Ballast railway ballast bed mechanical state monitoring method based on high-frequency environmental noise

Abstract

The invention discloses a ballasted railway ballast bed mechanical state monitoring method based on high-frequency environment noise. The method comprises the steps that S1, three-axis acceleration signals of a ballasted railway under the high-frequency environment noise are collected; s2, integrating the three-axis acceleration signal to obtain a three-axis speed signal, calculating horizontal / vertical power spectral density based on the three-axis speed signal, and calculating the horizontal / vertical power spectral density to obtain an average spectral ratio curve; and S3, constructing an initial basic structure model based on the prior information of the ballast bed and the roadbed, performing inversion on the average spectrum ratio curve by adopting a global inversion algorithm based on the initial basic structure model to obtain the ballast bed shear wave velocity, and judging the mechanical state of the ballast bed according to the ballast bed shear wave velocity. According to the method, the average spectrum ratio curve is established by collecting the signals under the high-frequency environmental noise, the ballast bed shear wave velocity is obtained through inversion, real-time monitoring and long-term tracking of the mechanical state of the ballast bed of the ballast railway are achieved, and the method has remarkable engineering application potential and popularization value.

Description

Technical Field

[0001] This invention relates to the field of ballast railway operation and maintenance technology, specifically to a method for monitoring the mechanical condition of ballast railway track bed based on high-frequency environmental noise. Background Technology

[0002] Ballasted railways are the most widely used track structure in my country's railway system, and ballast bed is an important component of the basic structure of ballasted railways. Reasonable rail support stiffness is a necessary condition to ensure the smoothness, comfort, and safety of train operation. However, the stiffness of ballast bed in conditions such as bridge-to-subgrade transition sections, tunnel entrance and exit sections, poor subgrade drainage sections, windy and sandy areas, and coal transport lines is affected by factors such as climate change, rainwater erosion, and repeated train loads, exhibiting significant time-varying characteristics.

[0003] Current standards for assessing the mechanical state of ballast tracks require the removal of sleeper fasteners for manual static loading tests. This invasive operation consumes significant time and labor costs, severely limiting testing efficiency and operational applicability. Current non-destructive testing methods used in railway infrastructure, such as mobile track loading vehicles, track inspection trolleys, and ground-penetrating radar, primarily focus on physical or track geometry indicators, failing to effectively evaluate the mechanical properties of the ballast track. Recent research has introduced elastic surface wave testing methods to evaluate the mechanical performance of the ballast track through the elastic wave velocity of the ballast medium layer. However, my country's high-speed railways use concrete sleepers, which have a significant difference in wave impedance compared to the ballast track. The interface elastic wave energy is predominantly reflected wave (78%), resulting in a complex wavefield beneath the track. Traditional surface wave testing and analysis methods, based on a one-dimensional horizontal layering assumption, cannot effectively obtain dispersion curves containing ballast layer wave velocity characteristics, leading to poor testing results and the inability to conduct continuous testing. Summary of the Invention

[0004] In view of one or more shortcomings of the existing technology, the present invention provides a method for monitoring the mechanical state of ballast railway track bed based on high-frequency environmental noise. This method can obtain the dynamic parameters of track bed and subgrade by utilizing the vibration and noise of the surrounding environment, without the need for active excitation, and realizes real-time monitoring and long-term continuous tracking of the mechanical state of track bed.

[0005] To achieve the above objectives, the present invention adopts one or more of the following technical solutions: A method for monitoring the mechanical state of ballasted railway track bed based on high-frequency environmental noise includes the following steps: S1. Collect triaxial acceleration signals under high-frequency environmental noise of ballast railway; S2. Integrate the triaxial acceleration signal to obtain a triaxial velocity signal, calculate the horizontal / vertical power spectral density based on the triaxial velocity signal, and calculate the average spectral ratio curve based on the horizontal / vertical power spectral density; S3. Construct an initial foundation structure model based on prior information of the track bed and subgrade. Based on the initial foundation structure model, use a global inversion algorithm to invert the average spectral ratio curve to obtain the track bed shear wave velocity. Determine the mechanical state of the track bed based on the track bed shear wave velocity.

[0006] Preferably, in step S1, the frequency range of the high-frequency environmental noise that reflects the track bed condition is 50Hz~1kHz.

[0007] Preferably, in step S1, a triaxial accelerometer is used to acquire its triaxial acceleration signal under high-frequency environmental noise. The triaxial accelerometer is a wideband response sensor with an effective frequency response range covering the 50Hz~1kHz frequency range.

[0008] Preferably, the triaxial accelerometer is arranged on the upper surface of the sleeper box of the ballasted railway and is coupled and fixed to the surface of the loose track bed by coupling nails.

