A method, system and storage medium for determining railway roadbed deformation warning level

Abstract

The present invention provides a method, system and storage medium for determining the early warning level of railway roadbed deformation, the method comprising: obtaining deformation monitoring data of railway roadbed, fitting deformation monitoring data with a curve trend prediction model to obtain continuous deformation trend features, and discretizing the continuous deformation trend features to obtain multiple deformation trend intervals; obtaining characteristic values ​​of multi-dimensional deformation statistical indicators of each deformation trend interval based on the roadbed settlement value of the data point in the deformation trend interval, comparing the characteristic values ​​of the deformation statistical indicators of each deformation trend interval for each deformation statistical indicator, thereby obtaining the deformation characteristic values ​​of the corresponding deformation statistical indicators; calculating the correlation coefficient between the deformation characteristic values ​​of each deformation statistical indicator and the deformation center characteristic values ​​corresponding to each deformation statistical indicator, integrating to obtain the total correlation coefficient, and determining the deformation early warning level of the railway roadbed based on the size of the total correlation coefficient. The present invention can achieve early warning of railway roadbed deformation.

Description

Technical Field

[0001] The present invention relates to the technical field of railway roadbed deformation monitoring, and in particular to a method and system for determining a railway roadbed deformation early warning level. Background Art

[0002] The deformation of existing railway subgrades can be monitored by using the interferometric synthetic aperture radar (InSAR) time-series monitoring technology, or by using the Beidou high-precision positioning system to conduct millimeter-level continuous automatic monitoring of railway subgrades, bridges, and slopes around the clock. However, most of the current monitoring systems are passive emergency control systems, and human intervention is only carried out after the accumulated deformation of the subgrade reaches the set system deformation threshold. This not only leads to high repair costs, but also great difficulty in repair.

[0003] Since roadbed settlement is a dynamic evolution process over a larger time scale, the roadbed has actually entered a creep deformation stage of slow, uniform, small-scale sinking or sliding before reaching the system alarm threshold. Therefore, it is urgent to conduct active risk warning before the deformation reaches the threshold, thereby realizing early warning of roadbed deformation. Summary of the invention

[0004] In view of this, an embodiment of the present invention provides a method and system for determining a railway roadbed deformation warning level, so as to achieve early warning of railway roadbed deformation and improve the efficiency of roadbed repair.

[0005] One aspect of the present invention provides a method for determining a railway roadbed deformation warning level, the method comprising the following steps:

[0006] Obtaining deformation monitoring data of the railway roadbed, fitting the deformation monitoring data with a curve trend prediction model to obtain continuous characteristics of the deformation trend, and discretizing the continuous characteristics of the deformation trend to obtain multiple deformation trend intervals; wherein each deformation trend interval includes multiple data points for indicating the corresponding relationship between the roadbed settlement value and time;

[0007] Based on the roadbed settlement value of the data point in the deformation trend interval, the characteristic value of the multi-dimensional deformation statistical index of each deformation trend interval is obtained. For each deformation statistical index, the characteristic value of the deformation statistical index of each deformation trend interval is compared to obtain the deformation characteristic value of the corresponding deformation statistical index.

[0008] The correlation coefficient between the deformation characteristic value of each deformation statistical indicator and the deformation center characteristic value corresponding to each deformation statistical indicator is calculated to obtain the correlation coefficient corresponding to each deformation statistical indicator, the correlation coefficient corresponding to each deformation statistical indicator is integrated to obtain the total correlation coefficient, and the deformation warning level of the railway roadbed is determined based on the size of the total correlation coefficient; wherein, the deformation center characteristic value corresponding to each deformation statistical indicator is obtained by clustering the deformation characteristic values ​​of the corresponding deformation statistical indicators calculated based on multiple groups of deformation monitoring data obtained in advance.

[0009] In some embodiments of the present invention, obtaining deformation monitoring data of a railway roadbed includes: obtaining original deformation monitoring data of the railway roadbed, performing data cleaning on the original deformation monitoring data to obtain initial deformation monitoring data, and performing noise reduction processing on the initial deformation monitoring data to obtain second noise reduction deformation monitoring data;

[0010] The initial deformation monitoring data is subjected to denoising to obtain second denoised deformation monitoring data, including:

[0011] Using a wavelet threshold denoising algorithm to denoise the initial deformation monitoring data to obtain first denoised deformation monitoring data;

[0012] The first noise reduction deformation monitoring data is divided into a plurality of intrinsic mode components and residual components by using an empirical mode decomposition algorithm, and the autocorrelation coefficient of each intrinsic mode component is calculated by using an autocorrelation function;

[0013] The wavelet threshold denoising algorithm is used to denoise the intrinsic mode components whose autocorrelation coefficients are higher than the set threshold, and the denoised intrinsic mode components, the intrinsic mode components whose autocorrelation coefficients are not higher than the set threshold, and the residual components are reconstructed to obtain the second denoised deformation monitoring data.

[0014] In some embodiments of the present invention, the method also includes: using the second noise reduction deformation monitoring data to form a corresponding frequency spectrum diagram and power spectrum density diagram, and evaluating the frequency spectrum diagram and power spectrum density diagram according to a set noise evaluation index to evaluate the effect of the noise reduction processing; wherein the set noise evaluation index includes one or more evaluation indicators of signal-to-noise ratio, waveform similarity parameter, mean square error and root mean square error.

[0015] In some embodiments of the present invention, the deformation trend interval is obtained by:

[0016] Based on the same monitoring time interval, the continuous feature of the deformation trend is discretized into multiple deformation trend intervals; or, based on the same number of data points, the continuous feature of the deformation trend is discretized into multiple deformation trend intervals.

[0017] In some embodiments of the present invention, the curve trend prediction model includes a hyperbola model, a Hoshino method model, an exponential curve model or a Gopperts model;

[0018] The deformation statistical indicators are mean, variance, range, slope, Rayleigh entropy, generalized Hurst index, long-range correlation characteristics or uncorrelation coefficient; among them, the slope is obtained by fitting the roadbed settlement values ​​of data points in the deformation trend interval using the least squares method.

[0019] In some embodiments of the present invention, for each deformation statistical indicator, the characteristic values ​​of the deformation statistical indicator in each deformation trend interval are compared to obtain the deformation characteristic value of the corresponding deformation statistical indicator, including:

[0020] For each deformation statistical indicator, the range of the characteristic value of the deformation statistical indicator in each deformation trend interval is calculated, and the range of the characteristic value is used as the deformation characteristic value of the deformation statistical indicator.

