Railway roadbed deformation risk identification method and device

Abstract

The invention provides a railroad bed deformation risk identification method and device, and the method comprises the steps: carrying out the partitioning of the historical deformation data of each monitoring point along a target railway according to the distribution similarity based on a gray wolf algorithm, building an independent principal component analysis model, obtaining the priori knowledge, obtaining a priori demixing matrix corresponding to each priori sub-block, and obtaining a priori demixing matrix corresponding to each priori sub-block; control limits corresponding to the priori # imgabs0 # statistical magnitude and the priori # imgabs1 # statistical magnitude are obtained; when data newly generated by each monitoring point is processed, grouping is carried out according to prior blocks obtained based on historical deformation data, a posterior # imgabs2 # statistical magnitude and a posterior # imgabs3 # statistical magnitude are rapidly calculated by using a corresponding prior unmixing matrix, local risk monitoring is carried out in combination with a control line obtained in prior knowledge, and the risk monitoring accuracy is improved. Bayesian comprehensive statistics are calculated for the posterior # imgabs4 # statistics and the posterior # imgabs5 # statistics, the global deformation risk is monitored by using a preset significance level value, local and global systematic risk identification for the hidden danger area is realized, and the identification efficiency and accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of railway risk supervision, and particularly to a method and device for identifying the risk of railway subgrade deformation. Background Art

[0002] In the prior art, by deploying Beidou high-precision positioning terminals in the risk hidden danger areas along the railway, the safety status of railway track infrastructure can be effectively monitored. The risk hidden danger areas along the railway mainly include geological disaster risk points, key infrastructure, etc. along the railway, and their spans range from several hundred meters to over a thousand meters. Therefore, multiple Beidou terminals are often deployed on site to jointly monitor the external risk hidden dangers along the railway. Existing research mainly gives the early warning method of Beidou deformation monitoring data at a certain monitoring point through data-driven methods. Although it can effectively identify the safety status of railway subgrade deformation, the analysis result only represents the risk hidden danger situation in a local area and cannot describe the overall safety status of the risk hidden danger area. Therefore, it is very urgent to jointly analyze the monitoring data of each group in the hidden danger area along the railway to realize the systematic risk identification of the hidden danger area. Moreover, the analysis and processing speed is slow and cannot meet the requirement of rapid response in a large range. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a method and device for identifying the risk of railway subgrade deformation to eliminate or improve one or more defects existing in the prior art and solve the problem that the prior art cannot perform rapid and efficient deformation risk monitoring on the local and overall parts of the railway using the prior knowledge of historical data.

[0004] One aspect of the present invention provides a method for identifying the risk of railway subgrade deformation, and the method includes the following steps: Obtain the historical deformation data of the target railway collected by each monitoring point in cooperation with the reference station; the reference station and the monitoring points are deployed at preset positions along the target railway; Group the historical deformation data of each monitoring point in pairs to construct a plurality of candidate sub-blocks, construct a plurality of grey wolves as the solutions corresponding to the grouped candidate sub-blocks based on the grey wolf algorithm, minimize the sum of the KL divergences of the historical deformation data corresponding to each sub-block within the solution, and solve to obtain a plurality of prior sub-blocks with optimal grouping; Construct an independent component analysis model for each of the prior sub-blocks, and output the prior statistic and its first type of control limit, prior statistic and its second type of control limit, and record the prior unmixing matrix corresponding to each prior sub-block; Obtain the deformation data to be analyzed collected by the monitoring points during the target time period, and construct sub-blocks to be analyzed according to the grouping of the prior sub-blocks; Calculate the posterior statistics and posterior statistics for the deformation data to be analyzed of each sub-block to be analyzed, in combination with the prior unmixing matrix; Fuse the posterior statistics and the posterior statistics of each sub-block to be analyzed through Bayesian comprehensive inference to obtain a Bayesian comprehensive statistic; Compare the posterior statistics and posterior statistics of each sub-block to be analyzed in combination with the first type of control limit and the second type of control limit to monitor the local deformation risk and its occurrence time; compare the Bayesian comprehensive statistic in combination with a preset significance level value to monitor the global deformation risk and its occurrence time.

[0005] In some embodiments, the historical deformation data and the deformation data to be analyzed are displacement data sequences collected at a set frequency.

[0006] In some embodiments, multiple grey wolves are constructed based on the grey wolf algorithm as the solutions corresponding to the grouping of the candidate sub-blocks, and the sum of the KL divergences of the historical deformation data corresponding to each sub-block within the solution is minimized to solve for multiple prior sub-blocks of the optimal grouping, including: Initialize the grey wolf group, where each grey wolf represents a potential solution for pairwise grouping of the historical deformation data of each monitoring point to obtain multiple sub-blocks; Calculate the KL divergence of the historical deformation data within each sub-block in the potential solution corresponding to each grey wolf and sum it as the fitness function; find the wolf with the highest fitness, the wolf with the second highest fitness, the wolf with the third highest fitness, and the remaining grey wolves as wolves; In the hunting stage, each grey wolf updates its own position according to the position of the wolves in this round, and the expression is as follows: ; ; ; ; where represents the position of the prey, X(t) represents the position of the grey wolf in the t-th round, D represents the distance the grey wolf needs to move, and both A and C are cooperation coefficients; represents the convergence factor, which linearly decays from 2 to 0 according to the number of iterations; and represent random numbers within 0 to 1; In the hunting stage, each grey wolf updates its own position according to the wolves and wolves in this round, and the expression is: ; ; ; ; ; ; ; where represents the current position of the wolf, represents the current position of the wolf, represents the current position of the wolf; represents the position of each grey wolf individual; and and represent random numbers between 0 and 2; represents the distance between the wolf and the prey, represents the distance between the wolf and the prey, represents the distance between the wolf and the prey; and and represent the distance coefficient vector; represents the updated position of the wolf, represents the updated position of the wolf, represents the updated position of the wolf; is the position of the grey wolf individual after the th iteration; Update the social hierarchy of the grey wolf pack; When the maximum number of iterations is reached or the KL divergence converges to the set threshold, the algorithm ends.

