A method and system for evaluating the risk of urban road collapse

Abstract

The present application discloses a method and system for evaluating the risk of urban road collapse, which relates to the field of road detection and monitoring, and includes: collecting surface microseismic signals through the passive source surface wave linear array exploration method; preprocessing the collected surface microseismic signals; performing dispersion imaging on the preprocessed surface microseismic signals to obtain dispersion curves; taking the dispersion curve at the initial time as a reference, and calculating the phase velocity reduction rate of the dispersion curve in a future observation time period relative to the reference; dividing the collapse risk level of the urban road according to the phase velocity reduction rate. Aiming at the low accuracy of the risk evaluation of urban road collapse in the prior art, the present application improves the evaluation accuracy by eliminating the noise interference and amplitude difference in the surface microseismic signals; adopting an improved dispersion imaging method, comprehensively using the information of the fundamental mode and high-order mode dispersion curves to obtain a high-precision surface wave dispersion curve covering the entire frequency band, etc.

Description

Technical Field

[0001] This application relates to the field of road monitoring, and particularly to a method and system for evaluating the risk of urban road collapse. Background Art

[0002] With the continuous advancement of urbanization, urban roads, as an important part of the urban transportation network, have attracted much attention for their safety and durability. Diseases of urban roads, especially road collapses, not only affect the normal operation of transportation but also may threaten the safety of pedestrians and vehicles. Therefore, regular evaluation of the collapse risk of urban roads to timely detect and warn of road diseases is of great significance for ensuring urban traffic safety and extending the service life of roads.

[0003] Currently, the commonly used methods for identifying the risk of urban road collapse mainly include borehole exploration, static cone penetration test, dynamic cone penetration test, and ground penetrating radar, etc. Although these methods can obtain local information of the road structure, there are some common problems: First, the exploration range is limited, making it difficult to comprehensively evaluate the overall condition of the road; Second, some methods require damage to the road structure, affecting the normal use of the road; Third, the on-site operation is complex, time-consuming and laborious, and the evaluation efficiency is low; Fourth, the evaluation results are highly subjective and the degree of quantification is not high. Therefore, there is an urgent need for a non-destructive, fast, and quantitative method for evaluating the collapse risk applicable to the urban road environment.

[0004] The surface wave exploration technology is a non-destructive testing method based on the dispersion characteristics of surface waves. By analyzing the variation law of the propagation speed of surface waves on the ground surface with frequency (i.e., the dispersion curve), the physical and mechanical parameters of the underground medium can be inverted, and then the integrity and stability of the engineering structure can be evaluated. In recent years, the surface wave exploration technology has been preliminarily applied in road engineering, but mainly concentrated in fields such as highways and tunnels, and its application in the evaluation of the risk of urban road collapse is still very few. This is mainly because the urban road environment is complex, the traffic flow is large, and the surface wave signals collected are often interfered by noise, with a low signal-to-noise ratio, making it difficult to obtain high-quality dispersion curves, resulting in low evaluation accuracy. In addition, traditional surface wave dispersion imaging methods mostly use the fundamental mode dispersion curve and are difficult to accurately characterize the dispersion characteristics of urban road structures. Therefore, how to obtain high-quality surface wave ground micro-motion signals in a complex urban road environment and accurately extract the dispersion curve information of the full frequency band is the key problem that needs to be solved for the application of surface wave exploration technology in the evaluation of the risk of urban road collapse.

[0005] For example, the related patent document CN118483743A discloses a method for extracting surface wave dispersion curves based on strain fields, including the steps of: extracting strain data from seismic surface microtremor signals; preprocessing the strain data according to the type of the seismic surface microtremor signals to obtain strain components; converting the strain components from the time-space domain to the frequency-space domain through Fourier transform; converting the strain components from the frequency-space domain to the frequency-phase velocity domain through frequency-Bessel transform to obtain the surface wave dispersion spectrum of the strain components; and picking up the surface wave dispersion curve according to the surface wave dispersion spectrum. However, this scheme uses Fourier transform and frequency-Bessel transform to extract the surface wave dispersion spectrum, mainly using the dispersion curve information of the fundamental mode, and it is difficult to accurately characterize the dispersion characteristics of urban road structures. Summary of the Invention

[0006] Aiming at the low accuracy of urban road collapse risk assessment in the prior art, the present application provides a method and system for urban road collapse risk assessment, which eliminates noise interference and amplitude differences in seismic surface microtremor signals; adopts an improved dispersion imaging method, comprehensively uses the dispersion curve information of the fundamental mode and high-order modes, and obtains a high-precision surface wave dispersion curve covering the entire frequency band, etc., thereby improving the assessment accuracy.

[0007] The object of the present application is achieved through the following technical solutions.

[0008] One aspect of the present application provides a method for urban road collapse risk assessment, including: S1, collecting seismic surface microtremor signals through a passive source surface wave linear array exploration method; the seismic surface microtremor signals include surface vertical vibration velocity and displacement; S2, preprocessing the collected seismic surface microtremor signals; S3, performing dispersion imaging on the preprocessed seismic surface microtremor signals to obtain a dispersion curve; S4, taking the dispersion curve at the initial time as a reference, and calculating the phase velocity reduction rate of the dispersion curve in a future observation time period relative to the reference; S5, according to the phase velocity reduction rate, determining the degree of change in the compactness state of soil bodies at different depths below the road, and thereby identifying the possibility of urban road collapse risk; at the same time, identifying the development characteristics of underground diseases causing collapse through the high-low order mode dispersion characteristics in the f-k domain of each road microtremor signal processing, and determining the road collapse risk.

[0009] Among them, the surface microtremor signal refers to the environmental vibration generated on the surface under natural conditions. This kind of vibration is usually caused by natural or human activities, such as wind, waves, traffic, industrial activities, etc. In this application, the surface vertical vibration velocity and displacement are collected as the surface microtremor signal through passive-source surface wave exploration. Dispersion imaging is a method of imaging the velocity structure of underground media using the dispersion characteristics of surface waves. The propagation velocity of surface waves is different at different frequencies, which is called the dispersion phenomenon. By processing and analyzing surface wave signals at different frequencies, a relationship curve between frequency (or period) and velocity (or wavelength), that is, the dispersion curve, can be obtained. Dispersion imaging is the process of extracting the dispersion curve from the surface microtremor signal and inversely inferring the underground velocity structure based on this. The phase velocity reduction rate characterizes the degree of change in the velocity structure of underground media. In this application, the initial dispersion curve is used as a reference, and the percentage of the velocity change of the dispersion curve relative to the reference after a certain period of time is calculated as the phase velocity reduction rate. Since the loosening of underground media will cause the reduction of surface wave velocity, the phase velocity reduction rate can indicate the change in the compactness state of underground soil. The larger the reduction rate, the looser the soil and the higher the collapse risk.