[0021] In some embodiments of the present invention, the greater the total correlation coefficient, the lower the deformation warning level;

[0022] The eigenvalues ​​of the deformation feature centers include the eigenvalues ​​of the intangible deformation cluster centers and the eigenvalues ​​of the tangible deformation cluster centers.

[0023] In some embodiments of the present invention, the correlation coefficient corresponding to each deformation statistical indicator is the degree of membership corresponding to each deformation statistical indicator;

[0024] The membership degree corresponding to each deformation statistical index is obtained in the following way:

[0025] The membership function is used to calculate the degree of membership of the deformation eigenvalue of each deformation statistical indicator to the eigenvalue of the non-deformation clustering center corresponding to the corresponding deformation statistical indicator, and the membership corresponding to each deformation statistical indicator is obtained.

[0026] Another aspect of the present invention provides a system for determining a railway roadbed deformation warning level, comprising a processor, a memory, and a computer program / instructions stored in the memory, wherein the processor is used to execute the computer program / instructions, and when the computer program / instructions are executed, the system implements the steps of the method described in any of the above embodiments.

[0027] Another aspect of the present invention provides a computer-readable storage medium having a computer program / instruction stored thereon, which implements the steps of the method described in any of the above embodiments when the computer program / instruction is executed by a processor.

[0028] The present invention proposes a method and system for determining the early warning level of railway roadbed deformation, which can use data-driven means to mine the trend characteristics of roadbed deformation monitoring data, and determine the roadbed deformation risk level through the correlation coefficient between data points, thereby realizing active early warning of roadbed deformation.

[0029] Additional advantages, purposes, and features of the present invention will be described in part in the following description, and will become apparent to those skilled in the art after studying the following, or may be learned from the practice of the present invention. The purposes and other advantages of the present invention may be achieved and obtained by the structures specifically indicated in the specification and the accompanying drawings.

[0030] Those skilled in the art will appreciate that the objectives and advantages that can be achieved with the present invention are not limited to the above specific description, and the above and other objectives that can be achieved by the present invention will be more clearly understood from the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The drawings described herein are used to provide a further understanding of the present invention, constitute a part of the present application, and do not constitute a limitation of the present invention. In the drawings:

[0032] Figure 1 The figure is a flow chart of a method for determining a railway roadbed deformation warning level in one embodiment of the present invention.

[0033] Figure 2 The figure is a schematic diagram of a process of obtaining second noise-reduced deformation monitoring data based on original deformation monitoring data in one embodiment of the present invention.

[0034] Figure 3 Schematic diagram of the calculation process of deformation characteristic values ​​of various deformation statistical indicators in one embodiment of the present invention.

[0035] Figure 4 The figure is a flow chart of obtaining the deformation center feature value by using the K-means clustering algorithm in one embodiment of the present invention.

[0036] Figure 5 The figure is a flow chart of simulation analysis using a method for determining a railway roadbed deformation warning level in one embodiment of the present invention.

[0037] Figure 6 This is a graph of original deformation monitoring data of a selected section during a simulation in one embodiment of the present invention.

[0038] Figure 7 This is an initial deformation monitoring data diagram obtained through preprocessing during the simulation process in one embodiment of the present invention.

[0039] Figure 8This is a frequency spectrum diagram formed based on initial deformation monitoring data during a simulation process in one embodiment of the present invention.

[0040] Fig. 9 It is a power spectrum density diagram formed based on the initial deformation monitoring data in the simulation process in one embodiment of the present invention.

[0041] Fig.10 This is a first noise reduction deformation monitoring data diagram obtained during a simulation process in one embodiment of the present invention.

[0042] Fig.11 It is a component graph obtained by decomposing the first denoised deformation monitoring data using the EMD algorithm during the simulation process in one embodiment of the present invention.

[0043] Fig.12 Schematic diagram of the autocorrelation function analysis results of each EMD component during the simulation process in one embodiment of the present invention.

[0044] Fig.13 This is a frequency spectrum diagram formed according to the second noise reduction deformation monitoring data in one embodiment of the present invention.

[0045] Fig.14 It is a power spectrum density diagram formed according to the second noise reduction deformation monitoring data in one embodiment of the present invention.

[0046] Fig.15 This is a result diagram of fitting the second noise reduction deformation monitoring data using a hyperbola model in one embodiment of the present invention.

[0047] Fig.16 This is a diagram showing the deformation statistical index analysis results of each deformation trend interval of a selected segment in an embodiment of the present invention.

[0048] Fig.17 It is a partial clustering result diagram of deformation characteristic values ​​of various deformation statistical indicators calculated based on multiple groups of deformation monitoring data obtained in advance in one embodiment of the present invention in three-dimensional coordinates. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments and the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0050] It should also be noted that, in order to avoid obscuring the present invention due to unnecessary details, only structures and / or processing steps closely related to the solutions according to the present invention are shown in the accompanying drawings, while other details that are not closely related to the present invention are omitted.

[0051] It should be emphasized that the term “include / comprises” when used herein refers to the presence of features, elements, steps or components, but does not exclude the presence or addition of one or more other features, elements, steps or components.

[0052] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the accompanying drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0053] The current railway roadbed deformation monitoring system mainly sets the system deformation threshold, and then conducts human intervention after the cumulative deformation of the roadbed reaches a certain level. It belongs to passive emergency control, so the later repair process is relatively complex, difficult and costly. Based on this, the present application proposes a method for determining the early warning level of railway roadbed deformation, which can realize active early warning of roadbed deformation in the creep deformation stage of the roadbed to improve the efficiency of roadbed repair. The present application can use the curve trend prediction model to fit the deformation trend characteristics of the roadbed deformation monitoring data, calculate the deformation characteristic values ​​corresponding to each indicator based on the deformation trend characteristics and multi-dimensional deformation statistical indicators, and determine the roadbed deformation risk level according to the correlation coefficient between the deformation characteristic value and the deformation center characteristic value.

[0054] Figure 1 FIG. 1 is a flow chart of a method for determining a railway roadbed deformation warning level in an embodiment of the present invention. Figure 1 As shown, the method includes steps S110 to S130.

[0055] Step S110: Obtain deformation monitoring data of the railway roadbed, and use the curve trend prediction model to fit the deformation monitoring data to obtain continuous features of the deformation trend, and discretize the continuous features of the deformation trend to obtain multiple deformation trend intervals.

[0056] The deformation monitoring data can show how the railway subsidence or displacement changes over time. For example, the present application can obtain the time-series monitoring data of railway subgrade deformation from the Beidou deformation monitoring system.