[0007] In some embodiments, an independent principal component analysis model is constructed for each of the prior sub-blocks, and the prior statistics and their first-class control limits, and the prior statistics and their second-class control limits corresponding to each prior sub-block are output, and the prior unmixing matrix corresponding to each prior sub-block is recorded, including: For the historical deformation data of the target railway It contains m variables, n samples, and the data set is divided into p prior sub-blocks. Each prior sub-block is expressed as ; Then the variables within the i-th prior sub-block are expressed as , denotes the -th measurement variable in the i-th prior sub-block; Decompose the i-th prior sub-block using the independent component analysis model, and express as linear combinations of unknown independent components , expressed as: ; In the formula, denotes the mixing matrix, denotes the independent component matrix, and E denotes the residual matrix; Ignoring the residual matrix, the independent component analysis model corresponding to the i-th prior sub-block is expressed as: ; Then the independent component matrix estimated from the historical deformation data is expressed as: ; In the formula, denotes the independent component matrix estimated from the original deformation monitoring data solved based on the FastICA algorithm, and then the prior demixing matrix is obtained.

[0008] In some embodiments, in combination with the prior demixing matrix, calculate the posterior statistic and the posterior statistic for the deformation data to be analyzed in each sub-block to be analyzed, including: Calculate the posterior statistic for each sub-block to be analyzed. The expression is: ; Wherein, denotes the posterior statistic of the i-th sub-block to be analyzed, denotes the -th newly input sample data of the sub-block to be analyzed, denotes the prior demixing matrix, T denotes matrix transpose, denotes the calculated independent component matrix; Calculate the posterior statistic for each sub-block to be analyzed. The expression is: ; Among them, represents the posterior statistic of the i-th sub-block to be analyzed, which is obtained by projecting the new sample onto the independent component space learned by the model and reconstructing based on these independent components. The expression is: .

[0009] In some embodiments, the first type of control limit of each of the prior sub-blocks is obtained by performing univariate kernel density estimation on the corresponding prior statistic; the second type of control limit of each of the prior sub-blocks is obtained by performing univariate kernel density estimation on the corresponding prior statistic.

[0010] In some embodiments, the posterior statistics and the posterior statistics of each sub-block to be analyzed are fused through Bayesian comprehensive inference to obtain the Bayesian comprehensive statistic, including: Converting the posterior statistics and the posterior statistics of each sub-block to be analyzed into conditional probabilities. The expression is: ; ; ; ; In the formula, represents the deformation data to be analyzed of the i-th sub-block to be analyzed, F represents the fault state, N represents the normal state, represents the SPE confidence limit of the prior sub-block corresponding to the i-th sub-block to be analyzed, represents the confidence limit of the prior sub-block corresponding to the i-th sub-block to be analyzed; represents the posterior statistics of the i-th sub-block to be analyzed, represents the posterior statistics of the i-th sub-block to be analyzed; represents the probability of being normal based on the statistic index , represents the prior probability of being abnormal based on the statistic index ; represents based on the SPE statistic index The normal probability, represents the prior probability of abnormality based on the SPE statistic index ; Then the posterior probability of abnormality can be expressed as: ; ; ; ; wherein, represents based on the statistic index the posterior probability of abnormality, represents the probability of abnormality, based on the statistic index the probability of abnormality; represents the posterior probability of abnormality based on the SPE statistic index ; represents the probability of abnormality based on the SPE statistic index ; Based on the preset significance level , define as , define as , then the Bayesian comprehensive statistic expression is: ; wherein, B represents the number of sub - blocks to be analyzed.

[0011] In some embodiments, by combining the first - type control limit and the second - type control limit, the posterior statistic and the posterior statistic of each sub - block to be analyzed are compared to monitor the local deformation risk and its occurrence time, including: Mark the positions above the first - type control limit in the posterior statistic of each sub - block to be analyzed as having local deformation risk and mark the corresponding risk period; And, mark the positions above the second - type control limit in the posterior statistic of each sub - block to be analyzed as having local deformation risk and mark the corresponding risk period; By combining the preset significance level value, the Bayesian comprehensive statistic is compared to monitor the global deformation risk and its occurrence time, including: Mark the positions above the significance level value in the Bayesian comprehensive statistic as having global deformation risk and mark the corresponding risk period.

[0012] On the other hand, the present invention also provides a computer-readable storage medium, on which a computer program / instructions are stored, and when the computer program / instructions are executed by a processor, the steps of the above method are implemented.

[0013] On the other hand, the present invention also provides a computer program product, including a computer program / instructions, characterized in that when the computer program / instructions are executed by a processor, the steps of the above method are implemented.