[0010] Further, S1, collecting the surface microtremor signal through the passive-source surface wave linear array exploration method, including: S11, selecting the survey line on the section to be evaluated according to the grade and traffic flow of the urban road; S12, setting the geophone layout parameters according to the road width and traffic flow, and arranging geophones linearly along the road direction on the survey line, and the value range of the adjacent geophone spacing is 1m to 3m; S13, using the passive-source surface wave exploration method for observation, setting the single observation duration, and obtaining the surface microtremor signal.

[0011] Further, S2, preprocessing the collected surface microtremor signal, including: S21, performing de-mean processing on the surface microtremor signal; S22, performing normalization processing on the surface microtremor signal after de-mean processing to eliminate the amplitude difference between different lanes and retain the up-and-down movement direction information of the surface vibration; S23, performing spectral whitening processing on the surface microtremor signal after normalization processing to suppress noise interference and obtain the preprocessed surface microtremor signal.

[0012] Further, S21, performing de-mean processing on the surface microtremor signal to eliminate the DC bias, including: calculating the mean value of each channel of the surface microtremor signal; subtracting the mean value of each channel of the surface microtremor signal from the corresponding mean value to obtain the surface microtremor signal after de-mean processing; adjusting the positive and negative signs of the surface microtremor signal after de-mean processing to be consistent with the original surface microtremor signal to retain the context movement direction information of the surface microtremor signal.

[0013] Among them, the context motion direction information of the surface microtremor signal refers to the vibration direction of the surface microtremor signal in the vertical direction, that is, the up-and-down motion trend of the surface. This information reflects the polarity or phase characteristics of the surface vibration.

[0014] Specifically, the positive and negative signs of the surface microtremor signal contain the context direction information of the surface motion and reflect the vibration direction of the surface. Traditional mean removal processing may change the positive and negative signs of the data, resulting in the loss or confusion of the motion direction information. In this embodiment, after the mean removal processing, the positive and negative signs of the processed data are adjusted to be consistent with the original data to ensure that the context direction information of the surface motion is retained. This helps to correctly interpret the physical process of the surface vibration and identify the trends and laws of the surface motion. The context motion direction information can also be used to distinguish different types of seismic waves, such as compressional waves and shear waves. The context motion directions of different types of waves are different.

[0015] Further, in S22, perform normalization processing on the surface microtremor signal after the mean removal processing, including: calculating the maximum absolute value of each channel of the surface microtremor signal after the mean removal processing; using the maximum absolute value to normalize the absolute value of each channel of the surface microtremor signal after the mean removal processing to the interval from 0 to 1; adjusting the positive and negative signs of the normalized surface microtremor signal to be consistent with the original surface microtremor signal to unify the amplitudes of the surface microtremor signals of different channels.

[0016] Further, in S23, perform spectral whitening processing on the normalized surface microtremor signal, including: dividing the frequency band of the normalized surface microtremor signal according to the formation conditions and surface wave characteristics of the target road to determine the main frequency range of the surface wave signal; performing two-dimensional Fourier transform on the surface microtremor signal after the frequency band division to transform the surface microtremor signal from the time-space domain to the frequency-wavenumber domain; in the frequency-wavenumber domain, set an elliptical filtering window according to the energy distribution characteristics of the surface wave signal, where the major axis of the elliptical filtering window is the first direction of the energy distribution of the surface wave signal, the minor axis is the second direction of the energy distribution of the surface wave signal, the first direction is the main direction, and the second direction is the secondary direction; use the set elliptical filtering window to filter the data in the frequency-wavenumber domain, and only retain the energy within the main frequency range of the surface wave signal within the elliptical window to suppress the noise component; perform inverse two-dimensional Fourier transform on the filtered data in the frequency-wavenumber domain to convert the data from the frequency-wavenumber domain back to the time-space domain to obtain the filtered surface microtremor signal; perform spectral whitening on the filtered surface microtremor signal, and only perform normalization processing on the frequency components within the main frequency range of the surface wave signal to obtain the preprocessed surface wave observation data.

[0017] Among them, the dominant frequency range refers to the frequency range where the energy of the surface wave signal is concentrated. Due to the dispersion phenomenon of surface waves during propagation, the propagation speeds of surface waves with different frequencies in the underground medium are different. Usually, the surface wave energy is mainly concentrated in a certain specific frequency range, and this frequency range is the dominant frequency range. The determination of the dominant frequency range needs to be judged according to the formation conditions of the target road and the characteristics of the surface waves.

[0018] The time - space domain refers to the data representation method with time and spatial position as independent variables. In this application, the original surface micro - motion signal is collected in the time - space domain, that is, the surface vibration information at different times and different positions is recorded. The data in the time - space domain can intuitively reflect the propagation process of waves in time and space.

[0019] The frequency - wavenumber domain refers to the data representation method with frequency and wavenumber as independent variables. By performing a two - dimensional Fourier transform on the data in the time - space domain, the data can be transformed from the time - space domain to the frequency - wavenumber domain. In the frequency - wavenumber domain, the data reflects the energy distribution characteristics of the signal in terms of frequency and wavenumber. Different frequencies and wavenumbers correspond to different wave speeds and propagation directions.

[0020] The main direction and the secondary direction are two directions that describe the energy distribution characteristics of the surface wave signal in the frequency - wavenumber domain. On the frequency - wavenumber spectrogram, the energy of the surface wave signal usually shows an elliptical distribution. The long - axis direction of the ellipse is called the main direction, representing the propagation direction where the surface wave signal energy is most concentrated; the short - axis direction of the ellipse is called the secondary direction, which is orthogonal to the main direction. By identifying the main direction and the secondary direction, the propagation characteristics of the surface wave signal in space can be determined.

[0021] Spectral whitening is a data - processing method for spectral equalization. Its purpose is to make the energy distribution of the signal relatively balanced at each frequency, suppress the peaks in the spectrum, and enhance the low - energy frequency components. In this application, spectral whitening processing is performed on the filtered surface micro - motion signal, and only the frequency components within the dominant frequency range of the surface wave signal are normalized to make the energies of the frequency components within the dominant frequency range equivalent. This can highlight the characteristics of the surface wave signal and suppress the influence of noise.

[0022] Specifically, on the one hand, traditional FK - domain filtering uses a fixed rectangular or circular window, which is difficult to adapt to the actual energy distribution characteristics of the surface wave signal in the frequency - wavenumber domain, resulting in poor filtering effects. In this application, according to the formation conditions of the target road and the characteristics of the surface waves, an elliptical filtering window is adaptively designed, and the long axis and short axis of the window are respectively aligned with the main direction and the secondary direction of the surface wave signal energy distribution, better matching the actual distribution of the surface wave signal. The adaptive elliptical filtering window can more accurately extract the surface wave signal, reduce noise interference, improve the signal quality and the reliability of subsequent analysis.