[0057] In some embodiments of the present invention, obtaining deformation monitoring data of a railway roadbed includes: obtaining original deformation monitoring data of the railway roadbed, performing data cleaning on the original deformation monitoring data to obtain initial deformation monitoring data, and performing noise reduction processing on the initial deformation monitoring data to obtain second noise reduction deformation monitoring data. Specifically, performing noise reduction processing on the initial deformation monitoring data to obtain the second noise reduction deformation monitoring data is as follows:

[0058] Using a wavelet threshold denoising algorithm to denoise the initial deformation monitoring data to obtain first denoised deformation monitoring data;

[0059] The first noise reduction deformation monitoring data is divided into a plurality of intrinsic mode components and residual components by using an empirical mode decomposition (EMD) algorithm, and the autocorrelation coefficient of each intrinsic mode component is calculated by using an autocorrelation function;

[0060] The wavelet threshold denoising algorithm is used to denoise the intrinsic mode components whose autocorrelation coefficients are higher than the set threshold, and the denoised intrinsic mode components, the intrinsic mode components whose autocorrelation coefficients are not higher than the set threshold, and the residual components are reconstructed to obtain the second denoised deformation monitoring data.

[0061] More specifically, the original deformation monitoring data of the railway roadbed is affected by interference factors such as multipath error and external noise, and the amplitude range of its settlement is much wider than the actual deformation of the roadbed. Therefore, after obtaining the original deformation monitoring data, data cleaning and denoising are required to eliminate the interference of external factors on the actual deformation trend. Considering that the wavelet threshold denoising algorithm can effectively eliminate the multipath error in the monitoring data, the empirical mode decomposition method can have a good denoising effect on the external noise, and the EMD decomposition and wavelet threshold denoising have the advantages of flexible basis function selection, this application proposes to integrate the wavelet threshold denoising algorithm and the EMD algorithm to perform denoising on the initial deformation monitoring data after data cleaning, thereby effectively retaining the trend characteristics of the monitoring data while removing external noise interference.

[0062] As an example, data cleaning preprocessing of the acquired raw deformation monitoring data includes steps such as removing outlier data (such as data such as abnormal values ​​and missing values).

[0063] The specific process of data cleaning and denoising of the original deformation monitoring data obtained in this application is as follows: Figure 2 As shown, including:

[0064] Since the higher the frequency, the greater the external noise interference, the initial deformation monitoring data can be divided into set high-frequency monitoring data and non-set high-frequency monitoring data. The wavelet threshold denoising algorithm is used to denoise the set high-frequency monitoring data, and the denoised set high-frequency monitoring data and the retained non-set high-frequency monitoring data (that is, the non-set high-frequency monitoring data are not denoised) are reconstructed to obtain the first denoised deformation monitoring data. The first denoised deformation monitoring data is decomposed by the EMD algorithm to obtain multiple intrinsic mode components (IMFs) and a residual component, and the autocorrelation function is used to calculate the autocorrelation coefficient of each intrinsic mode component; among them, if the autocorrelation coefficient of the IMF is higher than the set threshold, it is a noise-free IMF; if the autocorrelation coefficient of the IMF is lower than the set threshold, it is a noisy IMF. The wavelet threshold denoising algorithm is used to denoise the noisy IMFs, and the denoised noisy IMFs, the retained noise-free IMFs and the residual component are reconstructed to obtain the second denoised deformation monitoring data.

[0065] As an example, the wavelet basis, threshold function, threshold and other parameters in the wavelet threshold denoising algorithm can be selected according to the noise reduction requirements and sample data, and the threshold function and other parameters of the wavelet threshold denoising algorithm used when obtaining the first noise reduction deformation monitoring data and the second noise reduction deformation monitoring data in the noise reduction process of the present application can be different.

[0066] The frequency and threshold corresponding to the high-frequency monitoring data mentioned above can be determined according to the noise reduction requirements, and the high frequency is set relative to the overall frequency of the deformation monitoring data. In addition, the above-mentioned process of using the wavelet threshold denoising algorithm and the EMD algorithm to perform noise reduction on the initial deformation monitoring data is only an example, and the present invention can also use other methods to perform noise reduction, and the present invention is not limited to this.

[0067] Furthermore, the method for determining the railway roadbed deformation warning level proposed in the present application also includes: using the second noise reduction deformation monitoring data to form a corresponding frequency spectrum and power spectrum density diagram, and evaluating the formed frequency spectrum and power spectrum density diagram according to the set noise evaluation index to evaluate the effect of the noise reduction processing. Among them, the noise evaluation index can be set by itself, and the present invention does not specifically limit it. For example, the set noise evaluation index may include one or more evaluation indicators of signal-to-noise ratio, waveform similarity parameter, mean square error and root mean square error.

[0068] If the denoising effect cannot meet the set noise condition, reset the threshold and threshold function parameters in the wavelet threshold denoising algorithm and the corresponding parameters in the EMD algorithm, and repeat the same process using the reset wavelet threshold algorithm and EMD algorithm. Figure 2 The processing process shown is used to obtain the second noise reduction deformation monitoring data that meets the set noise conditions.

[0069] As an example, if the acquired original deformation monitoring data can meet the set noise condition, there is no need to perform noise reduction processing on it.

[0070] In some embodiments of the present invention, the present application selects deformation monitoring data acquired by fitting a curve trend prediction model to obtain continuous characteristics of deformation trend. Specifically, the parameters of the curve trend prediction model can be determined by parameter fitting and other methods, and then the fitting and prediction of the roadbed settlement amount in the roadbed creep deformation stage can be realized based on the deformation monitoring data (which can be the above-mentioned second noise reduction deformation monitoring data), so as to more clearly and intuitively display the roadbed deformation trend.

[0071] The deformation monitoring data in this application can represent the change of the monitored roadbed settlement over time; the deformation trend continuous feature is obtained by fitting the curve trend prediction model or predicting the roadbed settlement in the deformation monitoring data, so the deformation trend continuous feature can be used to represent the change of the predicted value or fitted value of the roadbed settlement over time. To simplify the expression, the predicted value of the roadbed settlement (or the fitted value of the roadbed settlement) is collectively referred to as the roadbed settlement value in the following expression.

[0072] As an example, the deformation trend continuous feature may be a deformation trend line obtained by fitting deformation monitoring data.

[0073] More specifically, the curve trend prediction method is also called the non-linear trend prediction method, which can be predicted using a curve trend model, and its prediction results can show the trend of continuous growth or decrease of the time series over a long period of time. Commonly used curve trend models are parabolic and simple exponential. For example, the curve trend prediction model in this application can be a variety of sedimentation fitting models such as a hyperbolic model, a Hoshino method model, an exponential curve model or a Gompertz model.

