Railway roadbed deformation monitoring data interval prediction method and device

Abstract

The invention provides a railroad bed deformation monitoring data interval prediction method and device, and relates to the technical field of electric digital data processing, and the method comprises the steps: inputting a target deformation monitoring data sample of a railroad bed in a current time period into a deformation value prediction model, and enabling the deformation value prediction model to output a deformation prediction value of the target deformation monitoring data sample in a next time period; according to the deformation prediction value, a model variance corresponding to the deformation value prediction model and a noise variance, obtaining deformation interval prediction result data of the railway roadbed in the next time period; the model variance and the noise variance are obtained through training in advance based on historical deformation monitoring data samples of the railroad bed in each historical time period and a deformation monitoring data interval prediction model containing a plurality of deformation value prediction models. According to the method, deformation interval prediction of the railway roadbed can be realized, and the automation degree, reliability and effectiveness of railway roadbed deformation monitoring data interval prediction can be effectively improved, so that the timeliness of railway roadbed deformation monitoring is improved.

Description

Technical Field

[0001] This application relates to the technical field of electrical digital data processing, and particularly to a method and device for interval prediction of railway subgrade deformation monitoring data. Background Art

[0002] During the service process of railway track infrastructure, due to the combined effects of various disasters such as train impact, material aging, environmental erosion, and fatigue effect, problems such as damage accumulation and bearing capacity degradation will inevitably occur. Once a large deformation occurs, serious safety accidents may be triggered. By using the Beidou deformation monitoring system to continuously and stably monitor the track infrastructure automatically, the health status and performance degradation trend of the track infrastructure structure can be effectively understood. By predicting the deformation trend of the track infrastructure, potential safety hazards can be detected in time, early warning and targeted countermeasures can be taken in advance, and the safe operation of trains can be effectively ensured. Since the monitoring equipment is deployed in complex geological areas along the railway line, the monitoring data will show typical non-Gaussian characteristics such as spikes, heavy tails, skewness, and bimodality. At the same time, due to the continuous stability of the Beidou monitoring equipment, the monitoring data will be accompanied by long-term trends and memory characteristics. However, because the traditional analysis model has a relatively simple form and does not have a unit structure for learning such typical state characteristics, it cannot learn or even will damage such long-term trends and memory characteristics of the monitoring data during the modeling process. In addition, due to external uncertainties and other factors, single deformation value prediction often cannot fully reflect the real situation. Therefore, it is very urgent to construct an interval prediction model for deformation monitoring data suitable for the characteristics of railway subgrade deformation monitoring data to meet the interval prediction requirements of railway subgrade deformation trends. Summary of the Invention

[0003] In view of this, the embodiments of this application provide a method and device for interval prediction of railway subgrade deformation monitoring data to eliminate or improve one or more defects existing in the prior art.

[0004] One aspect of this application provides a method for interval prediction of railway subgrade deformation monitoring data, including: Inputting the target deformation monitoring data sample of the railway subgrade in the current period into the deformation value prediction model, so that the deformation value prediction model outputs the deformation prediction value of the railway subgrade in the next period; According to the deformation prediction value of the railway subgrade in the next period, the model variance corresponding to the deformation value prediction model, and the noise variance, obtaining the deformation interval prediction result data of the railway subgrade in the next period, where the model variance and the noise variance are pre-trained based on the historical deformation monitoring data samples of the railway subgrade in each historical period and a deformation monitoring data interval prediction model including multiple deformation value prediction models.

[0005] In some embodiments of the present application, before inputting the target deformation monitoring data sample of the railway subgrade in the current period into the deformation value prediction model to enable the deformation value prediction model to output the deformation prediction value of the railway subgrade in the next period, the following steps are further included: Data resampling is performed on the historical deformation monitoring data samples of the railway subgrade in each historical period in the original dataset to obtain a plurality of sample datasets. Among them, each of the historical deformation monitoring data samples is provided with the true value of the railway subgrade deformation as a label, and each of the sample datasets contains the reconstructed deformation monitoring data samples corresponding to each of the historical deformation monitoring data samples; Based on the respective deformation value prediction models corresponding to each of the sample datasets one by one, the deformation prediction values corresponding to each of the reconstructed deformation monitoring data samples in each of the sample datasets are respectively obtained; and according to the preset significance level parameter value and the deformation prediction values corresponding to each of the reconstructed deformation monitoring data samples in each of the sample datasets, the model variance corresponding to the deformation value prediction model is obtained; Moreover, a deformation value prediction model is used to respectively obtain the deformation prediction values corresponding to each of the historical deformation monitoring data samples, and a target dataset is constructed according to the deformation prediction values corresponding to each of the historical deformation monitoring data samples and the true value of the railway subgrade deformation. Then, the target dataset is input into the deformation value prediction model, and the output result data of the deformation value prediction model is used as the noise variance corresponding to the deformation value prediction model; According to the significance level parameter value, the model variance, the noise variance, and the preset deformation prediction parameters, a deformation interval prediction formula is constructed.

[0006] In some embodiments of the present application, the step of obtaining the deformation interval prediction result data of the railway subgrade in the next period according to the deformation prediction value of the railway subgrade in the next period, the model variance and the noise variance corresponding to the deformation value prediction model includes: Substitute the deformation prediction value of the railway subgrade in the next period into the deformation prediction parameter in the deformation interval prediction formula and solve the deformation interval prediction formula to obtain the deformation interval prediction result data of the railway subgrade in the next period.

[0007] In some embodiments of the present application, the step of performing data resampling on the historical deformation monitoring data samples of the railway subgrade in each historical period in the original dataset to obtain a plurality of sample datasets includes: Based on a preset step size, data resampling is performed on the historical deformation monitoring data samples of the railway subgrade in each historical period in the original dataset in the Bootstrap resampling manner, so as to obtain each sample dataset that all contains different reconstructed deformation monitoring data samples corresponding to their respective historical deformation monitoring data samples.