[0014] The beneficial effects of the present invention are at least: The method and device for identifying the risk of railway subgrade deformation according to the present invention block the historical deformation data of each monitoring point along the target railway according to the distribution similarity based on the gray wolf algorithm, and respectively establish independent principal component analysis models to obtain prior knowledge, and obtain the prior unmixing matrix corresponding to each prior sub-block, and prior statistics and prior control limits corresponding to the statistics; when processing the newly generated data of each monitoring point, group them according to the prior blocks obtained based on the historical deformation data, and use the corresponding prior unmixing matrix to quickly calculate the posterior statistics and posterior statistics, combine the control lines obtained from prior knowledge for local risk monitoring, and calculate the Bayesian comprehensive statistic for the posterior statistics and posterior statistics, and use the preset significance level value to monitor the global deformation risk. The present invention can quickly calculate the posterior statistics and posterior statistics of each block by using the prior knowledge obtained from the historical deformation data, and based on the Bayesian comprehensive reasoning fusion, realize the systematic risk identification for the local and global hidden danger areas, and improve the discrimination efficiency and accuracy.

[0015] The additional advantages, objects, and features of the present invention will be partially described below, and will become partially apparent to those of ordinary skill in the art after studying the following, or can be learned from the practice of the present invention. The objects and other advantages of the present invention can be realized and obtained by the structure specifically pointed out in the specification and the drawings.

[0016] Those skilled in the art will understand that the objects and advantages that can be achieved by the present invention are not limited to the above specifically described, and the above and other objects that the present invention can achieve will be more clearly understood according to the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present invention, form a part of this application, and do not limit the present invention. In the drawings: Figure 1 It is a schematic flow chart of the method for identifying the deformation risk of railway subgrade according to an embodiment of the present invention.

[0018] Figure 2 It is a schematic diagram of the layout of monitoring points in the method for identifying the deformation risk of railway subgrade according to an embodiment of the present invention.

[0019] Figure 3 It is a schematic logic diagram of the method for identifying the deformation risk of railway subgrade according to an embodiment of the present invention.

[0020] Figure 4 In (a) - (j) are the graphs of the change in the deformation displacement amounts collected at monitoring points 1 - 10.

[0021] Figure 5 In (a) - (j) are the probability density fitting graphs of the deformation displacement amounts collected at monitoring points 1 - 10.

[0022] Figure 6 They are 10 independent components separated from the deformation displacement amounts of monitoring points 1 - 10 through the ICA method.

[0023] Figure 7 In (a) is Figure 6 the statistic graph of the ICA model in Figure 6 and (b) is the statistic graph of the ICA model in

[0024] Figure 8 In (a) is the statistic graph of the ICA model of sub - block 1, Figure 8 and (b) is the statistic graph of the ICA model of sub - block 1, Figure 9 In (a) is the statistic graph of the ICA model of sub - block 2.

[0025] Figure 9 And (b) is the statistic graph of the ICA model of sub - block 2.

[0026] Figure 10 In (a) is the statistic graph of the ICA model of sub - block 3, Figure 10 and (b) is the statistic graph of the ICA model of sub - block 3.

[0027] Figure 11 In (a) is the statistic graph of the ICA model of sub - block 4, Figure 11 and (b) is the statistic graph of the ICA model of sub - block 4.

[0028] Figure 12 In (a), it is the statistic graph of the ICA model of sub-block 5, Figure 12 and in (b), it is the statistic graph of the ICA model of sub-block 5.

[0029] Figure 13 It is the BIC statistic graph of monitoring points 1 to 10. Specific implementation manners

[0030] To make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with the implementation manners and the drawings. Herein, the illustrative implementation manners of the present invention and the descriptions thereof are used to explain the present invention, but do not limit the present invention.

[0031] Herein, it also needs to be noted that in order to avoid obscuring the present invention due to unnecessary details, only the structures and / or processing steps closely related to the solution according to the present invention are shown in the drawings, and other details less related to the present invention are omitted.

[0032] 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.

[0033] For the existing railway subgrade deformation monitoring schemes, most of them perform independent analysis on individual monitoring points, focusing on the monitoring of local deformation risks and lacking overall systematic evaluation. Facing the continuously expanding railway network, the demand for large-scale systematic deformation risk assessment is becoming increasingly prominent, and the demand for fast and efficient deformation risk assessment detection ability is getting higher and higher.

[0034] One aspect of the present invention provides a method for identifying railway subgrade deformation risks, as Figure 1 shown, the method includes the following steps S101 to S107: Step S101: Obtain the historical deformation data of the target railway collected by each monitoring point in cooperation with the reference station; the reference station and the monitoring points are deployed at preset positions along the target railway.

[0035] Step S102: Group the historical deformation data of each monitoring point in pairs to construct a plurality of candidate sub-blocks, construct a plurality of grey wolves as solutions corresponding to the grouped candidate sub-blocks based on the grey wolf algorithm, minimize the sum of the KL divergences of the historical deformation data corresponding to each sub-block within the solution, and solve to obtain a plurality of prior sub-blocks with optimal grouping.

[0036] Step S103: Construct an independent component analysis model for each prior sub-block and output the prior Statistics and their first - type control limits, prior Statistics and their second - type control limits, and record the prior unmixing matrix corresponding to each prior sub - block.