[0074] As an example, the hyperbolic model is a relatively simple two-parameter model that conforms to the actual change trend and is widely used in actual engineering. The expression of the hyperbolic model is as follows:

[0075] ;

[0076] in, Indicates that over time The total amount of sedimentation changes, represents the initial settlement, and and represent the initial and elapsed time of the settlement, respectively, and the parameters and is the coefficient to be determined.

[0077] The hyperbolic model can also be transformed into:

[0078] ;

[0079] From the above formula, we can see that the parameters and They correspond to the slope and intercept of the continuous characteristics of the deformation trend, which can be solved based on the deformation monitoring data. and Substituting the expression of the hyperbolic model, the fitted settlement at any time can be obtained.

[0080] Furthermore, the present application can use the existing feature discretization method to discretize the continuous feature of the deformation trend to obtain multiple deformation trend intervals. For example, the trend line can be discretized to obtain multiple data points, and the discretized data points can be divided into each deformation trend interval.

[0081] In some embodiments of the present invention, considering that roadbed settlement is a slowly changing process and the monitoring time may be long, in order to better express the deformation trend, the continuous characteristics of the deformation trend may be discretized into multiple deformation trend intervals, including:

[0082] Based on the same monitoring time interval, the continuous deformation trend feature is discretized into multiple deformation trend intervals; or based on the same number of data points, the continuous deformation trend feature is discretized into multiple deformation trend intervals. Each deformation trend interval includes multiple data points, which are used to indicate the corresponding relationship between the roadbed settlement value and time.

[0083] Step S120: Based on the roadbed settlement value of the data point in the deformation trend interval, the characteristic value of the multi-dimensional deformation statistical index of each deformation trend interval is obtained. For each deformation statistical index, the characteristic value of the deformation statistical index of each deformation trend interval is compared to obtain the deformation characteristic value of the corresponding deformation statistical index. The specific process of obtaining the deformation characteristic value of each deformation statistical index based on the roadbed settlement value can be as follows: Figure 3 shown.

[0084] The multidimensional deformation statistical index is a statistical index that can measure the change of roadbed settlement value, and may include at least two of the mean, variance, range, slope, Renyi entropy, generalized Hurst index (Multifractal Detrended Fluctuation Analysis, MFDFA), long-range correlation characteristics and irrelevance coefficient. Among them, the mean can be used to measure the change of the average settlement value of the sample points in each deformation trend interval, the range measures the overall change of the roadbed settlement value of the sample points in each deformation trend interval, the significance of the slope is to measure the change of the settlement trend of the sample points in each deformation trend interval, the significance of the variance is to measure the overall fluctuation of the roadbed settlement value of the sample points in each deformation trend interval, the Renyi entropy can be used to measure the stability of the roadbed settlement value of the sample points in each deformation trend interval, the significance of the generalized Hurst index (i.e., multi-fractal trend fluctuation analysis) is to measure the fractal situation of the sample points in each deformation trend interval, the long-range correlation characteristics can be used to quantitatively describe the autocorrelation of the time series of each deformation trend interval on a larger time scale, and the significance of the uncorrelated coefficient is to show the degree of dispersion of the sample points in each deformation trend interval. Among them, the characteristic value of the slope of each deformation trend interval can be obtained by fitting the roadbed settlement value of the sample points in each deformation trend interval using the least squares method.

[0085] The multi-dimensional deformation statistical indicators mentioned above are only examples, and other statistical indicators may also be selected, and the present application is not limited thereto.

[0086] In some embodiments of the present invention, for each deformation statistical indicator, the characteristic values ​​of the deformation statistical indicator in each deformation trend interval are compared to obtain the deformation characteristic value of the corresponding deformation statistical indicator, including:

[0087] For each deformation statistical index, the range of the deformation statistical index characteristic value in each deformation trend interval is calculated, and the calculated range is used as the deformation characteristic value of the deformation statistical index.

[0088] Taking mean and variance as deformation statistical indicators as an example, suppose there is a set of deformation monitoring data A, and three deformation trend intervals a1, a2 and a3 are obtained according to the deformation monitoring data A. The eigenvalues ​​of the means of the three deformation trend intervals are calculated according to the roadbed settlement values ​​of the data points in a1, a2 and a3, and the eigenvalues ​​corresponding to the variances of a1, a2 and a3 are calculated respectively, so as to obtain the range c of the eigenvalues ​​of the means of a1, a2 and a3 (assuming that the eigenvalues ​​of the means of a1, a2 and a3 are c1, c2 and c3 respectively, and c1>c2>c3, then c=c1-c3), and the range d of the eigenvalues ​​corresponding to the variances of a1, a2 and a3 (assuming that the eigenvalues ​​of the variances of a1, a2 and a3 are d1, d2 and d3 respectively, and d2>d1>d3, then d=d2-d3). Therefore, the deformation eigenvalue of the mean of the deformation monitoring data A is c, and the deformation eigenvalue of the variance is d.

[0089] The above-mentioned method of using the extreme difference of the characteristic value of the corresponding deformation statistical index as the deformation characteristic value is only an example, and other comparison methods can also be used, such as using a BP neural network to divide the deformation statistical index into two categories: unclear characteristics and obvious characteristics, and using a binary classification method to obtain the unclear characteristic center point and the obvious characteristic center point. The present invention does not make specific limitations on this.

[0090] As an example, the deformation characteristic value of the corresponding deformation statistical index obtained by comparing the characteristic value of each deformation statistical index in each deformation trend interval can be regarded as a one-dimensional coordinate point. Similarly, the characteristic value of the deformation cluster center described below can also be regarded as a one-dimensional coordinate point.

[0091] The key to classifying the warning level of roadbed deformation is to extract deformation characteristics. In step S120, the characteristic values ​​of deformation statistical indicators in each deformation trend interval are calculated to obtain the deformation characteristics of roadbed settlement within a specific time scale range.

[0092] Step S130: Calculate the correlation coefficient between the deformation characteristic value of each deformation statistical indicator and the deformation center characteristic value corresponding to each deformation statistical indicator to obtain the correlation coefficient corresponding to each deformation statistical indicator, integrate the correlation coefficients corresponding to each deformation statistical indicator to obtain the total correlation coefficient, and determine the deformation warning level of the railway roadbed based on the size of the total correlation coefficient; wherein the deformation center characteristic value corresponding to each deformation statistical indicator is obtained by clustering the deformation characteristic values ​​of the corresponding deformation statistical indicators calculated based on multiple groups of deformation monitoring data obtained in advance.