[0008] In some embodiments of the present application, before performing data resampling on the historical deformation monitoring data samples of the railway subgrade in each historical period in the original dataset to obtain a plurality of sample datasets, it further includes: Obtain the longitude and latitude data of the railway subgrade monitored by the Beidou satellite in each historical period; Perform position calculation and data frequency adjustment on the longitude and latitude data of each historical period to obtain the deformation monitoring data of the railway subgrade in each historical period; Preprocess the deformation monitoring data of each historical period to obtain the historical deformation monitoring data corresponding to each historical period, wherein the preprocessing includes at least one of outlier and missing value processing, wavelet denoising, and stationarity test; Perform fractional-order difference operation on each of the historical deformation monitoring data to obtain the historical deformation monitoring data samples of each historical period and form a corresponding original dataset.

[0009] In some embodiments of the present application, the deformation value prediction model includes: an LSTM network provided with a fractional-order difference processing unit, and the fractional-order difference processing unit is used to perform fractional-order difference operation on the obtained prediction result after the LSTM network performs deformation prediction on the input deformation monitoring data sample, so as to obtain the deformation prediction value corresponding to each of the deformation monitoring data samples, wherein the deformation monitoring data sample includes: the reconstructed deformation monitoring data sample or the historical deformation monitoring data sample.

[0010] In some embodiments of the present application, the railway subgrade deformation monitoring data interval prediction method further includes: When or after obtaining the deformation monitoring data of the railway subgrade in the next period, generate the potential safety hazard detection result data of the railway subgrade in the next period according to the deformation monitoring data of the next period and the deformation interval prediction result data of the railway subgrade in the next period.

[0011] Another aspect of the present application provides a railway subgrade deformation monitoring data interval prediction device, including: A deformation value prediction module, configured to input the target deformation monitoring data sample of the railway subgrade in the current period into the deformation value prediction model, so that the deformation value prediction model outputs the deformation prediction value of the railway subgrade in the next period; A deformation interval prediction module, configured to obtain the deformation interval prediction result data of the railway subgrade in the next time period according to the deformation prediction value of the railway subgrade in the next time period, the model variance corresponding to the deformation value prediction model, and the noise variance, where the model variance and the noise variance are pre-trained based on the historical deformation monitoring data samples of the railway subgrade in each historical time period and a deformation monitoring data interval prediction model including a plurality of deformation value prediction models.

[0012] The third aspect of the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the railway subgrade deformation monitoring data interval prediction method described above is implemented.

[0013] The fourth aspect of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the railway subgrade deformation monitoring data interval prediction method described above is implemented.

[0014] The fifth aspect of the present application provides a computer program product, including a computer program. When the computer program is executed by a processor, the railway subgrade deformation monitoring data interval prediction method described above is implemented.

[0015] The railway subgrade deformation monitoring data interval prediction method provided by the present application inputs the target deformation monitoring data sample of the railway subgrade in the current time period into the deformation value prediction model, so that the deformation value prediction model outputs the deformation prediction value of the railway subgrade in the next time period; according to the deformation prediction value of the railway subgrade in the next time period, the model variance corresponding to the deformation value prediction model, and the noise variance, the deformation interval prediction result data of the railway subgrade in the next time period is obtained, where the model variance and the noise variance are pre-trained based on the historical deformation monitoring data samples of the railway subgrade in each historical time period and a deformation monitoring data interval prediction model including a plurality of deformation value prediction models, which can realize the deformation interval prediction of the railway subgrade, effectively improve the automation degree and reliability of the railway subgrade deformation monitoring data interval prediction process, improve the effectiveness of the deformation interval prediction result, improve the timeliness of the railway subgrade deformation monitoring, and further provide more auxiliary information for relevant departments such as the operation and maintenance of railway track infrastructure, effectively realizing the prevention of potential safety hazards of track infrastructure.

[0016] The additional advantages, objectives, and features of the present application will be partially described below, and will become partially apparent to those of ordinary skill in the art after studying the following part, or can be learned from the practice of the present application. The objectives and other advantages of the present application can be achieved and obtained by the structures specifically pointed out in the specification and the drawings.

[0017] Those skilled in the art will understand that the objectives and advantages that can be achieved with this application are not limited to those specifically described above, and the above and other objectives that can be achieved with this application will be more clearly understood according to the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of this application, form a part of this application, and do not limit this application. The components in the drawings are not drawn to scale, but only to illustrate the principles of this application. To facilitate the illustration and description of some parts of this application, the corresponding parts in the drawings may be enlarged, that is, they may become larger relative to other components in the exemplary device actually manufactured according to this application. In the drawings: Figure 1 FIG. 8 is a first schematic flowchart of a method for interval prediction of railway subgrade deformation monitoring data in an embodiment of this application.

[0019] Figure 2 FIG. 12 is a second schematic flowchart of a method for interval prediction of railway subgrade deformation monitoring data in an embodiment of this application.

[0020] Figure 3 FIG. 16 is a schematic training logic diagram of the model variance and noise variance in the deformation interval prediction formula of the method for interval prediction of railway subgrade deformation monitoring data in an embodiment of this application.

[0021] Figure 4 FIG. 20 is a third schematic flowchart of a method for interval prediction of railway subgrade deformation monitoring data in an embodiment of this application.

[0022] Figure 5 FIG. 24 is a schematic flowchart of a method for interval prediction of railway subgrade deformation monitoring data respectively executed based on the FI-LSTM model and the LSTM model in an application example of this application.

[0023] Figure 6 FIG. 28 is a schematic diagram of the relationship between the prediction interval estimate with a confidence level of 95% and the predicted value (i.e., the deformation predicted value) in an application example of this application.

[0024] Figure 7 FIG. 32 is a schematic diagram of the relationship between the prediction interval estimate with a confidence level of 90% and the predicted value (i.e., the deformation predicted value) in an application example of this application.

[0025] Figure 8 FIG. 36 is a schematic structural diagram of a device for interval prediction of railway subgrade deformation monitoring data in an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] To make the objectives, technical solutions and advantages of the present application more clear and understandable, the following further details the present application in conjunction with the embodiments and the accompanying drawings. Herein, the illustrative embodiments of the present application and their descriptions are used to explain the present application, but do not limit the present application.

[0027] Herein, it should also be noted that in order to avoid obscuring the present application with unnecessary details, only the structures and / or processing steps closely related to the solution according to the present application are shown in the drawings, while other details less relevant to the present application are omitted.

[0028] It should be emphasized that the term "including / comprising" 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.

[0029] Herein, it should also be noted that if not otherwise specified, the term "connection" in this document can refer not only to a direct connection, but also to an indirect connection with an intermediate.