[0037] Step S104: Obtain the deformation data to be analyzed collected at the monitoring points during the target period, and construct sub - blocks to be analyzed according to the grouping of prior sub - blocks.

[0038] Step S105: Combine the prior unmixing matrix, and calculate the posterior statistics and posterior statistics for the deformation data to be analyzed of each sub - block to be analyzed.

[0039] Step S106: Through Bayesian comprehensive inference, fuse the posterior statistics and posterior statistics of each sub - block to be analyzed to obtain the Bayesian comprehensive statistic.

[0040] Step S107: Combine the first - type control limit and the second - type control limit to compare the posterior statistics and posterior statistics of each sub - block to be analyzed to monitor the local deformation risk and its occurrence time; combine the preset significance level value to compare the Bayesian comprehensive statistic to monitor the global deformation risk and its occurrence time.

[0041] In step S101, as Figure 2 shown, set monitoring points for the subgrade of the target railway or the building itself, which are used to collect displacement data as deformation data through Beidou positioning, GPS positioning or other positioning forms. In this process, in order to improve the detection accuracy, a reference station is introduced for auxiliary detection.

[0042] In some embodiments, the historical deformation data and the deformation data to be analyzed are displacement data sequences collected at a set frequency.

[0043] In step S102, referring to Figure 3 , in the offline mode, the historical deformation data already generated by each monitoring point is used to obtain prior knowledge. The risk - hidden area can be regarded as a whole with certain correlations. In order to improve the data utilization rate and the stability and robustness of deformation risk monitoring, for the historical deformation monitoring data generated by multiple monitoring points, they can be divided into several sub - blocks according to the distribution characteristics, and independent component analysis models (ICA early - warning models) are established for each sub - block to obtain the prior unmixing matrix, the control limits of the statistics, and the control limits of the

[0044] The monitoring points are grouped in pairs to construct sub - blocks. In the offline mode, the sub - blocks constructed using historical deformation data are prior sub - blocks, and the sub - blocks generated during the subsequent analysis process are sub - blocks to be analyzed. The sub - blocks to be analyzed are directly generated based on the grouping results of the prior sub - blocks without the need for additional processing, and thus the prior knowledge obtained from historical deformation data can be directly applied.

[0045] The core purpose of block division is to classify the data of monitoring points with similar data distribution characteristics into one category and monitor them simultaneously to optimize the detection efficiency.

[0046] In some embodiments, multiple grey wolves are constructed based on the grey wolf algorithm as the solutions for the corresponding groups of the candidate sub - blocks, and the sum of the KL divergences of the historical deformation data corresponding to each sub - block within the solution is minimized to obtain multiple prior sub - blocks with the optimal grouping, including: Step S201: Initialize the grey wolf group. Each grey wolf represents a potential solution for pairwise grouping of the historical deformation data of each monitoring point to obtain multiple sub - blocks.

[0047] Step S202: Calculate the sum of the KL divergences of the historical deformation data within each sub - block in the potential solution corresponding to each grey wolf as the fitness function; find the wolf with the highest fitness, the wolf with the second - highest fitness, the wolf with the third - highest fitness and regard the remaining grey wolves as wolves.

[0048] Step S203: In the hunting stage, each grey wolf updates its own position according to the position of the wolves in this round, and the expression is as follows: ; ; ; ; where, represents the position of the prey, X(t) represents the position of the grey wolf at the t - th round, D represents the distance the grey wolf needs to move, and both A and C are cooperation coefficients; represents the convergence factor, which linearly decays from 2 to 0 according to the number of iterations; and represent random numbers within 0 to 1; Step S204: In the hunting stage, each grey wolf updates its own position according to the wolves, wolves, wolves in this round, and the expression is: ; ; ; ; ; ; ; Among them, represents the current position of the wolf, represents the current position of the wolf, represents the current position of the wolf; represents the position of each individual grey wolf; , , represent random numbers between 0 and 2; represents the distance between the wolf and the prey, represents the distance between the wolf and the prey, represents the distance between the wolf and the prey; , , represent the distance coefficient vector; represents the updated position of the wolf, represents the updated position of the wolf, represents the updated position of the wolf; is the position of the individual grey wolf after the th iteration.

[0049] Step S205: Update the social rank of the grey wolf pack.

[0050] Step S206: When the maximum number of iterations is reached or the KL divergence converges to the set threshold, the algorithm ends.

[0051] Among them, the calculation method of the KL divergence is as follows: ; In the formula, represents the KL divergence, and represent the distribution probabilities of two historical deformation data random variables in each sub-block group. When the two distributions are similar, the KL divergence approaches 0, and vice versa, the KL divergence is larger.

[0052] In step S103, refer toFigure 3 For each of the multiple prior sub - blocks obtained after grouping, an independent component analysis model (ICA early - warning model) is established separately.

[0053] In some embodiments, an independent component analysis model is constructed for each prior sub - block, and the prior statistic corresponding to each prior sub - block, its first - type control limit, prior statistic and its second - type control limit are output, and the prior demixing matrix corresponding to each prior sub - block is recorded, including steps S301 - S303: Step S301: For the historical deformation data of the target railway containing m variables and n samples, the data set is divided into p prior sub - blocks, and each prior sub - block is expressed as ; then the variables within the i - th prior sub - block are expressed as , indicating the - th measurement variable in the i - th prior sub - block.