[0009] Preferably, in step S2, the specific steps for integrating the triaxial acceleration signal to obtain the triaxial velocity signal and calculating the horizontal / vertical power spectral density based on the triaxial velocity signal are as follows: S21. Perform detrending and mean-removing processing on the triaxial acceleration signal; S22. Integrate the processed triaxial acceleration signal to obtain a triaxial velocity signal, and perform detrending processing on the triaxial velocity signal; S23. Remove the window containing discontinuities and near-field disturbance noise from the processed triaxial velocity signal to obtain a new triaxial velocity signal, and calculate the horizontal / vertical power spectral density of the new triaxial velocity signal.

[0010] Preferably, in step S2, after iteratively calculating the average spectral ratio curve from the horizontal / vertical power spectral density, the average spectral ratio curve is smoothed to eliminate curve spikes.

[0011] Preferably, in step S22, a preset sliding window is used to perform detrending processing on the triaxial velocity signal, wherein the sliding window is set to a length of 100s and a step size of 25s.

[0012] Preferably, in step S23, the specific steps for calculating the horizontal / vertical power spectral density of the new triaxial velocity signal and iteratively calculating the average spectral ratio curve from the horizontal / vertical power spectral density are as follows: The new triaxial velocity signal is processed using a sliding time window method, dividing the continuous time period into multiple overlapping time periods, and the time spectrum of the signal is calculated by using short-time Fourier transform for each time period. The spectral power density in three directions is obtained from the time spectrum. After synthesizing the horizontal vibration power spectrum, the ratio curve of the horizontal to the vertical power spectrum is calculated using the following formula:

[0013] in, , : The power spectral density of the j-th window of two orthogonal horizontal components; : The vertical component power spectral density of the j-th window; The average spectral ratio curve is calculated using the following formula:

[0014] Where n: number of windows; : The ratio of the horizontal to the vertical power spectrum of the j-th window.

[0015] Preferably, in step S2, after obtaining the new sampling window, the average spectral ratio curve is iteratively updated, and the average spectral ratio curve of the t-th update (t≥2) is calculated using the following formula:

[0016] Where n: number of windows; i: sliding step size; The sliding window is calculated and updated in real time using the following recursive iterative formula:

[0017] Preferably, in step S3, for the spectral ratio curve updated in each iteration, a local optimization algorithm is used to perform rapid inversion based on the shear wave velocity of the track bed obtained from the initial inversion, and the mechanical parameters of the track bed are updated.

[0018] Preferably, in step S3, the initial basic structure model includes multiple initial parameters such as layer thickness, transverse wave velocity, longitudinal wave velocity, Poisson's ratio, and density; the prior information of the track bed and subgrade includes the layer thickness and material of the track bed and subgrade.

[0019] Preferably, in step S3, the global inversion algorithm adopts the Monte Carlo algorithm, and the local optimization inversion algorithm adopts the downhill method.

[0020] Preferably, in step S3, the specific process of determining the mechanical state of the track bed based on the shear wave velocity is as follows: Based on the relationship between shear wave velocity and shear modulus Conversion relationship between shear modulus and elastic modulus The elastic modulus and shear modulus of the ballasted track bed are calculated, and the mechanical state of the track bed is determined based on the elastic modulus and shear modulus.

[0021] The present invention also provides a ballast railway track bed mechanical condition monitoring system based on high-frequency environmental noise, comprising: The signal acquisition module is used to acquire triaxial acceleration signals under high-frequency environmental noise conditions on ballasted railways. The time-frequency processing module is used to integrate the triaxial acceleration signal to obtain a triaxial velocity signal, calculate the horizontal / vertical power spectral density based on the triaxial velocity signal, and calculate the average spectral ratio curve based on the horizontal / vertical power spectral density. The inversion evaluation module is used to construct an initial basic structure model based on prior information of the track bed and subgrade. Based on the initial basic structure model, a global inversion algorithm is used to invert the average spectral ratio curve to obtain the track bed shear wave velocity. The mechanical state of the track bed is determined based on the track bed shear wave velocity.

[0022] The present invention also provides an electronic device, including a memory and a processor, wherein the memory is used to store a program; the processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the method for monitoring the mechanical state of ballast railway track bed based on high-frequency environmental noise as described in any of the preceding claims.

[0023] The present invention also provides a non-transitory computer-readable storage medium for storing a readable computer program that, when executed by a processor, can implement the steps in the method for monitoring the mechanical state of ballast railway track bed based on high-frequency environmental noise as described in any of the preceding claims.

[0024] By adopting the above technical solution, the beneficial effects of the present invention are as follows: 1. This invention establishes an average spectral ratio curve by collecting signals under high-frequency environmental noise and obtains the shear wave velocity of the ballast track through inversion, thereby realizing the monitoring of the mechanical state of the ballast track. Furthermore, the sampling and calculation process in the monitoring method of this invention is continuous, enabling real-time monitoring and long-term tracking of the mechanical state of the ballast track. It has strong applicability to key nodes in the line or scenarios with large fluctuations in the ballast track state, and has significant engineering application potential and promotion value.