[0030] Hereinafter, embodiments of the present application will be described with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0031] Accurately predicting the future change trend of variables is one of the key elements in making decisions. However, due to factors such as external uncertainties, a single point prediction method often cannot comprehensively reflect the real situation. To further process and understand this uncertainty, the prediction interval theory has emerged. The prediction interval theory can provide a more comprehensive theoretical framework for variable trend prediction by introducing concepts such as prediction intervals based on statistical theory, so as to more accurately describe the future change range of variables.

[0032] Based on this, in order to solve problems such as the inability of existing methods for changing eye features of people in videos to automatically predict the deformation interval of railway subgrades, the embodiments of the present application respectively provide a method for predicting the interval of railway subgrade deformation monitoring data, a device for predicting the interval of railway subgrade deformation monitoring data for executing the method for predicting the interval of railway subgrade deformation monitoring data, an entity device, a computer-readable storage medium, a storage medium and a computer program product, which can realize the prediction of the deformation interval of railway subgrades, can effectively improve the automation degree and reliability in the process of predicting the interval of railway subgrade deformation monitoring data, and can improve the effectiveness of the prediction result of the deformation interval, so as to improve the timeliness of railway subgrade deformation monitoring.

[0033] Specific details are described in detail through the following embodiments.

[0034] Based on this, an embodiment of the present application provides a method for predicting the interval of railway subgrade deformation monitoring data that can be implemented by a device for predicting the interval of railway subgrade deformation monitoring data. Refer to Figure 1 The method for predicting the interval of railway subgrade deformation monitoring data specifically includes the following content: Step 100: Input the target deformation monitoring data sample of the railway subgrade in the current period into the deformation value prediction model, so that the deformation value prediction model outputs the deformation prediction value of the railway subgrade in the next period.

[0035] The deformation value prediction model in Step 100 can be implemented by an existing machine learning model that can predict the deformation prediction value of the railway subgrade in the next period based on the target deformation monitoring data sample of the railway subgrade in the current period. For example, a time series analysis model such as an LSTM network can be used. Among them, the LSTM network refers to the Long Short-Term Memory model.

[0036] In one or more embodiments of the present application, each type of period can be set to the same duration, that is, the current period, the next period, and the historical period, etc. can all be the same duration.

[0037] It can be understood that the target deformation monitoring data sample can be obtained by processing the longitude and latitude data of the railway subgrade monitored by the Beidou satellite in the current period. Specifically, it can include: obtaining the longitude and latitude data of the railway subgrade monitored by the Beidou satellite in the current period; performing position calculation and data frequency adjustment on the longitude and latitude data to obtain the deformation monitoring data of the railway subgrade in the current period; preprocessing the deformation monitoring data to obtain the corresponding target deformation monitoring data, where the preprocessing includes at least one of outlier and missing value processing, wavelet denoising, and stationarity test; if the deformation value prediction model in Step 100 uses a model such as an LSTM network equipped with a fractional-order difference processing unit, the preprocessed target deformation monitoring data can also be subjected to fractional-order difference processing on the target deformation monitoring data to eliminate the long-range correlation characteristics of the data and retain the trend characteristics of the data, so as to obtain the target deformation monitoring data sample of the current period that conforms to the LSTM modeling process; if the deformation value prediction model in Step 100 does not have a fractional-order difference processing unit, the target deformation monitoring data is the target deformation monitoring data sample.

[0038] In one or more embodiments of the present application, the LSTM network equipped with a fractional-order difference processing unit can be abbreviated as the FI-LSTM model, which is a recursive recurrent neural network model with self-similar characteristics and can flexibly simulate the random process of signals with typical long-term trend characteristics.

[0039] Step 200: Obtain the predicted deformation interval result data of the railway subgrade in the next time period according to the predicted deformation value of the railway subgrade in the next time period, the model variance corresponding to the deformation value prediction model, and the noise variance, where the model variance and the noise variance are pre-trained based on the historical deformation monitoring data samples of the railway subgrade in each historical time period and the deformation monitoring data interval prediction model including multiple deformation value prediction models.

[0040] The model variance and the noise variance corresponding to the deformation value prediction model adopted in Step 200 are pre-trained based on the historical deformation monitoring data samples of the railway subgrade in each historical time period and the deformation monitoring data interval prediction model including multiple deformation value prediction models. Among them, the model architectures of the multiple deformation value prediction models in the deformation monitoring data interval prediction model are the same. For example, they can all be LSTM networks equipped with fractional-order difference processing units; the deformation value prediction model adopted in Step 100 can be one of the multiple deformation value prediction models in the deformation monitoring data interval prediction model, or other deformation value prediction models that do not belong to the deformation monitoring data interval prediction model.

[0041] Specifically, first, it is necessary to explain the predicted deformation interval result data mentioned in Step 200. The predicted deformation interval result data is embodied as a numerical interval, which can be abbreviated as the prediction interval in one or more embodiments of the present application. The prediction interval has wide application value in many fields. First of all, it can help decision-makers evaluate the uncertainty of the prediction results, so as to avoid risk decisions made by over-relying on a single predicted value. Secondly, the prediction interval provides a tool for scientific researchers and data analysts to quantify the prediction accuracy, which helps them optimize the model and improve the prediction accuracy.

[0042] As can be seen from the above description, the railway subgrade deformation monitoring data interval prediction method provided by the embodiments of the present application can realize the deformation interval prediction of the railway subgrade, effectively improve the automation degree and reliability of the railway subgrade deformation monitoring data interval prediction process, improve the effectiveness of the predicted deformation interval result, so as to improve the timeliness of railway subgrade deformation monitoring, and further provide more auxiliary information for relevant departments such as the operation and maintenance of railway track infrastructure, and effectively realize the prevention of potential safety hazards of track infrastructure.

[0043] In the development and application process of the prediction interval theory, it is necessary to clarify the concepts of the confidence interval and the prediction interval. The confidence interval is the interval for parameter estimation, mainly for estimating the uncertainty factors of the population parameters, while the prediction interval is used to describe the uncertainty of the model prediction, considering the random error of the model, and is the estimation of the future change interval of the variable. This application mainly estimates the future change interval of the trend of deformation monitoring data, rather than estimating a certain population parameter. Therefore, the prediction interval theory of the variable is mainly described.