[0054] Step S302: Use the independent component analysis model to decompose the i - th prior sub - block, and express as a linear combination of unknown independent components , expressed as: ; In the formula, represents the mixing matrix, represents the independent component matrix, and E represents the residual matrix.

[0055] Step S303: Ignoring the residual matrix, the independent component analysis model corresponding to the i - th prior sub - block is expressed as: ; Then the independent component matrix estimated from the historical deformation data is expressed as: ; In the formula, represents the independent component matrix estimated from the original deformation monitoring data solved based on the FastICA algorithm, and then the prior demixing matrix is obtained.

[0056] In some embodiments, the first - type control limit of each prior sub - block is obtained by performing univariate kernel density estimation on the corresponding prior statistic.

[0057] The second - type control limit of each prior sub - block is obtained by performing univariate kernel density estimation on the corresponding prior statistic.

[0058] Hypothesis is an independent and identically distributed sample of the random variable y. The mathematical description of the univariate kernel density estimation is as follows: ; In the formula, is the kernel density estimation of the unknown probability density function , m represents the sample size, represents the process data of the sub-block, represents the kernel function, represents the smoothing parameter or bandwidth. For each sub-block, all the statistics and their corresponding control limits are established. If the statistic is less than the corresponding statistical limit, it indicates that the area represented by the sub-block is overall stable; otherwise, it indicates the existence of certain potential risks, that is: ; .

[0059] The prior statistic and the prior statistic of each prior sub-block are used to calculate the univariate kernel density estimation as the corresponding control limits, that is, the first type of control limit and the second type of control limit .

[0060] In step S104, referring to Figure 3 , during the online process, the monitoring point collects data to perform the railway subgrade deformation risk discrimination task, continuously generating new deformation data to be analyzed. Based on prior knowledge, the monitoring points are grouped according to the prior sub-blocks, and the corresponding deformation data to be analyzed are used to quickly solve the posterior statistic and the posterior statistic by using the prior unmixing matrix, and the local deformation risk is monitored based on the first type of control limit and the second type of control limit .

[0061] In some embodiments, in combination with the prior unmixing matrix, the posterior statistic and the posterior statistic of the deformation data to be analyzed for each sub-block to be analyzed are calculated, including steps S401~S402: Step S401: Calculate the posterior statistic of each sub-block to be analyzed. The expression is: ; Among them, represents the posterior statistic of the i-th sub-block to be analyzed, represents the -th newly input sample data of the sub-block to be analyzed, represents the prior unmixing matrix, and T represents the matrix transpose. represents the calculated independent component matrix.

[0062] Step S402: Calculate the posterior statistic for each sub-block to be analyzed, and the expression is: ; where represents the posterior statistic of the i-th sub-block to be analyzed, is obtained by projecting the new sample onto the independent component space learned by the model and reconstructing based on these independent components, and the expression is: .

[0063] In step S106, referring to Figure 3 , based on the posterior statistic and posterior statistic calculated for each sub-block to be analyzed in steps S401~S402, calculate the Bayesian comprehensive statistic to reflect the overall deformation situation of the target railway section.

[0064] In some embodiments, by Bayesian comprehensive inference, fuse the posterior statistic and posterior statistic of each sub-block to be analyzed to obtain the Bayesian comprehensive statistic, including steps S501~S50: Step S501: Convert the posterior statistic and the posterior statistic of each said sub-block to be analyzed into conditional probabilities, and the expression is: ; ; ; ; In the formula, represents the deformation data to be analyzed of the i-th sub-block to be analyzed, F represents the fault state, N represents the normal state, represents the SPE confidence limit of the prior sub-block corresponding to the i-th sub-block to be analyzed, represents the confidence limit of the prior sub-block corresponding to the i-th sub-block to be analyzed; represents the posterior statistic of the i-th sub-block to be analyzed, represents the posterior statistic of the i-th sub-block to be analyzed; Indicates the probability based on the statistic index being normal, Indicates the prior probability based on the statistic index being abnormal; Indicates the probability based on the SPE statistic index being normal, Indicates the prior probability based on the SPE statistic index being abnormal; Then the posterior probability of being abnormal can be expressed as: ; ; ; ; Among them, Indicates the posterior probability of being abnormal based on the statistic index , Indicates the probability of being abnormal, Based on the statistic index of being abnormal; Indicates the posterior probability of being abnormal based on the SPE statistic index , Indicates the probability of being abnormal based on the SPE statistic index ; Step S502: Based on a preset significance level , define as , define as , then the Bayesian comprehensive statistic expression is: ; Among them, B represents the number of sub - blocks to be analyzed.

[0065] In step S107, by combining the first - type control limit and the second - type control limit, compare the posterior statistic and the posterior statistic of each sub - block to be analyzed to monitor the local deformation risk and its occurrence time, including steps S601~S602: Step S601: Mark the posterior statistic of each sub - block to be analyzed that is higher than the first - type control limit as having a local deformation risk and mark the corresponding risk period.

[0066] Step S602: For the posterior statistics of each sub-block to be analyzed, those higher than the second type of control limit are marked as having a risk of local deformation, and the corresponding risk period is marked.

[0067] Compare the Bayesian comprehensive statistic with the preset significance level value to monitor the global deformation risk and its occurrence time, including Step S701: Mark those in the Bayesian comprehensive statistic higher than the significance level value as having a global deformation risk and mark the corresponding risk period.