[0093] In some embodiments of the present invention, when determining the deformation center eigenvalues ​​corresponding to various deformation statistical indicators through clustering, the deformation feature center eigenvalues ​​may be set to include eigenvalues ​​of intangible deformation cluster centers and eigenvalues ​​of tangible deformation cluster centers.

[0094] As an example, the K-means algorithm is selected to classify the deformation feature values ​​of the corresponding deformation statistical indicators calculated based on multiple sets of deformation monitoring data obtained in advance, and determine the center value of the deformation feature, so as to effectively distinguish between deformable data and non-deformable data. Figure 4 As shown, the specific process of the K-means algorithm can be described as: The deformation eigenvalue corresponding to a certain deformation statistical index ( is the number of groups of deformation monitoring data collected in advance), determine the number of clusters required for clustering (For example, two cluster centers can be set to divide the data into tangible clusters or intangible clusters, that is, ); Then the K-means clustering algorithm is used to cluster the features according to the distance between them (such as Manhattan distance). The samples are assigned to The distance between samples in different clusters is as large as possible, while the distance between deformation eigenvalues ​​in the same cluster is as small as possible, that is, the sum of squared errors within the cluster is as small as possible. The calculation formula for the sum of squared errors within the cluster is as follows:

[0095] ;

[0096] in, Deformation eigenvalue The cluster center of the cluster to which it belongs in the iterative process.

[0097] In addition, when using the K-means algorithm for clustering, it is necessary to repeat the iterative allocation step and the update step until the set convergence condition is met, and then the iterative process ends. Among them, the allocation step is used to select the distance deformation eigenvalue The nearest cluster The update step requires calculating the mean of the samples in the cluster at the beginning of each iteration and using it as the new cluster center. , the formulas of the allocation step and the update step can be expressed as:

[0098] ;

[0099] ;

[0100] in, Indicates The number of deformation eigenvalues ​​in a cluster.

[0101] As an example, the process of calculating the deformation feature values ​​of each deformation statistical index from multiple sets of deformation monitoring data obtained in advance can be obtained in the same manner as step S110 and step S120. In addition, to ensure the accuracy of the deformation feature center value, more monitoring data can be selected for clustering.

[0102] The above-mentioned method of using the K-means clustering algorithm to determine the deformation feature center value is only an example, and other clustering algorithms can be used. The present invention does not limit the specific method for determining the deformation center feature value.

[0103] Furthermore, after the deformation eigenvalues ​​corresponding to the deformation statistical indicators are determined in step S120, it is necessary to calculate the correlation coefficients between the deformation eigenvalues ​​and the deformation center eigenvalues ​​of the corresponding deformation statistical indicators according to the correlation between the data values, for example, to calculate the correlation coefficients between the deformation eigenvalues ​​of the MFDFA and the predetermined deformation center eigenvalues ​​of the MFDFA.

[0104] In some embodiments of the present invention, after the cluster analysis is completed, the membership degree of the deformation eigenvalues ​​to different cluster centers (i.e., the deformation center eigenvalues) can be quantified according to the membership function, so as to judge the stability of the railway subgrade and determine the risk level of the railway subgrade deformation.

[0105] As an example, if the above-mentioned K-means clustering algorithm is used in advance to obtain the invisible deformation cluster center and the visible deformation cluster center, the membership function can be used to calculate the degree of membership of the deformation eigenvalue of each deformation statistical index to the eigenvalue of the invisible deformation cluster center corresponding to the corresponding deformation statistical index, thereby obtaining the membership corresponding to each deformation statistical index; the higher the degree of membership, the greater the correlation coefficient between the deformation eigenvalue and the eigenvalue of the invisible deformation cluster center. This is because the deformation eigenvalues ​​in the invisible deformation cluster cluster are highly aggregated, while the deformation eigenvalues ​​in the visible deformation cluster cluster are more dispersed. Therefore, the size of the membership function of the deformation eigenvalue and the eigenvalue of the invisible deformation cluster center can be used as the main basis for judging the stability of the roadbed.

[0106] Membership function is a core concept in fuzzy mathematics. Its interval range is between [0,1]. It is mainly used to describe the degree of membership of an individual to a set. It can be expressed as:

[0107] ;

[0108] in, Indicates The deformation eigenvalues ​​and The membership degree of the eigenvalues ​​of the cluster centers ranges from: ,when , which represents the deformation eigenvalue Completely belongs to the category , that is, the deformation eigenvalue With The eigenvalues ​​of the cluster centers are completely correlated. When represents the deformation eigenvalue Not at all in the category , that is, deformation feature points With The eigenvalues ​​of the cluster centers are completely uncorrelated. represents the number of cluster centers (i.e. the number of deformation feature center points), Indicates The deformation eigenvalues ​​and The distance between the eigenvalues ​​of cluster centers, Indicates The deformation eigenvalues ​​and The distance between the eigenvalues ​​of cluster centers, is the fuzzy weighted index.

[0109] The membership function mentioned above is only an example, and the present invention is not limited thereto. Other algorithms for measuring the correlation coefficient between the deformation feature value and the feature value of the deformation feature center may also be used (for example, the distance is proportional to the correlation coefficient, and the farther the distance, the smaller the correlation coefficient).

[0110] After obtaining the correlation coefficients corresponding to the deformation statistical indicators, the correlation coefficients corresponding to the deformation statistical indicators are integrated to obtain the total correlation coefficient corresponding to the deformation monitoring data. The correlation coefficients corresponding to the deformation statistical indicators can be integrated by weighted average, or other integration methods can be used, and the present invention is not limited thereto.

[0111] In addition, the railway roadbed deformation risk level can be divided according to the rule that the larger the total correlation coefficient, the lower the deformation warning level.

[0112] Taking the above-mentioned K-means clustering algorithm for cluster analysis and using the membership function as an evaluation indicator of the warning level as an example, the warning levels shown in Table 1 can be obtained. As can be seen from Table 1, the degree of roadbed settlement can be divided into 9 categories, and the warning levels are determined according to the corresponding membership function values. Among them, non-deformation ≥ 0.9 means that the total membership (i.e., total correlation coefficient) of all deformation eigenvalues ​​to the eigenvalue of the non-deformation cluster center is not less than 0.9, and deformed deformation ≤ 0.1 means that the total membership of all deformation eigenvalues ​​to the eigenvalue of the deformed cluster center is not greater than 0.1.