[0044] Suppose the output value of a certain prediction model is , then the error of the prediction model can be expressed as: (1) In formula (1), represents the i-th observed value (i.e., the monitoring data). In the prediction method provided in the embodiments of this application, the output value of the prediction model is the predicted regression value of B FI-LSTM models, represents the true value. In the prediction method provided in the embodiments of this application, is the true value of the deformation of the railway subgrade as the label, represents the random error caused by noise. It is easy to observe that the error between the observed value and the true value can be represented by , that is, the prediction error; the error between the predicted value and the true value of the prediction model can be represented by , that is, the model error.

[0045] Among them, the estimated value of the distribution corresponding to the prediction error is the estimated value of the confidence interval of the model prediction result; and the estimated value of the distribution corresponding to the model error is the estimated value of the prediction interval of the model. If the model error and the random error are independent of each other, the model prediction variance (i.e., the variance of the prediction result of the prediction model) is expressed as: (2) In formula (2), represents the estimated value of the variance, that is, the model variance (i.e., the variance of the model uncertainty), while represents the estimated value corresponding to the noise and the interference signal, that is, the noise variance (i.e., the variance of the noise uncertainty). Therefore, the solution of the prediction interval of the prediction value of the prediction model can be divided into two parts: solving the variance of the model uncertainty and the variance of the noise uncertainty. The prediction interval can be directly estimated through . And the prediction interval theory fully considers factors such as noise and interference existing in the prediction result. Therefore, in the evaluation of the true prediction performance of the prediction model, constructing an interval prediction model will have more important practical significance.

[0046] Under the given significance level , the prediction interval corresponding to the prediction model can be expressed as: (3) In formula (3), represents the significance level, and respectively correspond to the upper and lower bounds of the prediction interval of the i-th predicted value of the prediction model. The upper and lower bounds can be specifically expressed as: (4) (5) In formulas (4) and (5), represents the upper quantile of the Gaussian distribution under the condition of the significance level , while represents the model prediction variance, which includes the model variance and the noise variance in two parts.

[0047] Based on this, in order to further improve the automation degree and reliability of the interval prediction process of railway subgrade deformation monitoring data, and improve the effectiveness of the deformation interval prediction result and the timeliness of railway subgrade deformation monitoring, in a method for interval prediction of railway subgrade deformation monitoring data provided in an embodiment of the present application, by using a prediction model and combining a data reconstruction method, an estimation of the overall distribution of variables can be realized. Among them, an LSTM network with a fractional-order difference processing unit is used as the deformation value prediction model. According to the above prediction interval theory, if the prediction interval of a variable is to be obtained, then first, the predicted value and the variance need to be obtained. And the model prediction variance includes two parts, namely the model variance and the noise variance, that is: (6) In formula (6), i represents the i-th moment; n represents the n-th moment; represents the i-th true value; represents the i-th predicted value; represents the average value.

[0048] Therefore, the key to solving the model prediction interval of a variable lies in solving the two parts of the model variance and the noise variance. The overall idea is to resample the original data using the data resampling method times, and use the FI-LSTM model for prediction to achieve the estimation of the overall distribution. The model variance can be obtained through further calculation. The method for solving the noise variance takes the original data as input, and the output uses the variance of the prediction results of the prediction model and the model variance to estimate the noise variance, and then the estimated noise variance can be obtained through the FI-LSTM model and further calculation.

[0049] Specifically, referring to Figure 2 , before step 100 in the interval prediction method for railway subgrade deformation monitoring data, the following specific content is further included: Step 010: Resample the historical deformation monitoring data samples of the railway subgrade in each historical period in the original dataset to obtain multiple sample datasets. Among them, each historical deformation monitoring data sample is provided with the true value of the railway subgrade deformation as a label, and each sample dataset contains the reconstructed deformation monitoring data samples corresponding to each of the historical deformation monitoring data samples.

[0050] Step 020: Based on each deformation value prediction model corresponding to each sample dataset respectively, obtain the deformation prediction values corresponding to each of the reconstructed deformation monitoring data samples in each sample dataset; and according to the preset significance level parameter value and the deformation prediction values corresponding to each of the reconstructed deformation monitoring data samples in each sample dataset, obtain the model variance corresponding to the deformation value prediction model.

[0051] In one or more embodiments of the present application, each historical deformation monitoring data sample corresponding to each historical period corresponds to a reconstructed deformation monitoring data sample in each of the 1st to Bth sample datasets, and the deformation prediction value corresponding to the ith reconstructed deformation monitoring data sample in the bth sample dataset is denoted as , where B is a positive integer greater than 2, preferably a positive integer greater than 1000, b is any positive integer from 1 to B; i is a positive integer greater than 1. The significance level parameter value can be abbreviated as the significance level, denoted as .

[0052] Referring to Figure 3 , the deformation prediction values corresponding to each of the reconstructed deformation monitoring data samples in sample dataset 1 constitute the deformation prediction value set corresponding to sample dataset 1 ; the FI-LSTM model corresponding to sample dataset 1 is denoted as FI-LSTM1; the deformation prediction values corresponding to each of the reconstructed deformation monitoring data samples in sample dataset 2 constitute the deformation prediction value set corresponding to sample dataset 2 ; The FI-LSTM model corresponding one-to-one to the sample data set 2 is denoted as FI-LSTM2; the deformation prediction values corresponding to each of the reconstructed deformation monitoring data samples in the B-th sample data set (i.e., the sample data set B) constitute the deformation prediction value set corresponding to the sample data set B ; The FI-LSTM model corresponding one-to-one to the sample data set B is denoted as FI-LSTMB.

[0053] In step 200, according to the preset significance level parameter value and the deformation prediction values corresponding to each of the reconstructed deformation monitoring data samples in each of the sample data sets, the specific process of obtaining the model variance corresponding to the deformation value prediction model is as follows: 1) Based on each of the deformation value prediction models (i.e., FI-LSTM1 to FI-LSTMB) corresponding one-to-one to each of the sample data sets, respectively obtain the deformation prediction values corresponding to each of the reconstructed deformation monitoring data samples in each of the sample data sets , to obtain the prediction regression values of the B FI-LSTM models : (7) 2) According to the deformation prediction values corresponding to each of the reconstructed deformation monitoring data samples in each of the sample data sets and the prediction regression values of the B FI-LSTM models obtain the model variance corresponding to the deformation value prediction model : (8) And, step 030: Use a deformation value prediction model to respectively obtain the deformation prediction values corresponding to each of the historical deformation monitoring data samples, and construct a target data set according to the deformation prediction values corresponding to each of the historical deformation monitoring data samples and the true value of the railway subgrade deformation, and then input the target data set into the deformation value prediction model, and use the output result data of the deformation value prediction model as the noise variance corresponding to the deformation value prediction model.