[0025] 2. This invention utilizes high-frequency environmental vibration and noise around the track to obtain dynamic parameters of the track bed and subgrade, enabling testing of the mechanical state of ballasted railway track beds. On the one hand, it eliminates the need for static loads. Compared to the manual static loading testing methods in current standards, this invention does not require the removal of sleeper fasteners when applied to track bed mechanical state monitoring, resulting in minimal intrusion and no impact on normal line operation. On the other hand, compared to traditional active excitation detection methods for elastic surface waves, this invention eliminates the need for artificial excitation sources, reducing costs. Furthermore, it overcomes the limitation of traditional active excitation detection methods for elastic surface waves in being unable to conduct continuous testing, enabling real-time, long-term tracking monitoring.

[0026] 3. Compared with non-destructive testing technologies such as mobile track loading vehicles, track inspection trolleys, and ground-penetrating radar, the ballast railway track bed mechanical condition monitoring method of the present invention can directly evaluate the track bed mechanical condition, and is more efficient and accurate in track bed mechanical condition monitoring. Attached Figure Description

[0027] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0028] Figure 1 This is a flowchart of the method for monitoring the mechanical state of ballasted railway track bed based on high-frequency environmental noise in an embodiment of the present invention; Figure 2 This is a schematic diagram of the coupling method between the sensor and the loose-bed track in an embodiment of the present invention; Figure 3 This is a schematic diagram of the sensor arrangement in an embodiment of the present invention; Figure 4 This is a diagram showing the processing effect of step S2 in an embodiment of the present invention; Figure 5 This is a time-frequency analysis of rail transit load noise in an embodiment of the present invention. Figure 1 ; Figure 6 This is a time-frequency analysis of ground traffic load noise in an embodiment of the present invention. Figure 2 ; Figure 7 This is a schematic diagram of the STA / LTA screening effect in an embodiment of the present invention; Figure 8 This is a schematic diagram of the iterative calculation of the horizontal / vertical spectral ratio curve sliding update during the monitoring process in an embodiment of the present invention; Figure 9 These are the power spectral density and horizontal / vertical spectral ratio curves at measurement point 1 in this embodiment of the invention; Figure 10 These are the power spectral density and horizontal / vertical spectral ratio curves at measurement point two in this embodiment of the invention; Figure 11 These are the power spectral density and horizontal / vertical spectral ratio curves at the three measurement points in this embodiment of the invention; Figure 12 These are the power spectral density and horizontal / vertical spectral ratio curves at measurement point four in this embodiment of the invention; Figure 13 This is a comparison chart of the global inversion effect and actual measurement of each measurement point in the embodiment of the present invention.

[0029] 1. Triaxial accelerometer; 2. Coupler pin; 3. Measurement point one; 4. Measurement point two; 5. Measurement point three; 6. Measurement point four. Detailed Implementation

[0030] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0031] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0032] Example 1 In one typical embodiment of this application, a method for monitoring the mechanical state of ballasted railway track bed based on high-frequency environmental noise is provided, such as... Figures 1-13 As shown, it includes the following steps: S1. Collect triaxial acceleration signals under high-frequency environmental noise of ballast railway.

[0033] Specifically, triaxial accelerometers are arranged on the upper surface of the sleeper box of the ballast railway. The triaxial accelerometers are used to collect the triaxial vibration acceleration signal under high-frequency environmental noise. The high-frequency environmental noise is mainly generated by human activities such as surrounding traffic loads and construction disturbances. Its frequency range reflecting the condition of the track bed is 50Hz~1kHz.

[0034] Among them, reference Figure 2 The triaxial accelerometer 1 includes three accelerometers, which are arranged to form an orthogonal triaxial sensor. In this embodiment, the accelerometer is a wide-band response sensor, such as a piezoelectric accelerometer with a frequency response range of 0.2 Hz to 1500 Hz. The effective frequency response range can cover the high-frequency noise range of 50 Hz to 1 kHz to ensure the effective acquisition of high-frequency environmental noise signals.

[0035] In this embodiment, the triaxial accelerometer 1 is coupled to the surface of the loose ballast bed through a coupling pin 2. The coupling pin 2 is 25cm long and 2cm in diameter. Its length is greater than twice the maximum particle size of the loose ballast, which can stably fix the triaxial accelerometer in the ballast bed pores.

[0036] S2. Integrate the triaxial acceleration signal to obtain the triaxial velocity signal, calculate the horizontal / vertical power spectral density based on the triaxial velocity signal, and calculate the average spectral ratio curve based on the horizontal / vertical power spectral density.

[0037] Further, the specific steps are as follows: S21. The acquired triaxial acceleration signals are simultaneously detrended and mean-reduced to reduce the sensor's own drift signal and extract pure vibration noise.

