems of traditional empirical judgment and threshold identification.

[0043] (3) High spatial resolution By continuously laying distributed optical fiber sensing cables along the road and bridge surface, the system can achieve high-precision detection with meter-level spatial resolution. Compared with discrete sensor deployment monitoring, this invention can achieve icing status identification across the entire road section without blind spots, significantly improving the spatial integrity and distribution uniformity of monitoring, and providing accurate data support for identifying icing risks on long-span bridges and long-distance roads.

[0044] (4) Strong anti-interference ability The system employs a wavelet denoising algorithm to perform multi-scale filtering on the original vibration signal, effectively suppressing interference from vehicle load, wind vibration, and environmental noise. Through joint analysis of short-time Fourier transform and fast Fourier transform, complementary feature extraction in the time and frequency domains is achieved, significantly improving the robustness and stability of signal recognition. Combined with a joint verification mechanism of static and dynamic features, instantaneous abnormal fluctuations can be automatically identified and eliminated, ensuring the authenticity and repeatability of spectral change trends.

[0045] (5) Intelligent linkage de-icing system to form an active prevention and control mechanism When the Mahalanobis distance value in the real-time monitoring results exceeds the set threshold and remains so for more than three time windows, the system automatically triggers an ice risk warning and activates the snow melting and de-icing system, achieving closed-loop control from monitoring and identification to proactive de-icing. This mechanism transforms the de-icing system from a "passive response" to a "proactive prevention and control" system, enabling automatic intervention in the early stages of icing formation, significantly reducing the risk of road icing and improving the operational safety of roads and bridges.

[0046] (6) Low false alarm rate, enabling intelligent self-calibration This invention employs a judgment mechanism that combines dynamic feature parameters with static feature verification, effectively reducing misjudgments and false alarms caused by environmental changes or instantaneous noise. The system features an adaptive feature library update function, capable of automatically correcting the baseline model based on seasonal changes, structural differences, and long-term monitoring data, achieving long-term stable operation and intelligent self-calibration, thereby ensuring high reliability and continuous accuracy of the system under complex environmental conditions.

[0047] This invention proposes a DAS-based method and system for monitoring snow melting and de-icing on road and bridge surfaces. The system uses high-sensitivity sensing optical cables laid along the road and bridge surface to collect vibration signals from the interaction between tires and the road surface during vehicle movement in real time. Wavelet denoising technology is used to reduce environmental interference, and Fast Fourier Transform (FFT) and Short-Time Fourier Transform (STFT) are combined to extract three characteristic parameters: low-frequency energy proportion, spectral centroid frequency, and frequency abrupt change rate, establishing a benchmark library for the uniced state. The system performs time-frequency analysis on the real-time signals, calculates the Mahalanobis distance between the signals and the feature vectors in the benchmark library, and automatically triggers an icing warning and activates the snow melting and de-icing system when the deviation exceeds a set threshold and persists for more than three time windows. This invention achieves highly sensitive detection and intelligent early warning of icing conditions by identifying the characteristic pattern of abrupt changes in tire-road contact vibration from mid-high frequencies to low frequencies, effectively improving road safety and reducing system energy consumption. It has significant engineering implications for improving the energy efficiency and optimizing the operation of road and bridge snow melting and de-icing systems.

[0048] Finally, the method described in this application is merely a preferred embodiment and is not intended to limit the scope of protection of this invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for monitoring snow melting and de-icing on road and bridge surfaces based on DAS, characterized in that, Includes the following steps: Step 1: Real-time acquisition of vibration data generated by the interaction between the tires and the road surface during vehicle operation using distributed fiber optic acoustic sensing technology, followed by noise reduction processing and extraction of vibration response components. Step 2: Perform spectral analysis on the vibration response components in the unfrozen state using Fast Fourier Transform and Short-Time Fourier Transform to obtain the spectral component set. Step 3: Extract three characteristic parameters from the spectral component set in Step 2: low-frequency energy ratio, spectral centroid frequency, and frequency mutation rate, and establish a static benchmark library and a dynamic benchmark library. Step 4: The vibration signal is collected in real time. Short-time Fourier transform sliding analysis is performed on the collected vibration signal to dynamically calculate the dynamic characteristic parameters. Step 5: The dynamic feature parameter results are combined with the Fast Fourier Transform to verify the static features; Step 6: Calculate the Mahalanobis distance between the real-time signal feature vector and the reference library feature vector in Step 3 to quantitatively determine the degree of deviation of the spectral energy distribution.