[0054] Specifically, step 030 can use any one of the LSTM networks (i.e., FI-LSTM models) with a fractional-order difference processing unit in step 020 to construct a target data set H according to the deformation prediction values corresponding to each of the historical deformation monitoring data samples and the true value of the railway subgrade deformation: (9) In formula (9), the value of the historical deformation monitoring data sample input at time i, y iDenote the true value of the railway subgrade deformation as the label of the historical deformation monitoring data sample, which can be simply referred to as the true value.

[0055] Specifically, interval prediction of a variable involves solving the model variance and the noise variance. First, the process of solving the noise method will be described. Since the noise method cannot be directly observed, the noise variance can be estimated based on the variance of the prediction result of the prediction model and the model variance. According to , the noise variance can be expressed as , that is: (10) Since the noise variance should be non - negative, so there can be , and then the target data set can also be abbreviated as , Denote the noise variance sent into the prediction model at time i. Resample this data set and repeat times, and training the deformation value prediction model can obtain an approximate noise variance, and then the prediction interval corresponding to the deformation value prediction model can be obtained.

[0056] Step 040: Construct a deformation interval prediction formula based on the significance level parameter value, the model variance, the noise variance, and a preset deformation prediction parameter.

[0057] Specifically, under the condition of a given significance level , a deformation interval prediction formula as shown in the following formula (11) can be constructed: (11) In formula (11), is the deformation interval prediction result data of the railway subgrade in the next time period, that is, the prediction interval.

[0058] Based on this, referring to Figure 2 , in a method for interval prediction of railway subgrade deformation monitoring data provided in an embodiment of the present application, the step 200 specifically includes the following contents: Step 210: Substitute the deformation prediction value of the railway subgrade in the next time period into the deformation prediction parameter in the deformation interval prediction formula and solve the deformation interval prediction formula to obtain the deformation interval prediction result data of the railway subgrade in the next time period.

[0059] In order to further improve the effectiveness and reliability of resampling, in a method for interval prediction of railway subgrade deformation monitoring data provided in an embodiment of the present application, referring to Figure 4 , the step 010 in the method for interval prediction of railway subgrade deformation monitoring data specifically includes the following contents: Step 011: Based on a preset step size, resample the historical deformation monitoring data samples of the railway subgrade in each historical period in the original dataset by the Bootstrap resampling method to obtain sample datasets, each of which contains different reconstructed deformation monitoring data samples corresponding to their respective historical deformation monitoring data samples.

[0060] Among them, the Bootstrap resampling method is a statistical method that randomly draws samples from the original dataset with replacement (that is, each sample has an equal probability of being selected and can be selected multiple times) to generate multiple new datasets (called bootstrap samples). These new datasets are used to estimate the distribution of statistics, so as to infer the statistical characteristics of the original dataset.

[0061] That is to say, the Bootstrap method can be used to resample the original dataset B times to obtain B sample datasets. Different FI-LSTM models are used to predict each sample dataset to obtain B predicted values. When B is large enough (B should be greater than 1000), the model variance can be estimated.

[0062] On this basis, in order to further improve the accuracy and result validity of the interval prediction of railway subgrade deformation monitoring data, in a method for interval prediction of railway subgrade deformation monitoring data provided in an embodiment of the present application, see Figure 4 , before step 010 in the method for interval prediction of railway subgrade deformation monitoring data, the following specific content is further included: Step 001: Obtain the longitude and latitude data of the railway subgrade monitored by the Beidou satellite in each historical period.

[0063] Step 002: Perform position calculation and data frequency adjustment on the longitude and latitude data of each historical period to obtain the deformation monitoring data of the railway subgrade in each historical period.

[0064] Step 003: Preprocess the deformation monitoring data of each historical period to obtain the historical deformation monitoring data corresponding to each historical period. Among them, the preprocessing includes at least one of outlier and missing value processing, wavelet denoising, and stationarity test.

[0065] Step 004: Perform fractional-order difference (i.e., d-order nabla derivative) operation on each of the historical deformation monitoring data to obtain the historical deformation monitoring data samples of each historical period and form a corresponding original dataset.

[0066] Railway Beidou monitoring equipment is usually deployed at positions such as railway subgrades for round-the-clock monitoring of the deformation of railway infrastructure. Due to its long-term, stable and continuous monitoring, the output signals of Beidou monitoring equipment usually exhibit long-range correlation characteristics, long-term trend characteristics and typical periodic characteristics. It can reflect certain dynamic characteristics in the evolution process of monitoring data and significantly affect the future change trend of monitoring data. Therefore, it is necessary to introduce this feature into the prediction model when constructing the prediction model to improve the model accuracy. The FI-LSTM model proposed in this application is a recurrent neural network model with self-similar characteristics, which can flexibly simulate the random process of signals with typical long-term trend characteristics.

[0067] Based on this, in a method for interval prediction of railway subgrade deformation monitoring data provided in an embodiment of this application, the deformation value prediction model includes: an LSTM network provided with a fractional difference processing unit, and this fractional difference processing unit is used to perform fractional difference (i.e., d-order inverse nabla derivative) operation on the obtained prediction result after the LSTM network performs deformation prediction on the input deformation monitoring data sample, so as to obtain the deformation prediction value corresponding to each of the deformation monitoring data samples, where the deformation monitoring data sample includes: the reconstructed deformation monitoring data sample or the historical deformation monitoring data sample, and the value of d is determined in advance based on the Hurst exponent of the preset deformation monitoring data sample.

[0068] Among them, the fractional difference processing unit refers to a functional module for performing fractional difference processing and is connected to the output layer of the LSTM network.