[0023] On the other hand, traditional FK domain filtering is prone to introducing artifacts. Especially when the energy distribution of the surface wave signal is irregular or there are multiple main directions, it is difficult for a fixed window to effectively separate the signal and noise. This application adopts an adaptive elliptical filtering window, which adjusts the window shape and size according to the actual energy distribution of the surface wave signal, can better suppress the noise component, and reduce the generation of artifacts. Through adaptive filtering, the signal-to-noise ratio of the surface wave signal can be effectively improved, providing more reliable data support for subsequent dispersion analysis and road collapse evaluation.

[0024] Furthermore, traditional spectral whitening processing usually normalizes the frequency components in the entire frequency range, which may amplify the noise components and affect the signal quality. Before spectral whitening processing, this application filters the data through an adaptive elliptical filtering window, only retaining the energy within the main frequency range of the surface wave signal. During spectral whitening processing, only the frequency components within the main frequency range of the filtered surface wave signal are normalized, avoiding the amplification of noise components and improving the pertinence and effectiveness of spectral whitening.

[0025] Preferably, a consistency correction step for supplementing the surface micro-motion signals of different channels is added: using the dispersion characteristics of surface waves, a suitable reference frequency is selected, and the phase difference of the surface micro-motion signals of each channel near this frequency is calculated. Then, based on this, the time offset of the data of different channels is corrected to make their travel times consistent. In addition, the amplitude attenuation law of surface waves can also be used to perform normalization compensation on the amplitudes of the data of different channels. Since the surface micro-motion signals of far channels have a longer propagation distance and more severe amplitude attenuation, a larger compensation coefficient needs to be given. Through the above consistency correction, the systematic deviation between the surface micro-motion signals of different channels can be eliminated, creating better conditions for coherent superposition in subsequent dispersion imaging. Specifically, using the dispersion characteristics of surface waves, a suitable reference frequency is selected, and the phase difference of the surface micro-motion signals of each channel after S23 processing near this frequency is calculated; according to the calculated phase difference, the time offset of the data of different channels after S23 processing is corrected to make their travel times consistent; using the amplitude attenuation law of surface waves, the amplitudes of the data of different channels after correction are normalized and compensated, and a larger compensation coefficient is given to the surface micro-motion signals of far channels.

[0026] Further, in S3, dispersion imaging is performed on the preprocessed surface micro-motion signals to obtain a dispersion curve, including: S31, performing a slowness-frequency domain transformation on the preprocessed surface micro-motion signals to obtain a slowness-frequency spectrum; S32, setting an energy threshold E in the slowness-frequency spectrum th(f), extract the local peak points of the slowness-frequency spectrum according to the frequency range and energy threshold as the reliable points of the surface wave dispersion curve; S33, based on the least squares method, fit the extracted reliable points to obtain the dispersion curves of the fundamental mode and the higher modes respectively; S34, perform smoothing processing on the fundamental mode dispersion curve to obtain the optimized fundamental mode dispersion curve; supplement the high-frequency band information of the optimized fundamental mode dispersion curve according to the higher mode dispersion curve to obtain the extended fundamental mode dispersion curve; splice the extended fundamental mode dispersion curve and the higher mode dispersion curve to obtain the final dispersion curve.

[0027] Specifically, the technical solutions for traditional extraction of dispersion curves usually adopt a single dispersion imaging method, such as the frequency-wavenumber (f-k) method, the slowness-frequency (p-f) method, etc., and then extract the dispersion curve through manual picking or simple peak searching. On the one hand, the dispersion maps obtained by traditional methods have low resolution. Especially in the high-frequency band, the energy distribution of the dispersion curve is relatively dispersed, making it difficult to accurately extract. This affects the integrity and reliability of the dispersion curve, especially the lack of high-frequency band information, which is crucial for the evaluation of the shallow structure of the road. Moreover, the urban road structure is complex, and higher modes are easily excited during the propagation of surface waves. Traditional solutions mainly focus on the fundamental mode and are difficult to effectively identify and extract higher mode information. The higher mode dispersion curve contains important information about the road structure and is of great value for comprehensively evaluating the road collapse risk.

[0028] This application uses the slowness-frequency domain transformation to obtain the slowness-frequency spectrum. Compared with the traditional frequency-wavenumber domain, the slowness-frequency domain can provide higher dispersion imaging resolution, especially in the high-frequency band, which is beneficial to extracting a complete and reliable dispersion curve. At the same time, in the slowness-frequency spectrum, this application sets an adaptive energy threshold E th (f), combined with local peak point extraction, to achieve automatic extraction of reliable points of the dispersion curve. This method improves the automation degree of dispersion curve extraction, reduces human intervention, and improves the extraction efficiency and consistency.

[0029] Secondly, this application uses the least squares method to fit the extracted reliable points to obtain the dispersion curves of the fundamental mode and the higher modes respectively. The fitting process takes into account the physical characteristics of the dispersion curve, effectively improving the continuity and smoothness of the dispersion curve, and laying a good foundation for subsequent surface wave inversion.

[0030] Finally, for the fundamental mode dispersion curve, the present application performs smoothing processing to optimize the curve quality. At the same time, the high-order mode dispersion curve is used to supplement the high-frequency band information and expand the frequency range of the fundamental mode dispersion curve. Finally, the expanded fundamental mode dispersion curve and the high-order mode dispersion curve are spliced to obtain the final dispersion curve. This processing process makes full use of the complementary information of the fundamental and high-order modes and comprehensively reflects the dispersion characteristics of the road structure.

[0031] On the other hand, in the traditional multi-channel surface wave analysis method, the dispersion curve is usually directly extracted in the slowness-frequency domain. However, the dispersion curve in the slowness-frequency domain is affected by the spatial aliasing effect. Especially in the high-frequency band, the resolution and accuracy of the dispersion curve may be reduced.

[0032] In the present application, by performing a two-dimensional Fourier transform on the slowness-frequency domain energy distribution map and converting the data to the wavenumber-frequency domain, the influence of the spatial aliasing effect can be effectively eliminated. In the wavenumber-frequency domain, the surface wave energy of different modes is concentrated in different wavenumber ranges, and the dispersion curve becomes clearer and more separated, which is beneficial to extracting high-precision and high-resolution dispersion curves. The extraction result of the dispersion curve in the wavenumber-frequency domain has a higher signal-to-noise ratio and resolution compared with the result directly extracted in the slowness-frequency domain. Especially in the high-frequency band, the aliasing and interference of different mode dispersion curves can be effectively avoided, and the reliability of the dispersion curve is improved.

[0033] Furthermore, the energy threshold E th (f) is set by the following formula: f ≤ f0, f > f0, where α1, n1, b1, α2, b2, c2 and f0 are parameters pre-fitted according to the surface microtremor signal, and w(f) is the data quality factor; the frequency f0 is the segmentation point. When the frequency f is less than or equal to f0, a power function form is adopted; when the frequency f is greater than f0, an exponential function form is adopted. The design of this piecewise function can better adapt to the energy distribution characteristics in different frequency ranges and improve the flexibility of the threshold function.