[0068] On the other hand, the present invention also provides a computer-readable storage medium, on which a computer program / instructions are stored, and when the computer program / instructions are executed by a processor, the steps of the above method are implemented.

[0069] On the other hand, the present invention also provides a computer program product, including a computer program / instructions, characterized in that when the computer program / instructions are executed by a processor, the steps of the above method are implemented.

[0070] The present invention is described below in conjunction with a simulation experiment: To verify that the proposed method can effectively achieve the active warning of railway track infrastructure, measured data of the track infrastructure in a typical section is selected for analysis. This section of the infrastructure is located in the North China Plain with complex geological conditions. To prevent major risk hazards and ensure the safe operation of the railway, the railway safety management department has deployed a total of 1 reference station and 10 monitoring stations in this risk hazard area to monitor the displacement of the track infrastructure. The maximum distance between the reference station and the monitoring stations does not exceed 300 meters, and the monitoring stations are evenly distributed on both sides of the railway line. Figure 2 It is a schematic diagram of the layout of the monitoring stations. The time span of the monitoring data is from September 1, 2022 to October 10, 2022, a total of 40 days. The data acquisition frequency is 1 epoch per 10 seconds, and one solution period is 10 minutes, with a total of 5760 data points.

[0071] Figure 4 In (a)-(j) are the original data time series diagrams of each monitoring point in this risk hazard area. It can be seen from the figure that the monitoring data can reflect the deformation trend of the railway track infrastructure within 40 days, and at the same time, it can also be observed that there are relatively serious spike features caused by external environmental factors. It can be observed from the figure that most of the monitoring points change relatively smoothly within 40 days, but obvious settlement trends can still be observed at monitoring points 5 and 9. For railway safety management personnel, it is also of great significance to judge the overall safety status of the hidden danger area through the status information of each monitoring point. Based on this, the present invention proposes to realize the systematic risk identification of the hidden danger area based on Bayesian and ICA theories.

[0072] First, perform statistical characteristic analysis on each group of monitoring data in the hidden danger area.Figure 5 In (a)-(j), the probability density fitting graphs of each monitoring point are shown. Among them, the blue bars represent the probability density distribution of the original data, the red represents the fitting of the traditional Gaussian probability density distribution, and the green represents the stable distribution fitting. Table 1 shows the fitting parameter values of the stable distribution for each monitoring point. From Figure 5 in (a)-(j), it can be seen that the stable distribution fitting can better describe the true probability density distribution of the data. The stable distribution curve and the traditional Gaussian distribution curve do not completely coincide, indicating that the monitoring data of each group in this area do not follow the Gaussian distribution. From the fitting results, no serious tailing or skewed peak characteristics appear in each group of data, indicating that the data in this hidden danger area has not undergone relatively serious deformation, which is consistent with the relatively stable results of the time series diagrams of each group of monitoring data. The stable distribution parameters of each group of monitoring data shown in Table 1 are also not equal to 2, further verifying that each group of monitoring data follows a non-Gaussian distribution. For such non-Gaussian distributed monitoring data, the ICA modeling method can be considered to realize the discrimination of abnormal states in the hidden danger area.

[0073] Table 1 Probability density fitting parameters Figure 6 are 10 independent components separated from 10 groups of monitoring data by the ICA method. It can be seen from the figure that the overall change trends of the 10 groups of independent components are relatively gentle, but obvious oscillations occur in the three groups of data ICA-1, ICA-2, and ICA-3 on September 11, indicating that these three components are significantly disturbed in the corresponding areas. Using the first 30 days of data as the training set to construct an ICA model aims to obtain and the control limits, that is, the discrimination indicators of the stability degree of the entire area, and then using the last 10 days of data as the test set to verify and discriminate the stability of this area. The principle of ICA is actually to calculate the statistic at each moment and then compare it with the control limit. If the statistic exceeds the corresponding control limit, it indicates that an abnormality has occurred in this area.

[0074] Figure 7 In (a) and (b), ICA modeling analysis is performed on 10 groups of monitoring data. Among them, Figure 7 in Figure (a) shows the statistics of each time point of the monitoring data, while Figure 7 in Figure (b) shows the statistics of each time point of the monitoring data. The black line in the figure represents the statistic corresponding to the time obtained by the ICA method, and the red line is the control limit obtained from the training data. It can be seen from the figure that whether it is The statistic is still the statistic, and overall, none of them completely exceed the corresponding control limits, indicating that the geology of this area is relatively stable. In addition, it can also be clearly observed that around October 3, and both statistics exceed the control limits, indicating that the geological area is relatively active during this time period. The results of the two statistics can corroborate each other, further proving the accuracy and reliability of this method. In addition, it can also be seen that the statistic also exceeds the control limit around October 10, and its control limit is relatively the control limit is also smaller, indicating that compared with it has a higher sensitivity to capturing abnormal signals.