[0069] To further illustrate the above FI-LSTM model, in an example, the structure of the FI-LSTM model is as follows: (12) In formula (12), represents the t-th signal output by the FI-LSTM model, represents the (t - 1)-th signal output by the FI-LSTM model, can be 0, 1, 2…N; represents the input signal subject to independent and identically distributed; represents the traditional LSTM model; represents order nabla derivative, and the order can be obtained through the Hurst exponent Get: , where the Hurst exponent Hurst can be obtained by the rescaled range method, represents the feedback difference operator, defined as , so can be expressed as: (13) In formula (13), represents the sampling period; k represents the order of the factorial power; that is: (14) Define , can be 0, 1, 2... N; then the output of the FI-LSTM model can be expressed as: (15) represents the number of fractional difference terms at the t-th moment, can be 0, 1, 2... N; Without loss of generality, for , ; When , , ; When , ; When , ; When , ; When , ; That is: (16) represents the input value of the model at the corresponding t moment; Among them, , , (17) In addition, in order to further realize the prevention of potential safety hazards in railway infrastructure, in a method for predicting the interval of railway subgrade deformation monitoring data provided in the embodiment of the present application, referring to Figure 4 , after step 200 in the method for predicting the interval of railway subgrade deformation monitoring data, the following specific content is included: Step 300: When or after obtaining the deformation monitoring data of the railway subgrade in the next time period, generate the potential safety hazard detection result data of the railway subgrade in the next time period according to the deformation monitoring data of the next time period and the predicted result data of the deformation interval of the railway subgrade in the next time period.

[0070] Specifically, if the value corresponding to the deformation monitoring data in the next time period is within the value range corresponding to the predicted result data of the deformation interval of the railway subgrade in the next time period (i.e., the prediction interval), it is determined that the infrastructure state is stable and there are no potential safety hazards in the next time period. Otherwise, it is confirmed that there are potential safety hazards. Then, the safety hazard detection result data indicating whether there are potential safety hazards in the railway subgrade in the next time period is output.

[0071] To further illustrate the above embodiments and verify the embodiments of the method for predicting the deformation monitoring data interval of the railway subgrade implemented by the Bootstrap resampling method and the FI-LSTM model, see Figure 5 , taking the FI-LSTM model as the residual model and the traditional LSTM model as the comparison model, the present application also provides an application example of the method for predicting the deformation monitoring data interval of the railway subgrade respectively executed based on the FI-LSTM model and the LSTM model. Since the FI-LSTM model shows better results in time series prediction, the application example of the present application proposes a set of trend prediction methods for time series based on this model, specifically including the following three parts: (1) Data preprocessing: By performing position calculation, denoising, outlier cleaning, and stationarity test on the original data to obtain normalized data; (2) FI-LSTM model construction: By calculating the Hurst exponent to determine the nabla derivative order and adjusting parameters to determine the model structure; (3) Trend prediction: Using the FI-LSTM model for lead prediction, and realizing interval prediction by resampling the model input data and calculating the variance of the corresponding predicted values. The application example of the present application specifically includes the following content: Step 1: 1. Data preprocessing - Position calculation and data frequency adjustment: Perform position calculation on the original longitude and latitude data (abbreviated as original data), and obtain displacement deformation data through data frequency adjustment, etc.; Step 2: 1. Data preprocessing - Wavelet denoising: Perform operations such as outlier and missing value processing and wavelet denoising on the deformation data to obtain normalized data; Step 3: 1. Data preprocessing - Stationarity test: Use the stable distribution to test the stationarity of the data. If it is not stationary, perform data aggregation processing, perform zero-mean processing on the data, and obtain standardized data; Step 3 specifically includes: 1) Stationarity discrimination: The application example of the present application mainly uses the stable distribution to estimate the stationarity of the deformation monitoring data. Since the stable distribution does not have a closed probability density function, it can only be described by the characteristic function .

[0072] 2) Data standardization: Perform extreme value normalization on the data: (18) In the formula represents the data after normalization processing, represents the maximum value in the array, represents the minimum value in the array.

[0073] Step 4: FI-LSTM model construction - dataset reconstruction: Calculate the Hurst exponent of the sequence, and perform order nabla derivative processing to eliminate the long-range correlation characteristics of the data and retain the trend characteristics of the data, so that it conforms to the LSTM modeling process; Specifically, since the deformation monitoring data is one-dimensional displacement data, in order to carry out further prediction model construction, the data needs to be intercepted and combined to be reconstructed into a two-dimensional matrix Z.

[0074] Step 5: FI-LSTM model construction - Train the model parameters of the FI-LSTM model and the LSTM model respectively, perform point prediction on the deformation monitoring data, and perform order inverse nabla derivative operation to output the predicted values of the FI-LSTM model and the LSTM model respectively; Perform times of Bootstrap resampling on the training data, train the corresponding prediction model and residual model respectively, estimate the model variance and noise variance, and solve the corresponding prediction interval.

[0075] Specifically, in order to further analyze the influence degree of each parameter of the LSTM model on the model accuracy, the application example of this application selects the number of hidden layers, the number of nodes in the hidden layer, the learning rate, and the batch size to test the model parameters respectively, and uses the root mean squared error (Root Mean Squared Error, RMSE), mean absolute error (Mean Absolute Error, MAE), mean absolute percentage error (Mean Absolute Percentage Error, MAPE), and coefficient of determination (Rsquared, R²) as reference indicators for accuracy discrimination. The purpose is to obtain the optimal parameters for subsequent trend prediction analysis. Finally, the model parameters are determined by the grid method to be 4 hidden layers, 64 nodes in each layer, a learning rate of 0.01, and a batch of 32.

[0076] Since there will be certain errors in the deformation monitoring system due to reasons such as noise, it is necessary to perform prediction interval estimation on the prediction results to provide more auxiliary information. After aggregation and normalization, the data follows a Gaussian distribution. The Bootstrap resampling method is used to obtain the corresponding data sets respectively, train the corresponding models and calculate the variances of the corresponding models and the noise variance. By selecting different , and then the prediction intervals with 95% and 90% confidence levels are obtained. The fitting results are shown in Figure 6 and Figure 7 respectively. It can be seen that the prediction intervals can basically cover the predicted values. The prediction intervals under different confidence levels are roughly the same as the deformation trend of the railway infrastructure, which can indicate the accuracy of the proposed prediction interval estimation method. In addition, the confidence interval is wider in places where the deformation changes violently, indicating that the prediction credibility becomes lower, and the results also conform to the actual situation.