[0034] Within each frequency band, the energy threshold function contains multiple undetermined parameters, such as α1, n1, b1, α2, b2, c2, etc. These parameters can be determined by pre-fitting the surface micro-motion signals to adapt to the data characteristics of different roads or regions. A data quality factor w(f) is introduced as a multiplicative modulation term for the energy threshold function. The value range of the data quality factor w(f) is (0, 1], which reflects the quality and reliability of the data at frequency f. When the signal-to-noise ratio SNR(f) is high, w(f) is close to 1 and has little impact on the energy threshold; when the signal-to-noise ratio is low, w(f) decreases, and the energy threshold is appropriately reduced to adapt to the change in data quality. In the calculation formula of the data quality factor w(f), d and e are empirical parameters that can be adjusted according to the characteristics of the actual data.

[0035] SNR(f) is the signal-to-noise ratio at frequency f, which can be estimated by time-frequency analysis or other signal processing methods.

[0036] Furthermore, w(f) is calculated according to the following formula: where f represents the frequency in Hz; f0 is the segmentation frequency that divides the frequency domain into a low-frequency band and a high-frequency band in Hz; α1 is the scale parameter of the power function in the low-frequency band, controlling the vertical stretching of the power function; n1 is the exponential parameter of the power function in the low-frequency band, controlling the shape and growth rate of the power function; b1 is the vertical offset parameter of the power function in the low-frequency band, controlling the vertical position of the power function; α2 is the scale parameter of the exponential function in the high-frequency band, controlling the vertical stretching of the exponential function; b2 is the exponential parameter of the exponential function in the high-frequency band, controlling the shape and growth rate of the exponential function; c2 is the vertical offset parameter of the exponential function in the high-frequency band, controlling the vertical position of the exponential function; w(f) represents the data quality factor, reflecting the quality and reliability of the data at frequency f, with a value range of (0, 1]; d is the empirical parameter in the calculation formula of the data quality factor, controlling the influence degree of the signal-to-noise ratio on the data quality factor; e is the empirical parameter in the calculation formula of the data quality factor, controlling the sensitivity of the signal-to-noise ratio to the data quality factor; SNR(f) represents the signal-to-noise ratio at frequency f, reflecting the quality of the signal at this frequency. The higher the signal-to-noise ratio, the closer the data quality factor is to 1.

[0037] Compared with the prior art, the advantages of this application are as follows:

[0038] On the one hand, the traditional processing of removing the mean value and maximum-minimum normalization of the surface micro-motion signals erases the positive and negative characteristics of the surface micro-motion signals. However, the positive and negative characteristics of the surface vertical vibration velocity or displacement data precisely reflect the up and down movement directions of the surface vibration and contain important physical information. Erasing the positive and negative characteristics may affect the accuracy and physical meaning of subsequent dispersion imaging.

[0039] In this application, when performing the mean removal process, the mean value of each surface microtremor signal can be extracted first, and its positive and negative signs can be recorded. Then, after the mean is removed, the sign of the original mean is restored to the processed data. In this way, both the DC bias can be eliminated, and the positive and negative characteristics of the surface microtremor signal can be retained. For the maximum-minimum normalization process, it can be changed to normalize the absolute value of the surface microtremor signal. After normalization, the normalized result is multiplied by the sign of the original data to restore its positive and negative characteristics. This improvement can eliminate the amplitude differences between different channels while retaining the up and down movement direction information of the surface vibration.

[0040] On the other hand, spectral whitening highlights the middle and high-frequency components rich in formation information by balancing the amplitudes of different frequency components. However, in actual surface microtremor signals, the high-frequency components far from the dominant frequency of the surface wave usually consist mainly of noise. Simply normalizing the entire frequency spectrum may inadvertently amplify the originally weak high-frequency noise, thus interfering with the effective surface wave signal.

[0041] In this application, before performing the spectral whitening process, a preliminary frequency band division is first performed on the surface microtremor signal. According to the formation conditions and surface wave characteristics of the target area, the dominant frequency range of the surface wave signal is roughly determined. During the spectral whitening process, only the frequency components within this dominant frequency range are normalized, while the high-frequency or low-frequency components far from the dominant frequency maintain their original amplitudes. In addition, before spectral whitening, the surface microtremor signal can be subjected to transform domain filtering to preliminarily suppress the noise components outside the surface wave frequency band. Then, spectral whitening is performed on the filtered data. This can avoid the risk of amplifying weak high-frequency noise and more specifically highlight the characteristics of the effective surface wave signal.

[0042] Furthermore, traditional dispersion imaging often uses the fundamental mode dispersion curve, mainly focusing on the extraction of the surface wave dispersion curve of the fundamental mode, while ignoring the information of the higher mode surface waves. Under certain complex geological conditions, the dispersion characteristics of the higher mode surface waves may contain important information about the formation structure and properties. If only relying on the fundamental mode surface waves, it may lead to the incompleteness and limitations of the evaluation results.

[0043] This application comprehensively utilizes the information of the fundamental mode and higher mode dispersion curves to obtain a high-precision surface wave dispersion curve covering the entire frequency band through dispersion curve splicing. At the same time, the improved dispersion imaging method introduces a piecewise non-linear energy threshold function and a data quality factor, improving the robustness of the reliable point extraction of the dispersion curve and overcoming the problems such as the discontinuity of the dispersion curve and susceptibility to noise interference existing in the existing methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] This application will be further described in the form of exemplary embodiments, which will be described in detail through the drawings. These embodiments are not restrictive. In these embodiments, the same numbers represent the same structures, where:

[0045] Figure 1 It is a schematic diagram of an exemplary application scenario of a method for evaluating the risk of urban road collapse shown in some embodiments of the present application;

[0046] Figure 2 It is a schematic diagram of a linear array observation system shown in some embodiments of the present application;

[0047] Figure 3 It is a schematic diagram of a data acquisition scenario shown in some embodiments of the present application;

[0048] Figure 4 It is the first data dispersion diagram in the first section of the present application;

[0049] Figure 5 It is the first data dispersion picking in the first section of the present application;

[0050] Figure 6 It is the first data dispersion curve in the first section of the present application;

[0051] Figure 7 It is the dispersion imaging judgment criterion for the first section of the present application;

[0052] Figure 8 It is the dispersion imaging judgment criterion for the second section of the present application;

[0053] Figure 9 It is the dispersion imaging judgment criterion for the third section of the present application;

[0054] Figure 10 It is the comparison of the survey line data for the first section of the present application;

[0055] Figure 11 It is the comparison of the survey line data for the second section of the present application;

[0056] Figure 12 It is the comparison of the survey line data for the third section of the present application. Detailed implementation manners

[0057] The methods and systems provided in the embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0058] As Figure 1 shown, the present application includes collecting surface micro-motion signals through a passive source surface wave linear array exploration method; the surface micro-motion signals include surface vertical vibration velocity and displacement; preprocessing the collected surface micro-motion signals; performing dispersion imaging on the preprocessed surface micro-motion signals to obtain a dispersion curve; using the dispersion curve at the initial time as a reference, calculating the phase velocity reduction rate of the dispersion curve in a future observation time period relative to the reference; and dividing the collapse risk level of the urban road according to the phase velocity reduction rate.