2. The method for monitoring snow melting and de-icing on road and bridge surfaces based on DAS as described in claim 1, characterized in that, In step 1, distributed fiber optic acoustic sensing technology is used to collect vibration data generated by the interaction between the tires and the road surface during vehicle operation in real time, including: Step 11: The sensing optical cable is laid in an S-shaped reciprocating pattern within the cement leveling layer of the road and bridge surface. Step 12: The sensing optical cable and the heat exchange pipe of the snow melting and de-icing system are arranged in parallel to cover the entire width of the road and bridge surface. Step 13: Send an optical pulse into the sensing optical cable and receive the Rayleigh scattering echo signal. Obtain the continuous vibration response along the length of the optical cable based on the amount of phase change of the scattered light. Step 14: Establish the original time-domain signal dataset corresponding to time and strain rate based on the vibration response.

3. The method for monitoring snow melting and de-icing on road and bridge surfaces based on DAS as described in claim 1, characterized in that, In step 1, the vibration data is denoised and the vibration response components are extracted, which includes: Step 111: The original time-domain signal is subjected to multi-scale wavelet decomposition transformation using the db wavelet basis to suppress environmental random noise and enhance the vibration signal response characteristics generated by the interaction between the tire and the road surface. Step 112: Determine the maximum decomposition scale n based on the number of signal sampling points N. max， To distinguish the characteristic components of different frequency bands; ； Step 113: Decompose the original time-domain signal into n levels, and use wavelet denoising algorithm to reconstruct the signal to provide high signal-to-noise ratio input data for subsequent spectrum analysis.

4. The method for monitoring snow melting and de-icing on road and bridge surfaces based on DAS as described in claim 1, characterized in that, Step 2 specifically includes: Step 21: Perform a fast Fourier transform on the collected vibration signal in the unfrozen state to obtain the amplitude spectrum and power spectrum distribution of the signal in the full frequency range; Step 22: Perform short-time Fourier transform on the same signal to divide the signal into multiple overlapping time windows. Perform Fourier transform on each time window to obtain instantaneous spectrum information that changes with time. Perform energy normalization and amplitude-frequency feature extraction on the spectrum components of different time windows to form a set of spectrum components containing energy distribution of multiple time windows and multiple frequency points. Step 23: The set of spectral components reflects the spectral energy distribution characteristics and time-varying patterns of the road and bridge surface under vehicle loads in the uniced state.

5. The method for monitoring snow melting and de-icing on road and bridge surfaces based on DAS as described in claim 4, characterized in that, Step 3 includes the following in detail: Step 31: Define the low-frequency band as 0-500Hz and the mid-high frequency band as 500Hz-3kHz; calculate the energy distribution curves of different frequency ranges based on the spectral component set, and extract three characteristic parameters: low-frequency band energy proportion, spectral centroid frequency, and frequency change rate. Step 32: Establish a static benchmark library by using the statistical mean and standard deviation of the feature parameters obtained from the Fast Fourier Transform, and establish a dynamic benchmark library by using the mean and standard deviation of the sliding window feature parameters obtained from the Short Time Fourier Transform.

6. The method for monitoring snow melting and de-icing on road and bridge surfaces based on DAS as described in claim 4, characterized in that, The formulas for calculating the low-frequency energy ratio, spectral centroid frequency, and frequency abrupt change rate in step 31 are as follows: Low-frequency energy ratio (R) = Low-frequency energy (L) / Mid-to-high frequency energy (MH); Where: L is the sum of the squares of the amplitudes of all frequency points in the low-frequency band, and MH is the sum of the squares of the amplitudes of all frequency points in the mid-to-high-frequency band; Spectral centroid frequency : ； in: For frequency, The amplitude (energy) at frequency f is the square of the amplitude. This represents the highest analysis frequency. Frequency mutation rate : ； Where: Δt is the time difference.