[0113] Table 1 Warning level classification standards

[0114]

[0115] Since the deformation eigenvalues ​​in the invisible deformation clusters are highly aggregated, while the deformation eigenvalues ​​in the visible deformation clusters are more dispersed, when dividing the roadbed deformation warning level, the total membership of the deformation eigenvalue and the eigenvalue of the invisible deformation cluster center can be used as the main basis for judging the stability of the roadbed, and the total membership of the deformation eigenvalue and the eigenvalue of the visible deformation cluster center should be used as an auxiliary basis for judging the stability of the roadbed. Therefore, the warning level division table can be simplified as Table 2.

[0116] Table 2 Warning level classification standards

[0117]

[0118] As an example, when the warning level is level 4, the warning signal may be blue; when the warning level is level 5, the warning signal may be yellow; when the warning level is level 6, the warning signal may be orange; when the warning level is level 7, the warning signal may be red. In addition, since the purpose of this application is to issue a warning when entering a creeping deformation stage of slow, uniform, small amount of sinking or sliding, no warning information may be sent in the safe and dangerous categories.

[0119] The above classification of the roadbed settlement degree into 9 categories is only an example, and the present invention is not limited thereto.

[0120] In a specific embodiment of the present invention, in order to verify the effect of the method for determining the railway roadbed deformation warning level proposed in this application, a typical section of the roadbed located in the northwest Loess Plateau with complex geological conditions and a long-term observation of overall subsidence risk was selected, and simulation analysis was performed based on the measured data of the section. Figure 5 As shown in the figure, the specific process of simulation analysis is as follows: the original deformation monitoring data is obtained from the Beidou deformation monitoring system, and the wavelet threshold denoising and EMD algorithm are integrated for noise reduction processing. The hyperbola model is used to fit the noise-reduced deformation monitoring data to obtain the roadbed deformation trend line. The mean, variance, range, Renyi entropy, slope and MFDFA are selected to extract multi-dimensional deformation trend feature information. The K-means clustering analysis model is used in advance to obtain the visible deformation clustering clusters and the invisible deformation clustering clusters, and the roadbed deformation risk level is determined according to the membership function, so as to realize the active early warning of roadbed deformation.

[0121] More specifically, during the data collection phase of the simulation analysis, the maximum distance between the base station and the monitoring station of the risk zone deformation monitoring system can be set to 300m, and the monitoring points can be evenly distributed on both sides of the railway. Using the Beidou deformation monitoring system to monitor the changes in roadbed settlement over time, and using a general high-precision static baseline solution program, the following can be obtained: Figure 6The original monitoring data shown in . The monitoring data spans from March 1, 2022 to March 31, 2022, a total of one month, with a data collection frequency of 10 seconds per epoch, a solution cycle of 10 minutes, and a total of 4320 data points.

[0122] right Figure 6 After the original monitoring data in the data cleaning preprocessing to remove outliers, the following Figure 7 The initial deformation monitoring data shown. Figure 6 The deformation trend of the monitoring data cannot be clearly observed in the data. After removing the outliers in the original deformation monitoring data (including removing bad points and supplementing lost points), Figure 7 It can be clearly observed that the amplitude range of the vertical axis is significantly reduced, the detailed characteristics of the monitoring data fluctuations are clearer, and the trend characteristics of the monitoring data at a larger time scale are more intuitive.

[0123] based on Figure 7 Initial deformation monitoring data formation in Figure 8 The spectrum diagram and Fig. 9 Power spectral density plot of . Figure 8 The energy distribution of the initial deformation monitoring data at different frequencies is given. Fig. 9 The power density of the initial deformation monitoring data at different frequencies is shown. Figure 8 and Fig. 9 It can be seen that the energy distribution of the initial deformation monitoring data is not uniform, which means that the initial deformation monitoring data is interfered by various noises.

[0124] Fig.10 The first denoised deformation monitoring data is obtained after the set high-frequency data in the initial deformation monitoring data is denoised using the wavelet threshold denoising algorithm. During the simulation process, the wavelet coefficients at different scales are calculated using the threshold modulus flattening method to obtain the threshold. Fig.10 It can be observed that the data burrs are less, and the data trend of the first noise reduction deformation monitoring data is consistent with the initial deformation monitoring data as a whole.

[0125] Fig.11 The figure shows the six IMF components with different frequencies and one residual component obtained after the first denoising deformation monitoring data is decomposed by the EMD algorithm. Considering that the main energy of most noisy data is concentrated in the lower frequency band, while the higher frequency band contains more noise, this application uses the difference in the autocorrelation function of the noise data as the basis for judging the denoising truncation order of the EMD component. The analysis results of the autocorrelation function of each EMD component are shown in Fig.12 shown.

[0126] Fig.15The blue line in the figure is the second denoised deformation monitoring data obtained after the first denoised deformation monitoring data is processed by the EMD algorithm and the wavelet threshold denoising algorithm. Fig.15 It can be observed that compared with the first denoised deformation monitoring data, the secondary denoising process using the EMD algorithm and wavelet threshold denoising algorithm can significantly reduce the noise interference in the monitoring data; compared with the original deformation monitoring data, the monitoring sequence (i.e. Fig.15 It can be clearly observed from the blue line in the figure that the roadbed in this section has a more obvious sinking trend during the one-month monitoring period, and has an even more obvious settlement trend after March 21.

[0127] Fig.13 and Fig.14 are respectively a spectrum diagram and a power spectrum density diagram formed according to the second noise reduction deformation monitoring data. Fig.13 and Fig.14 It can be seen that the power of all frequency bands except the main frequency (the lower frequency band of 0-50Hz) is suppressed, which indicates that the noise energy in the monitoring data is significantly weakened.

[0128] After denoising the deformation monitoring data, the hyperbolic model is used to further fit the second denoised deformation monitoring data, and the following is obtained: Fig.15 The hyperbola fitting results are shown. Fig.15 The blue line in the figure represents the second noise reduction deformation monitoring data, while the red line is the deformation trend line obtained by fitting the hyperbolic model. The deformation trend line obtained by fitting the hyperbolic model is highly consistent with the curve trend of the actual deformation monitoring data of the roadbed. Table 3 shows the results of the hyperbolic fitting parameters. In addition, since the deformation monitoring data obtained by the monitoring station is an absolute value, it is necessary to use the reference station position data to convert it into a relative value. Fig.15 The green dots in the figure are the relative position data of the base station and the monitoring station.