[0059] In this embodiment, a certain project in a certain city uses the method of this application for urban road monitoring, and the basic data is as follows:

[0060] Section Number Starting Longitude Starting Latitude End Longitude End Longitude ZZL-FV-1 116.4697284 39.91429118 116.4697525 39.91584234 ZZL-FV-2 116.463522 39.913408 116.463508 39.912824 ZZL-FV-3 116.4694655 39.91446193 116.4694628 39.91417803

[0061] Specifically, S1, collect surface micro - motion signals through the passive - source surface - wave linear - array exploration method. S11, according to the basic data, the three sections of the road to be evaluated are all located on Zhenzhi Road, and the road section lengths are similar. Therefore, a survey line is arranged on each section of the road, and the direction of the survey line is north - south.

[0062] S12, according to the road width and traffic flow of Zhenzhi Road, design the observation system as Figure 2 shown. Using the passive - source surface - wave linear - array exploration method, the geophone layout parameters are as follows: number of geophones: 18 geophones are arranged on each survey line; geophone spacing: the spacing between adjacent geophones is 2m to ensure the detection accuracy; total length of the survey line: 34m (18 geophones, spacing 2m).

[0063] S13, observation time: August 16, 2024 (the first time), October 11, 2024 (the second time), and November 12, 2024 (the third time); observation location: each time of observation selects the same road section and survey line layout as the first time (in August); observation method: use the ambient noise generated by vehicle loads as the passive source, and there is no need to arrange a seismic source; observation duration: the single - time observation duration is not less than 1 hour to obtain sufficient data; surface micro - motion signals: obtain the surface micro - motion signals on three survey lines ZZL - FV - 1, ZZL - FV - 2, and ZZL - FV - 3; through the above steps, the surface micro - motion signals of three sections of urban roads are obtained at three time points (August, October, and November 2024). By regularly (such as every 1 - 2 months) repeating the observation of the same road section and survey line, the dynamic monitoring of the urban road collapse risk can be realized. By comparing and analyzing the surface micro - motion signals in different periods, the change trend of the road structure can be found in time, providing a basis for road maintenance decision - making.

[0064] S2, pre - process the collected surface micro - motion signals, including:

[0065] S21, the original surface micro - motion signal is an m×n matrix D, where m is the number of channels (i.e., the number of geophones), and n is the number of sampling points of each channel of data. Each row of the matrix D represents a channel of surface micro - motion signal. Calculate the mean value of each row of the matrix D (i.e., each channel of surface micro - motion signal) respectively, and obtain an m×1 mean - value vector mean_D:

[0066] i = 1, 2,....., m, where mean_D[i] represents the mean value of the i-th ground microtremor signal. represents the sum of all elements in the i-th row.

[0067] Using the mean vector mean_D, each element of matrix D is de-meaned to obtain the de-meaned ground microtremor signal matrix D demeaned : D demeaned [i, j] = D[i, j] - mean_D[i], i = 1, 2,...., m; j = 1, 2,....., n, where D demeaned [i, j] represents the value of the j-th sampling point of the i-th de-meaned ground microtremor signal.

[0068] For each element of matrix D demeaned , it is adjusted according to the positive or negative sign of the corresponding element in the original ground microtremor signal matrix D: D demeaned [i, j]' = sign(D[i, j]) × abs(D demeaned [i, j]), i = 1, 2,...., m; j = 1, 2,...., n, where sign(*) is the sign function and abs(*) is the absolute value function. Through this step, it is ensured that the de-meaned ground microtremor signal has the same vibration direction (context motion direction) as the original data. The de-meaned ground microtremor signal matrix D_demeaned has the same size (m × n) as the original ground microtremor signal matrix D. The purpose of de-meaning is to eliminate the DC bias in the ground microtremor signal and make the data vibrate centered around zero mean. At the same time, by retaining the positive and negative signs of the data after de-meaning, the vibration direction information of the ground microtremor signal is ensured not to be lost, which is very important for analyzing the propagation characteristics of surface waves.

[0069] S22. The de-meaned ground microtremor signal matrix is D demeaned , whose size is m × n, where m is the number of channels and n is the number of sampling points per channel. For each row (i.e., each ground microtremor signal) of matrix D demeaned , calculate its maximum absolute value respectively to obtain a m × 1 maximum absolute value vector max_abs_D:

[0070] max_abs_D[i] = max(abs(D demeaned [i, :])), i = 1, 2,......, m, where max_abs_D[i] represents the maximum absolute value of the i-th ground microtremor signal, and abs(D demeaned [i, :]) represents the absolute value of all elements in the i-th row.

[0071] Using the maximum absolute value vector max_abs_D, for matrix D demeaned each element is normalized to obtain the normalized surface micro - motion signal matrix D normalized :

[0072] i = 1, 2,....., m; j = 1, 2,...., n, where D normalized [i, j] represents the absolute value of the j - th sampling point of the i - th surface micro - motion signal after normalization. By dividing by the maximum absolute value of each channel of data, the amplitude range of each channel of data is unified to the interval [0, 1].

[0073] For each element of matrix D normalized adjust according to the positive and negative signs of the corresponding position elements in the original surface micro - motion signal matrix D: D normalized [i, j]' = sign(D[i, j])×D normalized [i, j], i = 1, 2,....., m; j = 1, 2,...., n, where sign(*) is the sign function. Through this step, it is ensured that the surface micro - motion signal after normalization has the same vibration direction (context motion direction) as the original data. The surface micro - motion signal matrix D normalized after normalization, this matrix has the same size (m×n) as the original surface micro - motion signal matrix D.

[0074] The purpose of normalization is to eliminate the amplitude differences between the surface micro - motion signals of different channels, and unify the data amplitudes of all channels to the interval [0, 1]. In the urban road environment, due to the differences in traffic loads, pavement conditions, etc. of different lanes, there may be large differences in the amplitudes of the surface micro - motion signals obtained by different geophones. Through normalization, this difference can be eliminated, so that the data of all channels have a consistent scale in amplitude. At the same time, similar to the mean - removal process, normalization also needs to retain the positive and negative signs of the surface micro - motion signals to ensure that the vibration direction information is not lost. This is very important for analyzing the propagation characteristics of surface waves and the integrity of road structures.

[0075] S23, according to the formation conditions and surface wave characteristics of the target road, divide the frequency band of the surface micro - motion signal after normalization, and determine the main frequency range [f min , f max of the surface wave signal. Usually, the main frequency range of the surface wave signal is related to the velocity structure and wavelength of the formation, and can be estimated based on experience or prior knowledge.