[0038] S22. Integrate the processed triaxial acceleration signal to obtain the triaxial velocity signal, and use a preset sliding window to detrend the curve of the triaxial velocity signal. The purpose is to eliminate distortions such as secondary curvature of the integration result caused by sensor drift. During detrending, the velocity signal sliding window is set to a length of 100s and a step size of 25s. Discontinuities on the time history curve generated by fitting between windows can be removed using the short-long-short-time-window (STA / LTA) algorithm. The processing effect is as follows: Figure 4 As shown in the comparison.

[0039] S23. Remove the window containing discontinuities and near-field disturbance noise from the processed triaxial velocity signal to obtain a new triaxial velocity signal. Calculate the horizontal / vertical power spectral density of the new triaxial velocity signal.

[0040] Furthermore, the specific processing steps of step S23 include: S231. The STA / LTA algorithm is used to remove windows containing discontinuities and near-field disturbance noise from the triaxial velocity signal. Since the duration of noise generated by traffic loads is typically 25-30 seconds, this embodiment selects a 60-second LTA to avoid failing to filter long-term interference noise, and a 0.5-second STA is selected to filter transient noise. The STA / LTA ratio is set to a maximum of 2.0 and a minimum of 0.1 under strict conditions, which effectively eliminates abrupt changes in high-frequency environmental noise and discontinuities caused by sliding window detrending, extracting the effective environmental noise window. The velocity signals before and after processing are as follows: Figure 7 As shown.

[0041] S232. The new triaxial velocity signal is processed using a sliding time window method, dividing the continuous time period into multiple overlapping time periods. The time spectrum of the signal is calculated for each time period using a short-time Fourier transform (STFT). When performing time-frequency analysis on the new triaxial velocity signal, the velocity signal of each sampling period is analyzed using a short-time Fourier transform (STFT). The data is processed using a sliding window method, dividing the continuous time period (e.g., 12 hours) into multiple overlapping time periods, and the time spectrum of the signal in each time period is calculated. The short-time Fourier transform can segment non-stationary vibration signals into multiple quasi-stationary frames for spectrum analysis. Specifically, the velocity signal is first divided into multiple sliding time windows, with a certain proportion of overlapping time periods between adjacent windows. Then, a short-time Fourier transform is performed on each time period to obtain the time spectrum of the entire triaxial velocity signal. In this embodiment, the window length is set to 4.096s, the step size is 3.2768s, and adjacent windows are superimposed by 20%.

[0042] S232. From the time spectrum of the STFT results, the power density of the three channels (two horizontal components and a vertical component) in the horizontal and vertical directions is calculated respectively. After synthesizing the horizontal vibration power spectrum, the ratio curve of the horizontal and vertical power spectra is calculated using the following formula (1): (1) in, , : The power spectral density of the j-th window of two orthogonal horizontal components; : The vertical component power spectral density of the j-th window; Based on the sliding calculation method that includes a total of n windows, and assuming the sliding step size is i, the formula for calculating the geometric mean spectral ratio curve is: (2) Where n: number of windows; : The ratio of the horizontal to the vertical power spectrum of the j-th window.

[0043] During long-term monitoring, after obtaining a new sampling window, the average spectral ratio curve is iteratively updated. The average spectral ratio curve of the t-th (t≥2) update is calculated using the following formula (3): (3) Where n: number of windows; i: sliding step size.

[0044] like Figure 8 As shown, to achieve real-time sliding updates, the sliding window is calculated using the following recursive iterative formula: (4) S233. Logarithmic bandwidth is used to smooth the data after each update of the curve in order to reduce the spike fluctuations of the spectral ratio curve.

[0045] S3. Construct an initial basic structure model based on prior information of the track bed and subgrade. Based on the initial basic structure model, use a global inversion algorithm to invert the average spectral ratio curve to obtain the shear wave velocity of the track bed. Determine the mechanical state of the track bed based on the shear wave velocity.

[0046] Furthermore, the prior information for the track bed and subgrade includes information such as the layer thickness and materials of the track bed and subgrade. The initial foundation structure model includes multiple initial parameters such as layer thickness, transverse wave velocity, longitudinal wave velocity, Poisson's ratio, and density.

[0047] Furthermore, for each iteratively updated spectral ratio curve, based on the shear wave velocity of the track bed obtained from the initial inversion, a local optimization algorithm is used for rapid inversion to quickly update the track bed mechanical parameters and monitor changes in the mechanical state of the track bed. In this embodiment, the global inversion uses the Monte Carlo algorithm, and the local optimization inversion uses the downslope method.

[0048] Furthermore, based on the fundamental theoretical formulas, the following conversion relationships exist between shear wave velocity and shear modulus, and between shear modulus and elastic modulus: (5) (6) The quasi-static elastic modulus and shear modulus of the ballast track can be calculated based on formulas (5) and (6).

[0049] Example 2 This embodiment provides the implementation process of the high-frequency environmental noise-based mechanical condition monitoring method for ballasted railway track beds from Embodiment 1 when applied to the mechanical condition monitoring of railway track beds. (Refer to...) Figures 2-13 .