[0075] To further improve the robustness of the ICA model and the utilization rate of monitoring data, and to evaluate the risk status of potential hazard areas in multiple dimensions, the monitoring data of each group is grouped to establish a distributed ICA model. In the present invention, optimization is carried out through the grey wolf algorithm model. First, the grey wolf packs are set, that is, the pairwise permutations and combinations of the monitoring data of each group, and their corresponding KL divergences are calculated. The grouping corresponding to the minimum KL divergence is the alpha wolf, and then iteration is carried out until all the grouping results are selected. The purpose is to make the distributions of the data within the group similar. The grouping results are shown in Table 2. There are a total of 10 groups of monitoring data, which are evenly divided into 5 sub-blocks, that is, monitoring points 5 and 6 are in one group, monitoring points 1 and 10 are in one group, monitoring points 2 and 4 are in one group, monitoring points 8 and 9 are in one group, and monitoring points 3 and 7 are in one group.

[0076] Table 2 Sub-block results of each monitoring point in the potential hazard area Figures 8 to 12 For the and statistics corresponding to the five groups of monitoring data, it can be observed from the figure that the monitoring data within each grouping do not show the situation of exceeding the control limits in a large range, which is basically consistent with the ICA model results of the above 10 groups of monitoring data. In addition, it can also be observed from the figure that there are also local differences between the statistics of each grouping and the 10-group statistics. For example, the statistic of sub-block 4 shows an anomaly around October 8, while the Figure 8 statistic shown does not show an obvious anomaly in this section. The reason is that the Figure 8 statistic shown is a risk discrimination index given based on the entire potential hazard area, while each grouping is based on two of the variables, so it can better reflect the specific detailed characteristics. In addition, the and control limits of each sub-block and Figure 7The control limits shown are also different because the statistical limits of each sub-block are obtained based on the training data of their respective sub-blocks.

[0077] Figure 13 The distributed ICA model established according to the statistics of five sub-blocks, and its control limit is the corresponding confidence level. The red line in the figure is the corresponding confidence level control limit, and the black line is the statistic corresponding to each moment. It can be observed from the figure that the statistics within each group do not show a large range exceeding the control limit, which is basically consistent with the ICA model results of the above 10 groups of monitoring data. In addition, it can also be observed from the figure that the distributed ICA model shows local anomalies not only on October 3 and October 10, but also on October 8, which indicates that the distributed ICA model can reflect the local anomaly information of each group compared with the traditional ICA model. In addition, from the and statistics of each sub-block, the geology of this area was indeed relatively active on October 3, but the overall situation still did not reach a relatively serious standard. In fact, the control limit is only used as an alarm threshold, and the distance between the statistic and the control limit can also be used to determine the overall stability of the potential hazard area. Whether it is a traditional ICA model or a distributed ICA model, there are individual discrete points exceeding the control limit threshold. However, for the overall risk hazard area, it can be set that several consecutive statistics exceed the control limit threshold, and then it can be determined that the overall area enters a relatively serious risk state.

[0078] Corresponding to the above method, the present invention also provides a device / system, which includes a computer device. The computer device includes a processor and a memory. The memory stores computer instructions, and the processor is used to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the device / system implements the steps of the method described above.

[0079] The embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the foregoing edge computing server deployment method. The computer-readable storage medium can be a tangible storage medium, such as a random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, register, floppy disk, hard disk, removable storage disk, CD-ROM, or any other form of storage medium well-known in the technical field.

[0080] In summary, the method and device for identifying the risk of railway subgrade deformation according to the present invention block the historical deformation data of each monitoring point along the target railway according to the distribution similarity based on the gray wolf algorithm, and respectively establish an independent component analysis model to obtain prior knowledge, and obtain the prior demixing matrix corresponding to each prior sub-block, and prior Statistics and prior The control limits corresponding to the statistics; when processing the newly generated data of each monitoring point, group them according to the prior blocks obtained based on the historical deformation data, and use the corresponding prior unmixing matrix to quickly calculate the posterior Statistics and posterior Statistics, combine the control lines obtained from prior knowledge for local risk monitoring, for the posterior Statistics and posterior Statistics and posterior statistics, calculate the Bayesian comprehensive statistic, and use the preset significance level value to monitor the global deformation risk. The present invention can quickly calculate the posterior statistics of each block by using the prior knowledge obtained from the historical deformation data Statistics and posterior Statistics and posterior statistics, based on Bayesian comprehensive reasoning and fusion, realize the systematic risk identification for the local and global of the hidden danger area, and improve the discrimination efficiency and accuracy.

[0081] Those of ordinary skill in the art should understand that the various 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 implement in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed 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 functional 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 through a data signal carried in a carrier wave on a transmission medium or a communication link.

[0082] It should be clear that the present invention 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 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.

[0083] In the present invention, the features described and / or illustrated for one embodiment can be used in the same or similar manner in one or more other embodiments, and / or combined with the features of other embodiments or replace the features of other embodiments.