7. The method for monitoring snow melting and de-icing on road and bridge surfaces based on DAS as described in claim 1, characterized in that, Step 4 includes the following in detail: Step 41: The real-time acquired vibration signal is segmented according to a set sliding time window, and an overlap rate of 50% to 75% is set between adjacent time windows to improve time resolution. Step 42: Perform short-time Fourier transform on the signal within each sliding time window to obtain the instantaneous spectral distribution within each time window; Step 43: Based on the instantaneous spectrum distribution, calculate the real-time values ​​of the energy proportion of the mid-to-high frequency band, the frequency of the spectrum centroid, and the frequency change rate to form a dynamic feature parameter set containing time series information; Step 44: Suppress instantaneous noise fluctuations using a time series smoothing algorithm or a weighted moving average method to obtain the continuous variation curve of the characteristic parameters.

8. The method for monitoring snow melting and de-icing on road and bridge surfaces based on DAS as described in claim 1, characterized in that, Step 5 includes the following in detail: Step 51: Perform sliding comparison analysis on the dynamic characteristic parameter results obtained in the real-time monitoring stage. When the dynamic characteristics show instantaneous abnormal fluctuations, select the original vibration signal of the corresponding time window to perform fast Fourier transform to obtain the global spectrum distribution. Step 52: Compare the fast Fourier transform spectral characteristics of the time window with the average spectral characteristics in the static reference library, and calculate the deviation of the energy proportion of the mid-to-high frequency band and the frequency of the spectral centroid. When the characteristics obtained by the short-time Fourier transform are consistent with the characteristics obtained by the fast Fourier transform within the set tolerance range, the dynamic characteristic change is determined to be the true structural response. If the difference exceeds the tolerance threshold, it is determined to be local noise interference and is removed.

9. A method for monitoring snow melting and de-icing on road and bridge surfaces based on DAS as described in claim 1, characterized in that, Step 6 includes the following in detail: Step 61: Combine the three characteristic parameters calculated in the real-time monitoring phase—the energy proportion of the mid-to-high frequency band, the frequency of the spectral centroid, and the frequency change rate—into a real-time signal feature vector. Step 62: Extract the corresponding reference eigenvector and covariance matrix from the dynamic reference library; calculate the Mahalanobis distance value based on the difference between the real-time signal eigenvector and the reference eigenvector. : ； ； in: For real-time signal feature vectors, The mean of the baseline feature vectors, The covariance matrix of the baseline eigenvectors, Set a threshold for Mahalanobis distance. The number of feature vectors, It is a chi-square distribution of the squared Mahalanobis distance.

10. A monitoring system for snow melting and de-icing systems on road and bridge surfaces, characterized in that, include: The data acquisition module is used to lay high-sensitivity distributed sensing optical cables along the road and bridge surface, and to use distributed optical fiber acoustic wave sensing technology to collect vibration signals generated by the interaction between the tires and the road surface during vehicle operation in real time, forming a raw time domain signal dataset. The signal preprocessing module is used to perform wavelet noise reduction on the original time-domain signal dataset to reduce environmental noise and structural noise interference, and extract effective vibration response components containing vehicle load response characteristics. The time-frequency analysis module is used to perform spectral analysis on a large number of signals in the unfrozen state using fast Fourier transform and short-time Fourier transform, and obtain a set of spectral components that include energy distribution characteristics, time-varying characteristics and spectral stability. The feature extraction module is used to extract feature parameters such as the low-frequency energy ratio, the frequency of the spectral centroid, and the frequency change rate based on the set of spectral components, and to form a multi-dimensional feature vector characterizing the vibration characteristics of the road and bridge surface. The benchmark library creation module is used to create static and dynamic benchmark libraries based on characteristic parameters under different environmental conditions, in order to describe the standard vibration characteristics of road and bridge surfaces in the non-icing state. The real-time monitoring module is used to perform short-time Fourier transform sliding analysis on the vibration signals acquired in real time during system operation, dynamically calculate real-time characteristic parameters, and generate corresponding real-time feature vectors. The static verification module is used to verify and compare the real-time feature parameter results with the static spectrum results obtained by the fast Fourier transform, so as to suppress local noise interference and improve the reliability and accuracy of the monitoring features. The decision-making and early warning module is used to calculate the Mahalanobis distance between the real-time feature vector and the feature vector of the benchmark library to quantitatively determine the degree of deviation of the spectrum energy distribution. When the Mahalanobis distance exceeds the set threshold and continues for more than three time windows, the road and bridge surface freezing risk warning is automatically triggered, and the snow melting and de-icing system is activated to perform de-icing operations.

Images