[0077] That is to say, the application example of this application uses the FI-LSTM model and combines the Bootstrap method to realize the interval prediction of the deformation trend of the railway subgrade. By setting the confidence level, the fluctuation interval of the monitoring data under different confidence levels is determined, providing more auxiliary information for relevant departments such as the operation and maintenance of railway infrastructure, and effectively realizing the prevention of potential safety hazards of railway infrastructure.

[0078] From a software perspective, this application also provides a device for predicting the interval of railway subgrade deformation monitoring data for executing all or part of the methods for predicting the interval of railway subgrade deformation monitoring data. Refer to Figure 8 , and the device for predicting the interval of railway subgrade deformation monitoring data specifically includes the following contents: The deformation value prediction module 10 is used to input the target deformation monitoring data sample of the railway subgrade in the current period into the deformation value prediction model, so that the deformation value prediction model outputs the deformation prediction value of the railway subgrade in the next period.

[0079] The deformation interval prediction module 20 is used to obtain the deformation interval prediction result data of the railway subgrade in the next period according to the deformation prediction value of the railway subgrade in the next period, the model variance and the noise variance corresponding to the deformation value prediction model, where the model variance and the noise variance are pre-trained based on the historical deformation monitoring data samples of the railway subgrade in each historical period and the deformation monitoring data interval prediction model including multiple deformation value prediction models.

[0080] The embodiment of the device for predicting the interval of railway subgrade deformation monitoring data provided by this application can specifically be used to execute the processing flow of the embodiment of the method for predicting the interval of railway subgrade deformation monitoring data in the above embodiment, and its functions will not be elaborated here. For details, reference can be made to the detailed description of the embodiment of the method for predicting the interval of railway subgrade deformation monitoring data above.

[0081] The part of the device for predicting the interval of railway subgrade deformation monitoring data to perform the interval prediction of railway subgrade deformation monitoring data can be executed in the server or completed in the client device. Specifically, it can be selected according to the processing capacity of the client device and the limitations of the user usage scenario, etc. This application does not make any limitations in this regard. If all operations are completed in the client device, the client device may further include a processor for specific processing of the interval prediction of railway subgrade deformation monitoring data.

[0082] The above-mentioned client device may have a communication module (i.e., a communication unit), which can be communicatively connected to a remote server to achieve data transmission with the server. The server may include a server on the side of the task scheduling center, and in other implementation scenarios, it may also include a server of an intermediate platform, such as a server of a third-party server platform having a communication link with the task scheduling center server. The server may include a single computer device, or a server cluster composed of multiple servers, or a server structure of a distributed device.

[0083] Any suitable network protocol can be used for communication between the above-mentioned server and the client device side, including network protocols not yet developed on the filing date of this application. The network protocol may, for example, include TCP / IP protocol, UDP / IP protocol, HTTP protocol, HTTPS protocol, etc. Of course, the network protocol may also include, for example, the RPC protocol (Remote Procedure Call Protocol) and REST protocol (Representational State Transfer) used on top of the above-mentioned protocols.

[0084] As can be seen from the above description, the device for predicting the interval of railway subgrade deformation monitoring data provided by the embodiments of this application can realize the interval prediction of railway subgrade deformation, effectively improve the automation degree and reliability of the process of predicting the interval of railway subgrade deformation monitoring data, improve the effectiveness of the interval prediction result of deformation, so as to improve the timeliness of railway subgrade deformation monitoring, and further can provide more auxiliary information for relevant departments such as the operation and maintenance of railway track infrastructure, and effectively realize the prevention of potential safety hazards of track infrastructure.

[0085] The embodiments of this application also provide an electronic device, which may include a processor, a memory, a receiver and a transmitter. The processor is used to execute the method for predicting the interval of railway subgrade deformation monitoring data mentioned in the above embodiments. Among them, the processor and the memory may be connected through a bus or other means. Taking the connection through the bus as an example. The receiver can be connected to the processor and the memory in a wired or wireless manner.

[0086] The processor can be a Central Processing Unit (CPU). The processor can also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., in the form of chips, or combinations of the above types of chips.

[0087] As a non-transitory computer-readable storage medium, the memory can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as the program instructions / modules corresponding to the railway subgrade deformation monitoring data interval prediction method in the embodiments of the present application. By running the non-transitory software programs, instructions, and modules stored in the memory, the processor executes various functional applications and data processing of the processor, that is, implements the railway subgrade deformation monitoring data interval prediction method in the above method embodiments.

[0088] The memory can include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created by the processor, etc. In addition, the memory can include high-speed random access memory, and can also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory optionally includes a memory remotely set relative to the processor, and these remote memories can be connected to the processor through a network. Examples of the above networks include, but are not limited to, the Internet, enterprise intranets, local area networks, mobile communication networks, and combinations thereof.

[0089] The one or more modules are stored in the memory and, when executed by the processor, execute the railway subgrade deformation monitoring data interval prediction method in the embodiments.

[0090] In some embodiments of the present application, the user equipment can include a processor, a memory, and a transceiver unit. The transceiver unit can include a receiver and a transmitter. The processor, the memory, the receiver, and the transmitter can be connected through a bus system. The memory is used to store computer instructions, and the processor is used to execute the computer instructions stored in the memory to control the transceiver unit to transmit and receive signals.

[0091] As an implementation, the functions of the receiver and the transmitter in the present application can be considered to be implemented by a transceiver circuit or a dedicated transceiver chip, and the processor can be considered to be implemented by a dedicated processing chip, a processing circuit or a general-purpose chip.

[0092] As another implementation, it can be considered to use a general-purpose computer to implement the server provided in the embodiments of the present application. That is, the program codes for implementing the functions of the processor, the receiver and the transmitter are stored in the memory, and the general-purpose processor realizes the functions of the processor, the receiver and the transmitter by executing the codes in the memory.

[0093] The embodiments of the present application further provide a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the foregoing method for interval prediction of railway subgrade deformation monitoring data are implemented. The computer-readable storage medium may 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 art.

[0094] The embodiments of the present application further provide a computer program product, including a computer program. When the computer program is executed by a processor, the steps of the foregoing method for interval prediction of railway subgrade deformation monitoring data are implemented.