[0129] Table 3 Hyperbolic fitting parameters of deformation monitoring data

[0130]

[0131] In the simulation analysis process, this application selects six deformation statistical indicators, namely mean, variance, range, slope, Renyi entropy and MFDFA, to measure the settlement characteristics of the roadbed, and discretizes the deformation trend line into 50 deformation trend intervals according to the same monitoring time interval. The deformation statistical indicator analysis results of each deformation trend interval are shown in Figure 2. Fig.16 As shown, Fig.16 (a) in the figure represents the characteristic value of the slope of each deformation trend interval. Fig.16 (b) in the figure represents the characteristic value of the extreme value of each deformation trend interval. Fig.16 (c) in the figure represents the characteristic value of the mean of each deformation trend interval. Fig.16 (d) in the equation represents the characteristic value of the variance of each deformation trend interval. Fig.16 The (e) in the equation represents the characteristic value of Renyi entropy in each deformation trend interval. Fig.16 The (f) in represents the characteristic value of MFDFA in each deformation trend interval. Fig.16 It can be seen that the characteristic values ​​of each deformation statistical index in different intervals are obviously different, and their trends are similar to Fig.15 The monitoring data in the simulation are consistent, that is, they all have obvious upward or downward trends, which shows that the deformation statistical indicators selected in the simulation process can well extract the trend characteristics of the deformation monitoring data of the roadbed. After obtaining the characteristic values ​​of the deformation statistical indicators, the extreme values ​​of the characteristic values ​​of each indicator in the monitoring data are calculated to obtain the overall trend characteristics of the deformation monitoring data.

[0132] During the simulation process, this application selects 80 groups of pre-collected deformation monitoring data, and the corresponding deformation feature values ​​belong to 40 groups of deformation monitoring data with obvious deformation clusters and 40 groups of deformation monitoring data without obvious deformation clusters (the method proposed in this application does not require that the number of deformation monitoring data groups with corresponding deformation feature values ​​belonging to different clusters be consistent). After calculating the deformation feature points of 80 groups of deformation monitoring data according to the above simulation steps, the deformation feature values ​​are subjected to K-means cluster analysis. The eigenvalues ​​of the mean, variance, range, slope, Renyi entropy and MFDFA invisible deformation cluster center are [0.148, 0.391, 2.763, 0.415, 0.421, 0.045], and the eigenvalues ​​of the mean, variance, range, slope, Renyi entropy and MFDFA visible deformation cluster center are [2.692, 1.877, 10.2, 3.195, 1.922, 0.105].

[0133] Fig.17 Three deformation statistical indicators are selected as coordinate axes (six deformation statistical indicators are selected in the simulation analysis, which cannot be represented in the three-dimensional graph. Therefore, this application can arbitrarily select three indicators as coordinate axes and form three-dimensional coordinate points according to the corresponding deformation feature values). Partial clustering results of 80 sets of pre-obtained deformation monitoring data are displayed in three-dimensional coordinates. Fig.17 (a) in the figure is the clustering result with Renyi entropy, range and variance as coordinate axes. Fig.17 (b) in the figure is the clustering result with Renyi entropy, mean and variance as coordinate axes. Fig.17 (c) in the figure is the clustering result with Renyi entropy, variance and mean as coordinate axes. Fig.17The blue dots represent the data points formed by the deformation feature values ​​of the deformation monitoring data in the three-dimensional coordinate system, and the red squares represent the three-dimensional data points formed by the feature values ​​of the cluster centers (the cluster centers with deformation or the cluster centers without deformation). Fig.17 It can be clearly seen that by selecting the above six deformation feature dimensions, the cluster analysis method can distinguish the tangible deformation data from the intangible deformation data very well.

[0134] In addition, the present application uses cluster analysis to distinguish between deformable data and non-deformable data. The core of this is that cluster analysis belongs to unsupervised learning, and the cluster analysis results can further verify whether the deformation statistical indicators selected by the present application are reasonable and feasible.

[0135] The deformation monitoring data of the selected section in the simulation analysis is weighted averaged for the membership of the deformation characteristic values ​​formed by each deformation statistical index and the characteristic value of the invisible deformation clustering center of the corresponding index, and the total membership of the deformation monitoring data of the selected section to the invisible deformation clustering center is 0.64. According to Table 2, it can be seen that the roadbed has reached a low warning level.

[0136] As an example, the MFDFA algorithm can be expressed as follows:

[0137] Deformation trend continuous feature Follow the same monitoring interval Divide interval, The probability that a fractal in an interval is in this interval is , Different scaling indices available for:

[0138] ;

[0139] in, Indicates The interval length of the interval.

[0140] Taking the logarithm of the above formula and finding the limit gives:

[0141] ;

[0142] in, Represents the fractal dimension of a certain area of ​​the fractal, that is, the local fractal.

[0143] Multiply the formula obtained by finding the limit by And summing them up we get:

[0144] ;

[0145] definition Secondary information dimension for:

[0146] ;

[0147] from It is easy to get from the formula, The change can be achieved by choosing different scales to achieve this, in other words, by changing the scale Different fractal dimensions can be distinguished, thereby realizing the estimation of multi-fractal characteristics of monitoring data.

[0148] For the time series of deformation monitoring of civil engineering infrastructure To calculate its multifractal characteristic index, first reconstruct its cumulative deviation sequence:

[0149] ;

[0150] in, represents the mean value of deformation monitoring data, Represents the interval of the reconstructed sequence. The new sequence to be reconstructed is divided into equal parts of length subsequence, subsequence The number can be expressed as ,in is the floor function. Not necessarily the length of the interval Integer division, in order not to miss the tail data, the sub-intervals will be re-divided equally from the tail data, so the number of sub-intervals is 2 .

[0151] For each subinterval conduct The local trend function is obtained by least squares fitting:

[0152] ;

[0153] in, represents the polynomial coefficients, is the highest degree of the polynomial. By calculating the average variance To eliminate the subinterval Local trends within.

[0154] when hour:

[0155] ;

[0156] when hour:

[0157] ;

[0158] Calculation sequence Order fluctuation trend function:

[0159] ;

[0160] In order When the value is 0 or 2, the fluctuation trend function has an exponential power anomaly. It can be expressed as follows.

[0161] when hour:

[0162] ;

[0163] when hour:

[0164] ;

[0165] It is easy to observe that the above fluctuation trend function And monitoring data length and fractional order Related, and With interval length It increases at a power rate, that is ,in This is the required generalized Hurst index. and The power law relation can be rewritten as , taking its logarithm, we get:

[0166] ;

[0167] Where A is a constant. The slope in the above formula is Represents the generalized Hurst index, which is an important indicator for measuring the long-range correlation characteristics of monitoring data. It has the following conclusions: when different scales are selected hour, When the data also changes, it indicates that the monitoring data has multiple fractal characteristics, that is, the overall change trend of the data is unstable, and The greater the degree of change, the greater the overall change trend of the deformation monitoring data. hour, When it does not change accordingly, it indicates that this set of monitoring data has a single fractal characteristic, that is, the overall change trend of the data is relatively stable, and the future change trend is consistent with the past change trend.