[0076] For the surface micro - motion signal matrix D after frequency - band division normalizedPerform a two-dimensional Fourier transform (2D-FFT) to transform the surface micro-motion signal from the time-space domain to the frequency-wavenumber domain, obtaining the frequency-wavenumber domain matrix D fk :

[0077] D fk = fft2(D normalized ), where fft2(*) represents the two-dimensional Fourier transform function.

[0078] In the frequency-wavenumber domain, set an elliptical filtering window according to the energy distribution characteristics of the surface wave signal. The major axis direction of the elliptical window is the main direction of the surface wave signal energy distribution, and the minor axis direction is the secondary direction of the surface wave signal energy distribution. The size and direction of the elliptical window can be adjusted according to the characteristics of the surface wave signal energy distribution in the frequency-wavenumber domain image.

[0079] Use the set elliptical filtering window to filter the frequency-wavenumber domain matrix D fk and only retain the energy within the main frequency range [f min , f max of the surface wave signal within the elliptical window to suppress the noise components. After filtering, the filtered frequency-wavenumber domain matrix D fk,filtered is obtained.

[0080] Perform an inverse two-dimensional Fourier transform (2D-IFFT) on the filtered frequency-wavenumber domain matrix D fk,filtered to convert the data back from the frequency-wavenumber domain to the time-space domain, obtaining the filtered surface micro-motion signal matrix D filtered :

[0081] D filtered = ifft2(D fk,filtered ), where ifft2(*) represents the inverse two-dimensional Fourier transform function.

[0082] Perform spectral whitening on the filtered surface micro-motion signal matrix D filtered and only normalize the frequency components within the main frequency range

[0083] [f min , f max . The specific steps are as follows: Perform a Fourier transform on each trace of the filtered surface micro-motion signal to obtain spectral data; calculate the amplitude spectrum of each frequency component within the main frequency range; normalize the amplitude spectrum of each frequency component so that its amplitude is 1; perform an inverse Fourier transform on the normalized amplitude spectrum to obtain the time-domain data after spectral whitening. After spectral whitening, the preprocessed surface micro-motion signal matrix D preprocessed is obtained, and its size is the same as that of the original surface micro-motion signal matrix D.

[0084] The purpose of spectral whitening processing is to suppress the noise interference in the surface micro - motion signals and improve the signal - to - noise ratio of the surface wave signals. By filtering in the frequency - wavenumber domain using the characteristics of the energy distribution of the surface wave signals, the noise components can be effectively removed, and the main energy of the surface wave signals can be retained. At the same time, spectral whitening is performed on the filtered data to balance the amplitudes of different frequency components, so that the surface wave signals have similar energy levels within the main frequency range.

[0085] In this embodiment, the surface micro - motion signals collected at three time points (the first time, the second time, and the third time) for three road sections (the first road section, the second road section, and the third road section) are pre - processed, and the waveform characteristics before and after pre - processing are compared and analyzed. The original surface micro - motion signals generally have varying degrees of noise interference and signal quality problems, which directly affect the identification and analysis of surface wave signals. After pre - processing steps such as mean removal, normalization, and spectral whitening, the quality of the surface micro - motion signals is significantly improved, the surface wave signals become clearer and more regular, and the noise is effectively suppressed. There are certain differences in the quality of surface micro - motion signals for different road sections and different time points, but after pre - processing, the data quality has been significantly improved, indicating that the pre - processing method has good applicability and robustness.

[0086] S3. Perform dispersion imaging on the pre - processed surface micro - motion signals to obtain dispersion curves, including: S31. Perform a slowness - frequency domain transformation on the pre - processed surface micro - motion signals to convert the surface micro - motion signals in the time - space domain into the slowness - frequency domain. The slowness - frequency domain transformation can be achieved by methods such as phase - shift method, τ - p transformation, etc. After the transformation, a slowness - frequency spectrum is obtained, which reflects the propagation speed (slowness) characteristics of the surface wave signals at different frequencies.

[0087] S32. In the slowness - frequency spectrum, set an energy threshold E th (f). According to the frequency range and the energy threshold, extract the local peak points of the slowness - frequency spectrum as the reliable points of the surface wave dispersion curve. The energy threshold E th (f) is divided into a low - frequency band and a high - frequency band according to different frequencies f, and piece - wise functions are used for setting respectively: f≤f0, When f>f0, an exponential function form is used. Among them, α1, n1, b1, α2, b2, c2, and f0 are parameters pre - fitted according to the surface micro - motion signals, which control the shape and position of the energy threshold function. w(f) is the data quality factor, which reflects the quality and reliability of the data at frequency f and is calculated according to the signal - to - noise ratio SNR(f): d and e are empirical parameters that control the influence degree and sensitivity of the signal - to - noise ratio on the data quality factor. By reasonably setting the energy threshold and the data quality factor, the peak points of the surface wave signals in the slowness - frequency spectrum can be effectively identified, and the reliability of the dispersion curve extraction can be improved.

[0088] S33. Based on the least squares method, the reliable points extracted are fitted to obtain the dispersion curves of the fundamental mode and the higher modes respectively. First, fit the dispersion curve of the fundamental mode. From the reliable points extracted in step S32, select the data points corresponding to the fundamental mode, which are usually a set of points with the strongest energy and best continuity in the slowness - frequency spectrum. Use the least squares method to fit the selected data points to obtain the dispersion curve of the fundamental mode. The dispersion curve of the fundamental mode is usually fitted with a polynomial function or a piecewise function. For example: Polynomial function: c(f) = a0 + a1f + a2f 2 +......+ a n f n ; Piecewise function: where c(f) is the phase velocity corresponding to the frequency f, a0, a1,....., a n are the coefficients of the polynomial function, c1(f), c2(f),....., c n (f) are the sub - functions of the piecewise function, and f1, f2,...., f n-1 are the break points of the piecewise function. Through the least squares fitting, determine the coefficients of the polynomial function or the piecewise function to minimize the difference between the fitting curve and the selected data points. The obtained dispersion curve of the fundamental mode reflects the basic propagation velocity characteristics of the surface wave signal at different frequencies and corresponds to the dominant velocity structure of the formation.

[0089] Then, fit the dispersion curve of the higher modes. From the reliable points extracted in step S32, select the data points corresponding to the higher modes, which are usually a set of points with relatively stronger energy and relatively better continuity in the slowness - frequency spectrum. Use a method similar to that for fitting the dispersion curve of the fundamental mode to perform least - squares fitting on the selected data points to obtain the dispersion curve of the higher modes. The dispersion curve of the higher modes is usually fitted with a polynomial function or a piecewise function, and the specific form is the same as that for fitting the dispersion curve of the fundamental mode. Through the least squares fitting, determine the function coefficients of the dispersion curve of the higher modes to minimize the difference between the fitting curve and the selected data points. The obtained dispersion curve of the higher modes reflects the higher - order propagation velocity characteristics of the surface wave signal at different frequencies and contains detailed information about the formation velocity structure.