[0050] like Figure 2 As shown, in an embodiment of the present invention, three piezoelectric accelerometers with a frequency response range of 0.2 to 1500 Hz are used to form an orthogonal triaxial accelerometer 1. The triaxial accelerometer 1 is coupled to the surface of the loose ballast bed through a coupling pin 2. The coupling pin 2 is 25 cm long and 2 cm in diameter. Its length is greater than twice the maximum particle size of the loose ballast, and it can be stably fixed in the pores of the ballast bed.

[0051] In this embodiment of the invention, a dynamic acquisition instrument is used to upload the sensor's micro-vibration electrical signal to an industrial control computer at a sampling frequency of 2kHz for continuous recording and real-time calculation and processing.

[0052] This embodiment uses a ballasted railway test line as the test object. During the test, the line is located in the suburbs of an city, with no operating vehicles on the line. Within a 2km radius, there are facilities including subway, high-speed railway, conventional railway, and urban expressway. Preliminary site investigation showed that the ballast gradation of the line was severely deteriorated and there were signs of fine-grained soil intrusion. The thickness of the loose ballast layer is about 0.4m, the surface layer of the roadbed is filled with fine-grained soil mixed with 10% lime, and the thickness is not less than 0.4m. The subgrade layer is filled with medium-coarse sand. Based on the above prior information, to improve the robustness of the inversion process, the search range of initial parameters and the number of layers are reasonably expanded to avoid getting trapped in local optima or incorrect convergence paths, thereby improving the stability and adaptability of the inversion results. Initial parameters for the ballasted railway layers can be preset to complete the construction of the initial basic structure model. The specific parameters are shown in Table 1. Table 1

[0053] Where H is the layer thickness of each layer of the track bed and subgrade, Vp is the longitudinal wave velocity of the shear wave, Vs is the transverse wave velocity of the shear wave, Poisson is the Poisson ratio, and Rho represents the mass density of the material.

[0054] like Figure 3 As shown, in this embodiment, four measuring points are set inside the sleeper box. Measuring point 1 (3) is located at the center of the sleeper box; measuring point 2 (4) is 55cm away from measuring point 1 (3) and is the main area for tamping operations on ballast railways; measuring point 3 (5) is located at the edge of the sleeper; measuring point 4 (6) is located at the bottom of the track bed slope and is 50cm away from measuring point 3 (5) at a horizontal distance.

[0055] In the embodiments provided by this invention, the triaxial acceleration signals are first detrended and mean-reduced synchronously, then integrated to obtain the triaxial velocity signals, and the velocity curves are further detrended using a preset sliding window. In this embodiment, the vertical vibration channel processing for measurement point 46 is as follows: Figure 4 As shown, the sliding window is set to a length of 100s and a step size of 25s; discontinuities on the time history curves generated by fitting between windows can be removed by using the short-long-short-time-window (STA / LTA) algorithm to eliminate windows containing discontinuities.

[0056] To verify the high-frequency environmental noise sources generated by human activities in the embodiments, and to verify the feasibility and reliability of the invention in the field, the velocity signal was converted to the time-frequency domain using short-time Fourier transform and matched with the spectral characteristics of surrounding human activities for analysis, referring to... Figure 5 and Figure 6 The following example illustrates the effect of a typical noise segment in the time-frequency domain of the vertical vibration at measuring point 46: Rail transit loads involve periodic wheel-rail dynamic interactions, and the noise they generate typically exhibits a wide-band, equally spaced line spectrum. The first two peak frequencies are distributed in the ranges of 0–50 Hz and 200–300 Hz. The former is mainly generated by vertical bending vibration of the track structure, car body resonance, wheel-rail contact noise, and fastener system vibration; the latter is mainly dominated by wheel-rail contact irregularities (such as rail corrugation and weld seams). Meanwhile, the amplitude and frequency of the noise are lower than those at the center of the signal when the train approaches and departs. Figure 5 As shown, in this embodiment, the closest distance between the site and the surrounding operating rail lines is approximately 810m. (During the test...) Figure 5 The displayed time-frequency characteristics include the noise from rail traffic loads generated when trains pass by.

[0057] Ground vehicle loads are non-periodic, random tire-road dynamic interactions. The resulting excitation frequencies are dominated by factors such as road surface roughness, tire stiffness, and vehicle speed. Specifically, the low-frequency noise range (5-30Hz) is mainly generated by vehicle suspension system vibration and the vehicle's own modal vibrations during driving; the mid-frequency noise range (30-100Hz) is mainly generated by road surface irregularities, vehicle speed, and engine vibration; and the high-frequency noise range (>100Hz) is mainly dominated by road surface microtexture excitation and brake disc vibration. For example... Figure 6 As shown, in this embodiment, the closest distance between the site and the surrounding surface expressway is approximately 17m. (During the test...) Figure 6 The displayed time-frequency characteristics include the ground traffic load generated when various types of vehicles pass through the two-way three-lane road.