[0095] Those of ordinary skill in the art should understand 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 both. Specifically, whether to execute in hardware or software depends on the specific application and design constraints of the technical solution. A professional technician can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application. 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 application are programs or code segments for performing the required tasks. The program or code segment can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave on a transmission medium or a communication link.

[0096] It should be clear that the present application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, the detailed description of known methods is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application 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 application.

[0097] In the present application, features described and / or illustrated for one embodiment can be used in the same way or in a similar way in one or more other embodiments, and / or combined with the features of other embodiments or replace the features of other embodiments.

[0098] The above are only the preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the embodiments of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. A method for interval prediction of railway roadbed deformation monitoring data, characterized in that: include: Inputting a target deformation monitoring data sample of the railway embankment in the current period into a deformation value prediction model so that the deformation value prediction model outputs a deformation prediction value of the railway embankment in the next period; According to the deformation prediction value of the railway subgrade in the next time period, the model variance and noise variance corresponding to the deformation value prediction model, the deformation interval prediction result data of the railway subgrade in the next time period is obtained, wherein the model variance and noise variance are pre-trained based on the historical deformation monitoring data samples of the railway subgrade in each historical time period and the deformation monitoring data interval prediction model including multiple deformation value prediction models.

2. The railway roadbed deformation monitoring data interval prediction method according to claim 1 is characterized in that: Before inputting the target deformation monitoring data sample of the railway subgrade in the current period into the deformation value prediction model so that the deformation value prediction model outputs the deformation prediction value of the railway subgrade in the next period, the method further includes: Resampling the historical deformation monitoring data samples of the railway subgrade in each historical period in the original data set to obtain multiple sample data sets, wherein each of the historical deformation monitoring data samples is provided with a true value of the railway subgrade deformation as a label, and each of the sample data sets contains reconstructed deformation monitoring data samples corresponding to each of the historical deformation monitoring data samples; Based on each deformation value prediction model corresponding to each of the sample data sets, respectively, the deformation prediction value corresponding to each of the reconstructed deformation monitoring data samples in each of the sample data sets is obtained; and according to the preset significance level parameter value and the deformation prediction value corresponding to each of the reconstructed deformation monitoring data samples in each of the sample data sets, the model variance corresponding to the deformation value prediction model is obtained; And, using a deformation value prediction model to respectively obtain the deformation prediction values ​​corresponding to each of the historical deformation monitoring data samples, and constructing a target data set according to the deformation prediction values ​​corresponding to each of the historical deformation monitoring data samples and the true value of the railway roadbed deformation, and then inputting the target data set into the deformation value prediction model, and using the output result data of the deformation value prediction model as the noise variance corresponding to the deformation value prediction model; A deformation interval prediction formula is constructed based on the significance level parameter value, the model variance, the noise variance and preset deformation prediction parameters.

3. The railway roadbed deformation monitoring data interval prediction method according to claim 2 is characterized in that: The step of obtaining deformation interval prediction result data of the railway subgrade in the next time period according to the deformation prediction value of the railway subgrade in the next time period, the model variance and the noise variance corresponding to the deformation value prediction model, comprises: The predicted deformation value of the railway subgrade in the next time period is substituted into the deformation prediction parameter in the deformation interval prediction formula and the deformation interval prediction formula is solved to obtain the deformation interval prediction result data of the railway subgrade in the next time period.

4. The railway roadbed deformation monitoring data interval prediction method according to claim 2 is characterized in that: The historical deformation monitoring data samples of the railway subgrade in each historical period in the original data set are resampled to obtain multiple sample data sets, including: Based on a preset step size, the historical deformation monitoring data samples of the railway subgrade in each historical period in the original data set are resampled by Bootstrap resampling to obtain each sample data set containing different reconstructed deformation monitoring data samples corresponding to each of the historical deformation monitoring data samples.

5. The railway roadbed deformation monitoring data interval prediction method according to claim 2 is characterized in that: Before resampling the historical deformation monitoring data samples of the railway subgrade in each historical period in the original data set to obtain multiple sample data sets, the method further includes: Obtain the longitude and latitude data of the railway embankment in various historical periods based on Beidou satellite monitoring; Performing position calculation and data frequency adjustment on the latitude and longitude data of each historical period to obtain the deformation monitoring data of the railway subgrade in each historical period; Preprocessing the deformation monitoring data of each historical period to obtain the historical deformation monitoring data corresponding to each historical period, wherein the preprocessing includes: at least one of abnormal value and missing value processing, wavelet denoising and stationarity test; A fractional-order difference operation is performed on each of the historical deformation monitoring data to obtain historical deformation monitoring data samples of each historical period and form a corresponding original data set.

6. The railway roadbed deformation monitoring data interval prediction method according to claim 5 is characterized in that: The deformation value prediction model includes: an LSTM network provided with a fractional-order difference processing unit, wherein the fractional-order difference processing unit is used to perform a fractional-order difference operation on the obtained prediction result after the LSTM network performs deformation prediction on the input deformation monitoring data sample, so as to obtain the deformation prediction value corresponding to each of the deformation monitoring data samples, wherein the deformation monitoring data samples include: the reconstructed deformation monitoring data samples or the historical deformation monitoring data samples.

7. The method for interval prediction of railway roadbed deformation monitoring data according to any one of claims 1 to 6, characterized in that: Also includes: When or after the deformation monitoring data of the railway subgrade in the next period is obtained, the safety hazard detection result data of the railway subgrade in the next period is generated according to the deformation monitoring data of the next period and the deformation interval prediction result data of the railway subgrade in the next period.

8. A device for predicting intervals of railway roadbed deformation monitoring data, characterized in that: include: A deformation value prediction module, used for inputting a target deformation monitoring data sample of the railway embankment in the current period into a deformation value prediction model, so that the deformation value prediction model outputs a deformation prediction value of the railway embankment in the next period; A deformation interval prediction module is used to obtain the deformation interval prediction result data of the railway subgrade in the next time period based on the deformation prediction value of the railway subgrade in the next time period, the model variance and noise variance corresponding to the deformation value prediction model, wherein the model variance and noise variance are pre-trained based on historical deformation monitoring data samples of the railway subgrade in various historical time periods and a deformation monitoring data interval prediction model including multiple deformation value prediction models.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method for interval prediction of railway subgrade deformation monitoring data as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for interval prediction of railway roadbed deformation monitoring data as described in any one of claims 1 to 7 is implemented.

Images