[0090] In some cases, there may be multiple dispersion curves of the higher modes corresponding to different higher modes. For each higher mode, repeat the fitting process to obtain the corresponding dispersion curve of the higher mode. Sort and label all the dispersion curves of the higher modes according to the frequency range and phase velocity magnitude for subsequent analysis. The dispersion curve of the fundamental mode provides information about the dominant velocity structure of the formation, while the dispersion curves of the higher modes provide detailed information about the formation velocity structure.

[0091] S34. Optimize and extend the fundamental mode dispersion curve, and splice it with the higher mode dispersion curves to obtain the final dispersion curve. First, smooth and optimize the fundamental mode dispersion curve. Smooth the fundamental mode dispersion curve obtained in step S33 to remove the influence of outliers and noise. Common smoothing methods include the moving average method, the Savitzky-Golay filtering method, the spline interpolation method, etc. Moving average method: For each point on the dispersion curve, take a certain number of surrounding points and perform an arithmetic average to obtain the smoothed point. Savitzky-Golay filtering method: For each point on the dispersion curve, use a polynomial to fit a certain number of surrounding points to obtain the smoothed point. Spline interpolation method: Use a piecewise polynomial function to interpolate the dispersion curve to obtain the smoothed curve. Through the smoothing process, outliers and noise in the dispersion curve can be effectively removed, and the smoothness and continuity of the curve can be improved. The optimized fundamental mode dispersion curve is obtained, which reflects the smooth dispersion characteristics of the surface wave signal in the fundamental mode.

[0092] High-frequency band extension of the fundamental mode dispersion curve: The high-frequency band information of the fundamental mode dispersion curve is usually incomplete and needs to be extended according to the higher mode dispersion curves. Select a section of the curve from the higher mode dispersion curve that is connected to the high-frequency band of the fundamental mode dispersion curve as the extension section. Match and align the frequency range and phase velocity values of the extension section with the high-frequency band of the fundamental mode dispersion curve. Use a suitable fitting method (such as polynomial fitting, spline interpolation, etc.) to smoothly connect the extension section with the high-frequency band of the fundamental mode dispersion curve. The extended fundamental mode dispersion curve is obtained, with a wider frequency range and including the dispersion information in the high-frequency band.

[0093] Splice the fundamental mode dispersion curve and the higher mode dispersion curves: Splice the extended fundamental mode dispersion curve with the higher mode dispersion curves to obtain the final dispersion curve. The splicing process needs to consider the frequency range, phase velocity magnitude, and continuity of different mode dispersion curves. Usually, according to the frequency range and phase velocity magnitude, the higher mode dispersion curves are spliced after the extended fundamental mode dispersion curve in sequence. At the splicing points, use a suitable smoothing method (such as weighted average, spline interpolation, etc.) to ensure the continuity and smoothness of the dispersion curve. The final dispersion curve is obtained, which contains the complete surface wave dispersion characteristic information, covering the dispersion characteristics of the fundamental mode and the higher modes. The optimization and extension process improve the smoothness, continuity, and frequency range of the fundamental mode dispersion curve, ensuring the integrity of the dispersion information. The finally spliced dispersion curve contains the complete dispersion characteristics of the surface wave signal in different modes.

[0094] In this embodiment, the first data: Perform dispersion imaging on the surface micro-motion signal collected for the first time to obtain a dispersion diagram, as Figure 4 shown. For Figure 4Perform dispersion picking on the dispersion diagram, extract the energy peak points of the surface wave signals, and obtain the dispersion diagram after picking, as shown in Figure 5 . From the dispersion diagram after picking ( Figure 5 ), extract the dispersion curves, including the fundamental mode and higher modes, as shown in Figure 6 . Similarly, process the second and third data in the first section, and perform similar processing on the second and third sections. By performing dispersion imaging and extracting dispersion curves on the surface microtremor signals collected at different times in the three sections, the dispersion characteristics and dispersion curves at different time points for each section can be obtained. The dispersion diagram shows the energy distribution of the surface wave signals in the frequency-velocity domain and reflects the propagation velocity characteristics at different frequencies. Dispersion picking obtains the main dispersion characteristics of the surface wave signals by extracting the energy peak points in the dispersion diagram. The dispersion curves reflect the phase velocity changes of the surface wave signals at different frequencies, including the fundamental mode and higher modes.

[0095] Dispersion imaging and dispersion curve extraction are important steps in road structure assessment, providing crucial data support for subsequent surface wave inversion and road condition analysis. By comparing the dispersion characteristics and dispersion curves of different sections and at different times, the spatial variation and temporal evolution of the road structure can be evaluated, and potential road damage and collapse risks can be identified. The quality of the dispersion diagram and dispersion curves directly affects the accuracy and reliability of the assessment results. Therefore, it is necessary to strictly follow the standardized process for dispersion imaging and dispersion picking and optimize and verify the dispersion curves.

[0096] Take the three sections of data collected for the first time as the reference data for each section. During the entire three-month experiment, data collection and processing were carried out three times, and each collection interval was a fixed time, and the original three sections were re-collected and processed. In this embodiment, the phase velocity reduction rate is used as the risk assessment index, and the calculation formula is: Phase velocity reduction rate = (Phase velocity at the 1st time - Phase velocity at the i-th time) / Phase velocity at the 1st time. The phase velocity reduction rate reflects the degree of change in the phase velocity of the section relative to the reference data at different time points. According to the phase velocity reduction rate, the risk levels are classified as: Low-risk section: The phase velocity reduction rate is less than 10%; Medium-risk section: The phase velocity reduction rate is between 10% and 30%; High-risk section: The phase velocity reduction rate is between 30% and 100%. To more intuitively display the experimental results, we use the upper and lower 10% and 30% of the first measurement data as the measurement thresholds to generate the risk assessment result diagram. Figures 7 to 9 Respectively show the risk assessment results of the three sections.

[0097] Risk assessment of the first section ( Figure 10 ): Take the measurement data of the first survey line on August 16, 2024 as the original data, and the upper and lower 10% as the low-risk measurement index. Figure 10Dispersion diagrams of the first section at three time points are shown: August 16, 2024: black curve (reference data); October 11, 2024: green curve; November 12, 2024: blue curve. The dispersion results of the second and third measurements are almost within the range of plus or minus 10% based on the first measurement result. Accordingly, it is determined that the first section of the survey line is a low-risk section.

[0098] Risk assessment of the second section ( Figure 11 ): Using the measurement data of the second section of the survey line on August 16, 2024 as the original data, plus or minus 10% is used as the low-risk measurement index. Figure 11 Dispersion diagrams of the second section at three time points are shown: August 16, 2024: black curve (reference data); October 11, 2024: green curve; November 12, 2024: blue curve. The dispersion results of the second and third measurements are mainly within the range of plus or minus 10% of the first measurement result. Accordingly, it is determined that the second section of the survey line is a low-risk section.