[0058] The high-frequency noise sources in the environment were verified and identified through methods such as comparing time-frequency characteristics, monitoring the directionality of vibration sources, and converting acceleration signals into sound pressure signals for monitoring. In addition, high-frequency environmental noise generated by human activities in this embodiment also includes sources such as surrounding construction loads and near-field pedestrian disturbances. Since these factors are not the dominant noise sources in the test, they will not be elaborated upon in this embodiment.

[0059] To eliminate abrupt changes and discontinuities caused by sliding window detrending in high-frequency environmental noise, this embodiment uses the STA / LTA algorithm for rapid filtering. Since traffic load noise typically lasts 25-30 seconds, this embodiment selects a 60-second LTA to avoid failing to filter prolonged interference noise; a 0.5-second STA is selected to filter transient noise. The ratio is set to a maximum of 2.0 and a minimum of 0.1, according to recommendations under stringent conditions. Figure 7 The figure shows the effective environmental noise window extracted by the STA / LTA algorithm at measurement point 46 based on the above parameter settings.

[0060] In this embodiment, during the monitoring process, power spectral density calculations are simultaneously performed on the four measuring points using horizontal and vertical three-channel methods. (Setting 2...) 12Each sampling point corresponds to a Tukey window function with a window length of 4.096s and a window superposition of 20%. A short-time Fourier transform is performed, and the ratio curve of the horizontal and vertical power spectra is calculated based on formula (7): (7) in, , : The power spectral density of the j-th window of two orthogonal horizontal components; : The vertical component power spectral density of the j-th window.

[0061] Based on a sliding calculation method involving a total of n windows, and assuming a sliding step size of i, the formula for calculating the geometric mean spectral ratio curve in the t-th update is: (8) Where n: number of windows; i: sliding step size; : The ratio of the horizontal to the vertical power spectrum of the j-th window.

[0062] like Figure 8 As shown, to achieve real-time sliding updates, the following recursive iterative formula is used for calculation: (9) Logarithmic bandwidth smoothing is applied to the curve after each update to reduce spikes and fluctuations in the spectral ratio curve.

[0063] The results, with a total recording time of 12 hours, are referenced. Figures 9-12 ,in, Figure 9 In the middle, (a), (b), (c), and (d) are the vertical channel spectral ratio curve, the horizontal north-south channel spectral ratio curve, the horizontal east-west channel spectral ratio curve, and the H / V spectral ratio curve, respectively, for measuring point one. Figure 10 In the middle, (a), (b), (c), and (d) are the vertical channel spectral ratio curve, the horizontal north-south channel spectral ratio curve, the horizontal east-west channel spectral ratio curve, and the H / V spectral ratio curve, respectively, for measuring point 2. Figure 11 In the middle, (a), (b), (c), and (d) are the vertical channel spectral ratio curve, the horizontal north-south channel spectral ratio curve, the horizontal east-west channel spectral ratio curve, and the H / V spectral ratio curve, respectively, for measuring point three. Figure 12 Figures (a), (b), (c), and (d) show the vertical channel spectral ratio curve, the horizontal north-south channel spectral ratio curve, the horizontal east-west channel spectral ratio curve, and the H / V spectral ratio curve for measuring point four, respectively. In the H / V spectral ratio method, the ratio of surface wave wavelength to formation thickness can be used to distinguish between high-frequency and low-frequency components. When the wavelength is less than or equal to the formation thickness, high-frequency components dominate, and the wavelength is more sensitive to layered structures; in this case, both surface waves and volume waves have energy components. When the wavelength is much greater than the formation thickness, low-frequency components dominate, and the surface wave fundamental mode (especially Rayleigh waves) is the dominant component. Analysis of the power spectrum of each channel in this embodiment indicates that the roadbed layer mainly exhibits the surface wave fundamental mode.

[0064] This invention employs a global inversion algorithm to invert the horizontal / vertical spectral ratio curves obtained from initial monitoring to obtain the shear wave velocity of the track bed; the spectral ratio curves updated in each subsequent iteration are based on the initial inversion results, and a local optimization algorithm is used for rapid inversion to update and obtain the changes in the mechanical parameters of the track bed.

[0065] Since the site has no train operation for a long period of time in this embodiment, the track bed condition can be regarded as unchanged during the monitoring time span. Therefore, this embodiment takes the geometric mean spectral ratio curve calculated from all windows as an example, and uses the Monte Carlo algorithm to perform iterative inversion based on the fundamental mode of surface waves to obtain the basic structural model m containing the track bed layer. The inversion objective function is as follows: (10) In the formula, E(m): the objective function value of model m; j: the index of the frequency sampling point; : The observed HVSR spectral ratio at the j-th frequency point, derived from measured data; : The HVSR spectral ratio at the j-th frequency point, obtained by model m based on the forward modeling of the diffusion field; : The standard deviation of the j-th frequency point.