[0168] In the present application, the eigenvalues ​​corresponding to the Renyi entropy, long-range correlation characteristics and irrelevant coefficient can be calculated using existing technologies, which will not be elaborated here. For example, the eigenvalues ​​of the irrelevant coefficient can be calculated using a time series correlation analysis algorithm, but the present invention is not limited to this.

[0169] Corresponding to the above method, the present invention also provides a system for determining the railway roadbed deformation warning level, the system comprising a computer device, the computer device comprising a processor and a memory, the memory storing a computer program / instructions, the processor being used to execute the computer program / instructions stored in the memory, and when the computer program / instructions are executed by the processor, the system implements the steps of the method described above.

[0170] The embodiment of the present invention also provides a computer-readable storage medium on which a computer program / instruction is stored, and when the computer program / instruction is executed by a processor, the steps of the aforementioned edge computing server deployment method are implemented. The computer-readable storage medium can be a tangible storage medium, such as a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a floppy disk, a hard disk, a removable storage disk, a CD-ROM, or any other form of storage medium known in the technical field.

[0171] It should be understood by those skilled in the art that the exemplary components, systems and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software or a combination of the two. Whether it is performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention. When implemented in hardware, it can be, for example, an electronic circuit, an application-specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present invention are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link via a data signal carried in a carrier.

[0172] It should be clear that the present invention is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present invention.

[0173] In the present invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with features of other embodiments or replace features of other embodiments.

[0174] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the embodiments of the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for determining a railway roadbed deformation warning level, characterized in that: The method comprises the following steps: Acquire deformation monitoring data of the railway roadbed, use a curve trend prediction model to fit the deformation monitoring data to obtain a continuous deformation trend feature, and discretize the continuous deformation trend feature to obtain a plurality of deformation trend intervals; wherein each deformation trend interval includes a plurality of data points for indicating a corresponding relationship between a roadbed settlement value and time; Based on the roadbed settlement value of the data point in the deformation trend interval, the characteristic value of the multi-dimensional deformation statistical index of each deformation trend interval is obtained. For each deformation statistical index, the characteristic value of the deformation statistical index of each deformation trend interval is compared to obtain the deformation characteristic value of the corresponding deformation statistical index. Calculate the correlation coefficient between the deformation characteristic value of each deformation statistical indicator and the deformation center characteristic value corresponding to each deformation statistical indicator to obtain the correlation coefficient corresponding to each deformation statistical indicator, integrate the correlation coefficients corresponding to each deformation statistical indicator to obtain the total correlation coefficient, and determine the deformation warning level of the railway roadbed based on the size of the total correlation coefficient; wherein the deformation center characteristic value corresponding to each deformation statistical indicator is obtained by clustering the deformation characteristic values ​​of the corresponding deformation statistical indicators calculated based on multiple groups of deformation monitoring data obtained in advance; The correlation coefficients corresponding to the deformation statistical indicators are used to indicate the degree of membership of the deformation characteristic value to the deformation center characteristic value for each deformation statistical indicator; For each deformation statistical indicator, the characteristic values ​​of the deformation statistical indicators in each deformation trend interval are compared to obtain the deformation characteristic value of the corresponding deformation statistical indicator, including: For each deformation statistical indicator, the range of the characteristic value of the deformation statistical indicator in each deformation trend interval is calculated, and the range of the characteristic value is used as the deformation characteristic value of the deformation statistical indicator.

2. The method according to claim 1, characterized in that The obtaining of the deformation monitoring data of the railway roadbed comprises: obtaining the original deformation monitoring data of the railway roadbed, performing data cleaning on the original deformation monitoring data to obtain initial deformation monitoring data, and performing noise reduction processing on the initial deformation monitoring data to obtain second noise reduction deformation monitoring data; The process of performing noise reduction processing on the initial deformation monitoring data to obtain second noise-reduced deformation monitoring data includes: Using a wavelet threshold denoising algorithm to denoise the initial deformation monitoring data to obtain first denoised deformation monitoring data; Using an empirical mode decomposition algorithm to divide the first noise reduction deformation monitoring data into a plurality of intrinsic mode components and residual components, and using an autocorrelation function to calculate the autocorrelation coefficient of each intrinsic mode component; The wavelet threshold denoising algorithm is used to denoise the intrinsic mode components whose autocorrelation coefficients are higher than the set threshold, and the denoised intrinsic mode components, the intrinsic mode components whose autocorrelation coefficients are not higher than the set threshold, and the residual components are reconstructed to obtain second denoised deformation monitoring data.

3. The method according to claim 2, characterized in that The method also includes: using the second noise reduction deformation monitoring data to form a corresponding frequency spectrum diagram and power spectrum density diagram, and evaluating the frequency spectrum diagram and power spectrum density diagram according to a set noise evaluation index to evaluate the effect of the noise reduction processing; wherein the set noise evaluation index includes one or more evaluation indicators of signal-to-noise ratio, waveform similarity parameter, mean square error and root mean square error.

4. The method according to claim 1, characterized in that: The deformation trend interval is obtained by: Based on the same monitoring time interval, the continuous characteristics of deformation trend are discretized into multiple deformation trend intervals; Alternatively, based on the same number of data points, the deformation trend continuous feature is discretized into multiple deformation trend intervals.

5. The method according to claim 1, characterized in that The curve trend prediction model includes a hyperbola model, a Hoshino method model, an exponential curve model or a Gopperts model; The deformation statistical indicators are mean, variance, range, slope, Rayleigh entropy, generalized Hurst index, long-range correlation characteristics or uncorrelation coefficient; wherein the slope is obtained by fitting the roadbed settlement values ​​of data points in the deformation trend interval using the least squares method.

6. The method according to claim 1, characterized in that The larger the total correlation coefficient is, the lower the deformation warning level is; The deformation center eigenvalues ​​include eigenvalues ​​of intangible deformation cluster centers and eigenvalues ​​of tangible deformation cluster centers.

7. The method according to claim 6, characterized in that The degree of membership is obtained in the following way: The membership function is used to calculate the degree of membership of the deformation eigenvalue of each deformation statistical index to the eigenvalue of the non-deformation clustering center corresponding to the corresponding deformation statistical index.

8. A system for determining a railway roadbed deformation warning level, comprising a processor, a memory, and a computer program / instruction stored in the memory, characterized in that: The processor is used to execute the computer program / instructions, and when the computer program / instructions are executed, the system implements the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method as claimed in any one of claims 1 to 7 are implemented.

Images