[0099] Risk assessment of the third section ( Figure 12 ): Using the measurement data of the third section of the survey line on August 16, 2024 as the original data, plus or minus 10% is used as the low-risk measurement index. Figure 12 Dispersion diagrams of the third section at three time points are shown: August 16, 2024: black curve (reference data); October 11, 2024: green curve; November 12, 2024: blue curve; The dispersion results of the second and third measurements are both within the range of plus or minus 10% based on the first measurement. Accordingly, it is determined that the third section of the survey line is a low-risk section.

[0100] By conducting risk assessments on the multiple measurement data of the three sections within three months, we obtained the road collapse risk levels of each section. In this embodiment, all three sections are evaluated as low-risk sections, indicating that their road structures are relatively stable and the collapse risk is low. Visualizing the risk assessment results makes the results more intuitive and understandable, facilitating the road management department to quickly understand the road health status.

Claims

1. A method for evaluating the risk of urban road collapse, characterized in that Including: S1, collecting surface micro - motion signals through the passive - source surface - wave linear - array exploration method; the surface micro - motion signals include surface vertical vibration velocity and displacement; S2, pre - processing the collected surface micro - motion signals; S3, extracting surface - wave signals from the pre - processed micro - motion signals and performing dispersion imaging to obtain dispersion curves; S4, taking the dispersion curve at the initial time as a reference, calculating the phase - velocity reduction rate of the dispersion curve in a future observation period relative to the reference; S5, dividing the degree of change in the compactness state of the urban - road underground soil according to the phase - velocity reduction rate, and identifying the underground collapse - risk level; S3, performing dispersion imaging on the pre - processed surface micro - motion signals to obtain dispersion curves, including: S31, performing slowness - frequency domain transformation on the pre - processed surface micro - motion signals to obtain a slowness - frequency spectrum; S32. Set an energy threshold in the slowness-frequency spectrum. Extract the local peak points of the slowness-frequency spectrum as reliable points of the surface wave dispersion curve according to the frequency range and the energy threshold. S33, based on the least - squares method, fitting the extracted reliable points to obtain dispersion curves of the fundamental mode and higher - order modes respectively; S34, performing smoothing processing on the fundamental - mode dispersion curve to obtain an optimized fundamental - mode dispersion curve; Supplementary high - frequency - band information of the optimized fundamental - mode dispersion curve according to the higher - order - mode dispersion curve to obtain an extended fundamental - mode dispersion curve; Stitching the extended fundamental - mode dispersion curve and the higher - order - mode dispersion curve to obtain the final dispersion curve; Set the energy threshold through the following formula :[[]]END]] ； ； Calculated according to the following formula: Among them, f represents frequency; represents the segmented frequency, which divides the frequency domain into a low-frequency band and a high-frequency band; represents the scale parameter of the low-frequency band power function; represents the exponential parameter of the low-frequency band power function; represents the vertical offset parameter of the low-frequency band power function; represents the scale parameter of the high-frequency band exponential function; represents the exponential parameter of the high-frequency band exponential function; represents the vertical offset parameter of the high-frequency band exponential function; represents the data quality factor; d represents the parameter controlling the influence degree of the signal-to-noise ratio on the data quality factor; e represents the parameter controlling the sensitivity of the signal-to-noise ratio to the data quality factor; represents the signal-to-noise ratio at frequency f.

2. The urban - road collapse - risk assessment method according to claim 1, wherein: S1, collecting surface micro - motion signals through the passive - source surface - wave linear - array exploration method, including: S11, selecting a survey line on the section to be evaluated according to the grade and traffic flow of the urban road; S12, designing geophone layout parameters according to the road width and traffic flow, and arranging geophones linearly along the road on the survey line, and the value range of the distance between adjacent geophones is from 1m to 3m; S13, using the passive - source surface - wave exploration method for observation, setting the single - observation duration to obtain surface micro - motion signals.

3. The urban - road collapse - risk assessment method according to claim 1, wherein: S2, pre - processing the collected surface micro - motion signals, including: S21, performing mean - removal processing on the surface micro - motion signals; S22, performing normalization processing on the surface micro - motion signals after mean - removal processing to eliminate the amplitude difference between different lanes and retain the up - and - down movement - direction information of the surface vibration; S23, performing spectral whitening processing on the surface micro - motion signals after normalization processing to suppress noise interference and obtain pre - processed surface micro - motion signals.

4. The urban - road collapse - risk assessment method according to claim 3, wherein: S21, performing mean - removal processing on the surface micro - motion signals, including: Calculating the mean of each channel of surface micro - motion signals; Subtracting the mean of each channel of surface micro - motion signals from the corresponding mean to obtain surface micro - motion signals after mean - removal processing; Adjusting the positive and negative signs of the surface micro - motion signals after mean - removal processing to be the same as those of the original surface micro - motion signals to retain the context - movement - direction information of the original surface micro - motion signals.

5. The urban - road collapse - risk assessment method according to claim 4, wherein: S22. Normalize the surface microtremor signal after mean removal, including: Calculate the maximum absolute value of each channel of the surface microtremor signal after mean removal; Using the maximum absolute value, normalize the absolute value of each channel of the surface microtremor signal after mean removal to the range of 0 to 1; Adjust the positive and negative signs of the normalized surface microtremor signal to be consistent with the original surface microtremor signal to unify the amplitudes of the surface microtremor signals of different channels.

6. The urban road collapse risk assessment method according to claim 3, characterized in that: S23. Perform spectral whitening on the normalized surface microtremor signal, including: According to the formation conditions and surface wave characteristics of the target road, divide the frequency band of the normalized surface microtremor signal to determine the main frequency range of the surface wave signal; Perform two-dimensional Fourier transform on the surface microtremor signal after frequency band division to transform the surface microtremor signal from the time-space domain to the frequency-wavenumber domain; In the frequency-wavenumber domain, set an elliptical filter window according to the energy distribution characteristics of the surface wave signal, where the major axis of the elliptical filter window is the first direction of the energy distribution of the surface wave signal, the minor axis is the second direction of the energy distribution of the surface wave signal, the first direction is the main direction, and the second direction is the secondary direction; Use the set elliptical filter window to filter the frequency-wavenumber domain data, and only retain the energy within the main frequency range of the surface wave signal within the elliptical window to suppress the noise component; Perform two-dimensional inverse Fourier transform on the filtered frequency-wavenumber domain data to convert the data back from the frequency-wavenumber domain to the time-space domain to obtain the filtered surface microtremor signal; Perform spectral whitening on the filtered surface microtremor signal, and only normalize the frequency components within the main frequency range of the surface wave signal to obtain the preprocessed surface wave observation data.

7. An urban road subsidence risk assessment system, characterized in that, Including: At least one processing unit; configured to execute instructions to implement the urban road collapse risk assessment method according to any one of claims 1 to 6.

Images