[0066] The calculation of the aforementioned inversion objective function involves diffusion field theory, the core idea of ​​which is that in a diffusion wave field, the autocorrelation and cross-correlation of displacement signals can be represented by the imaginary part of the Green's function. When the source and receiver coincide, the displacement autocorrelation reflects the energy density in each direction, and the H / V spectral ratio is the ratio of the sum of the energy densities in the two horizontal directions to the square root of the energy density in the vertical direction. Theoretically, H / V can be directly written as the ratio of the imaginary parts of the Green's function:

[0067] The Green function is determined by the geometry and physical properties of the medium. Here, we assume the site is a horizontally layered, homogeneous, isotropic, elastic half-space. In forward modeling, the Green function is integrated in the radial wavenumber domain, decomposing it into contributions from surface waves (Rayleigh waves, Love waves) and volume waves (P waves, S waves). Surface wave component: Rayleigh and Love wave modes are extracted analytically (using dispersion curves and pole locations).

[0068] Body wave component: obtained by numerical integration over a finite wavenumber interval.

[0069] The ellipticity of the Rayleigh wave and the response amplitude of each mode are determined by the dispersion curve. Finally, the imaginary parts of the Green's function in the horizontal and vertical directions are combined to obtain the theoretical H / V curve of the corresponding layered model.

[0070] Since this part pertains to basic knowledge in the field of geophysical exploration and is not part of the core technology of this application, it will not be elaborated upon.

[0071] This embodiment uses variance to determine the inversion effect, setting the initial number of random models to 50,000; allowing low-velocity regions in the formation; setting the regularization factor to 0.001; setting the initial perturbation range for Monte Carlo inversion to 40%, and gradually reducing the range as the inversion converges until the final inversion result tends to stabilize. The HV curve within the first 5% of the inversion error range is compared with the measured values, such as... Figure 13 As shown in (a), (b), (c), and (d), the HV curves obtained by inversion from the four measurement points (point 1, point 2, point 3, and point 4) are basically consistent with the measured curves in shape. The optimal solutions for all measurement points and the average error of the top 5% are summarized in Table 2. Table 2

[0072] Where H is the layer thickness, E represents the elastic modulus, and G represents the shear modulus.

[0073] Referring to existing research in this field: Clean ballast: elastic modulus E is 210~275 MPa; Dirty ballast: elastic modulus E is in the range of 345~380MPa; Dirty ballast under humid conditions: elastic modulus E is in the range of 135~170MPa.

[0074] Based on the above inversion results and existing standards in this field, it can be determined that the mechanical properties of the ballast at measuring points 2 (4), 3 (5), and 4 (6) are significantly higher than normal values. The tested ballast may have local characteristics such as gradation deterioration and contamination. The verification is consistent with the actual state of the ballast on site, which is severely degraded and has fine soil intrusion.

[0075] When used for long-term monitoring of the mechanical state of the track bed in the field, the spectral ratio curve updated in each subsequent iteration is based on the initial inversion results and a local optimization algorithm is used for rapid inversion to update and obtain the changes in the mechanical parameters of the track bed.

[0076] In another typical embodiment of the present invention, a ballast railway track bed mechanical condition monitoring system based on high-frequency environmental noise is provided, the system comprising: The signal acquisition module is used to acquire triaxial acceleration signals under high-frequency environmental noise conditions on ballasted railways. The time-frequency processing module is used to integrate the triaxial acceleration signal to obtain a triaxial velocity signal, calculate the horizontal / vertical power spectral density based on the triaxial velocity signal, and calculate the average spectral ratio curve based on the horizontal / vertical power spectral density. The inversion evaluation module is used to construct an initial basic structure model based on prior information of the track bed and subgrade. Based on the initial basic structure model, a global inversion algorithm is used to invert the average spectral ratio curve to obtain the track bed shear wave velocity. The mechanical state of the track bed is determined based on the track bed shear wave velocity.

[0077] On the other hand, the present invention also provides an electronic device, including a memory and a processor, wherein the memory is used to store a program; the processor is coupled to the memory and is used to execute the program stored in the memory to complete the steps in the method for monitoring the mechanical state of ballast railway track bed based on high-frequency environmental noise provided in Embodiment 1.

[0078] In another aspect, the present invention also provides a non-transitory computer-readable storage medium for storing a readable computer program that, when executed by a processor, can perform the steps in the method for monitoring the mechanical state of ballast railway track bed based on high-frequency environmental noise provided in Embodiment 1.

[0079] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for monitoring the mechanical state of ballasted railway track bed based on high-frequency environmental noise, characterized in that, Includes the following steps: S1. Collect triaxial acceleration signals under high-frequency environmental noise of ballast railway; S2. Integrate the triaxial acceleration signal to obtain a triaxial velocity signal, calculate the horizontal / vertical power spectral density based on the triaxial velocity signal, and calculate the average spectral ratio curve based on the horizontal / vertical power spectral density; S3. Construct an initial foundation structure model based on prior information of the track bed and subgrade. Based on the initial foundation structure model, use a global inversion algorithm to invert the average spectral ratio curve to obtain the track bed shear wave velocity. Determine the mechanical state of the track bed based on the track bed shear wave velocity.

2. The method for monitoring the mechanical state of ballasted railway track bed based on high-frequency environmental noise as described in claim 1, characterized in that, In step S1, the frequency range of the high-frequency environmental noise that reflects the track bed condition is 50Hz~1kHz.

3. The method for monitoring the mechanical state of ballasted railway track bed based on high-frequency environmental noise as described in claim 1, characterized in that, In step S2, the triaxial acceleration signal is integrated to obtain the triaxial velocity signal, and the specific steps for calculating the horizontal / vertical power spectral density based on the triaxial velocity signal are as follows: S21. Perform detrending and mean-removing processing on the triaxial acceleration signal; S22. Integrate the processed triaxial acceleration signal to obtain a triaxial velocity signal, and perform detrending processing on the triaxial velocity signal; S23. Remove the window containing discontinuities and near-field disturbance noise from the processed triaxial velocity signal to obtain a new triaxial velocity signal, and calculate the horizontal / vertical power spectral density of the new triaxial velocity signal.

4. The method for monitoring the mechanical state of ballasted railway track bed based on high-frequency environmental noise as described in claim 3, characterized in that, In step S22, a preset sliding window is used to perform detrending processing on the triaxial velocity signal. The sliding window is set to a length of 100s and a step size of 25s.

5. The method for monitoring the mechanical state of ballasted railway track bed based on high-frequency environmental noise as described in claim 3, characterized in that, In step S23, the specific steps for calculating the horizontal / vertical power spectral density of the new triaxial velocity signal and iteratively calculating the average spectral ratio curve from the horizontal / vertical power spectral density are as follows: The new triaxial velocity signal is processed using a sliding time window method, dividing the continuous time period into multiple overlapping time periods, and the time spectrum of the signal is calculated by using short-time Fourier transform for each time period. The spectral power density in three directions is obtained from the time spectrum. After synthesizing the horizontal vibration power spectrum, the ratio curve of the horizontal to the vertical power spectrum is calculated using the following formula: in, , : The power spectral density of the j-th window of two orthogonal horizontal components; : The vertical component power spectral density of the j-th window; The average spectral ratio curve is calculated using the following formula: Where n: number of windows; : The ratio of the horizontal to the vertical power spectrum of the j-th window.

6. The method for monitoring the mechanical state of ballasted railway track bed based on high-frequency environmental noise as described in claim 1, characterized in that, In step S2, after obtaining the newly sampled window, the average spectral ratio curve is iteratively updated, and the average spectral ratio curve of the t-th update is calculated using the following formula: Where n: number of windows; i: sliding step size; The sliding window is calculated and updated in real time using the following recursive iterative formula: 。 7. The method for monitoring the mechanical state of ballasted railway track bed based on high-frequency environmental noise as described in claim 1, characterized in that, In step S3, for the spectral ratio curve updated in each iteration, the shear wave velocity of the track bed obtained from the initial inversion is used to perform a fast inversion using a local optimization algorithm to update the track bed mechanical parameters; the global inversion algorithm uses the Monte Carlo algorithm, and the local optimization inversion uses the downslope method.

8. A monitoring system for the mechanical state of ballasted railway track bed based on high-frequency environmental noise, characterized in that, include: The signal acquisition module is used to acquire triaxial acceleration signals under high-frequency environmental noise conditions on ballasted railways. The time-frequency processing module is used to integrate the triaxial acceleration signal to obtain a triaxial velocity signal, calculate the horizontal / vertical power spectral density based on the triaxial velocity signal, and calculate the average spectral ratio curve based on the horizontal / vertical power spectral density. The inversion evaluation module is used to construct an initial basic structure model based on prior information of the track bed and subgrade. Based on the initial basic structure model, a global inversion algorithm is used to invert the average spectral ratio curve to obtain the track bed shear wave velocity. The mechanical state of the track bed is determined based on the track bed shear wave velocity.

9. An electronic device comprising a memory and a processor, characterized in that, The memory is used to store a program, and the processor is coupled to the memory to execute the program stored in the memory to complete the steps in the method for monitoring the mechanical state of ballast railway track bed based on high-frequency environmental noise as described in any one of claims 1-7.

10. A non-transitory computer-readable storage medium for storing a readable computer program, characterized in that, When executed by a processor, the computer program is able to complete the steps in the method for monitoring the mechanical state of ballast track bed based on high-frequency environmental noise as described in any one of claims 1-7.

Images