[0084] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the embodiments of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A railway roadbed deformation risk identification method, characterized in that: The method comprises the following steps: Acquire historical deformation data of the target railway collected by each monitoring point in cooperation with a reference station; the reference station and the monitoring point are deployed at preset positions along the target railway; The historical deformation data of each monitoring point are grouped into multiple candidate sub-blocks in pairs, and multiple gray wolves are constructed as solutions corresponding to the groups of the candidate sub-blocks based on the gray wolf algorithm, and the sum of the KL divergence of the historical deformation data corresponding to each sub-block in the solution is minimized to obtain multiple a priori sub-blocks of the optimal grouping; Construct an independent principal component analysis model for each prior sub-block, and output the prior corresponding to each prior sub-block Statistics and their first-type control limits, priors Statistics and their second kind control limits, and recording the prior unmixing matrix corresponding to each of the prior sub-blocks; Acquire the deformation data to be analyzed collected by the monitoring point within the target time period, and construct the sub-blocks to be analyzed according to the grouping of the priori sub-blocks; Combined with the prior unmixing matrix, the posterior is calculated for the deformation data to be analyzed of each sub-block to be analyzed Statistics and Posterior Statistics; The posterior of each sub-block to be analyzed is integrated through Bayesian comprehensive reasoning Statistics and the posterior Statistics, get the Bayesian comprehensive statistics; Combine the first type of control limit and the second type of control limit to compare the posterior of each sub-block to be analyzed Statistics and Posterior The statistics monitor the local deformation risk and its occurrence time; and the Bayesian comprehensive statistics are compared with the preset significance level value to monitor the global deformation risk and its occurrence time.

2. The railway roadbed deformation risk identification method according to claim 1 is characterized in that: The historical deformation data and the deformation data to be analyzed are displacement data sequences collected at a set frequency.

3. The railway roadbed deformation risk identification method according to claim 1 is characterized in that: Based on the gray wolf algorithm, multiple gray wolves are constructed as solutions for the grouping of the candidate sub-blocks, the sum of the KL divergences of the historical deformation data corresponding to each sub-block in the solution is minimized, and multiple a priori sub-blocks of the optimal grouping are obtained by solving the solution, including: Initialize the wolf pack, each wolf represents the potential solution of multiple sub-blocks obtained by grouping the historical deformation data of each monitoring point in pairs; Calculate the KL divergence of the historical deformation data in each sub-block in the potential solution corresponding to each gray wolf and sum them as the fitness function; find the one with the highest fitness according to the fitness function corresponding to each gray wolf Wolves, second most adaptable Wolves, third most adaptable Wolf and the remaining gray wolves as Wolf; In the hunting phase, each wolf The wolf's position updates its own position. The expression is as follows: ； ； ； ； in, represents the position of the prey, X(t) represents the position of the wolf in round t, D represents the distance the wolf has to move, and A and C are both coordination coefficients; represents the convergence factor, which decays linearly from 2 to 0 according to the number of iterations; and Represents a random number between 0 and 1; During the hunting phase, each wolf Wolf, Wolf, The wolf updates its position, the expression is: ； ； ； ； ； ； ； in, express The wolf's current location, express The wolf's current location, express The wolf's current location; Indicates the location of each individual gray wolf; , , Represents a random number between 0 and 2; express The distance between the wolf and its prey, express The distance between the wolf and its prey, express the distance between the wolf and its prey; , , represents the distance coefficient vector; express Updated positions of wolves, express Updated positions of wolves, express Updated positions of wolves; For the The position of the individual gray wolf after iterations; Updates the social hierarchy of gray wolf packs; The algorithm ends when the maximum number of iterations is reached or the KL divergence and convergence reach the set threshold.

4. The railway roadbed deformation risk identification method according to claim 2 is characterized in that: Construct an independent principal component analysis model for each prior sub-block, and output the prior corresponding to each prior sub-block Statistics and their first-type control limits, priors Statistics and their second-class control limits, and record the prior unmixing matrix corresponding to each of the prior sub-blocks, including: The historical deformation data of the target railway Containing m variables and n samples, the data set is divided into p prior sub-blocks, each of which is represented by ; then the variables in the i-th prior sub-block are expressed as , represents the first in the ith prior sub-block measured variables; The independent principal component analysis model is used to decompose the i-th prior sub-block. Expressed as Unknown independent components The linear combination of is expressed as: ； In the formula, represents the mixing matrix, represents the independent component matrix, and E represents the residual matrix; Ignoring the residual matrix, the independent principal component analysis model corresponding to the i-th prior sub-block is expressed as: ； Then the independent component matrix estimated by the historical deformation data is expressed as: ； In the formula, represents the independent component matrix estimated from the original deformation monitoring data based on the FastICA algorithm, and the prior unmixing matrix is ​​obtained by solving .

5. The railway roadbed deformation risk identification method according to claim 4 is characterized in that: Combined with the prior unmixing matrix, the posterior is calculated for the deformation data to be analyzed of each sub-block to be analyzed Statistics and Posterior Statistics, including: Calculate the posterior for each sub-block to be analyzed Statistics, the expression is: ； in, represents the posterior of the i-th sub-block to be analyzed Statistics, Indicates The newly input sample data of the sub-block to be analyzed, represents the prior unmixing matrix, T represents the matrix transpose, represents the calculated independent component matrix; Calculate the posterior for each sub-block to be analyzed Statistics, the expression is: ； in, represents the posterior of the i-th sub-block to be analyzed Statistics, By adding the new sample Projected onto the independent component space learned by the model and reconstructed based on these independent components, the expression is: 。 6. The railway roadbed deformation risk identification method according to claim 1 is characterized in that: The first type of control limit of each a priori sub-block is a The statistic is obtained by univariate kernel density estimation; The second type of control limit for each of the a priori sub-blocks is a The statistic is obtained by univariate kernel density estimation.

7. The railway roadbed deformation risk identification method according to claim 1 is characterized in that: The posterior of each sub-block to be analyzed is integrated through Bayesian comprehensive reasoning Statistics and the posterior Statistics, get Bayesian comprehensive statistics, including: The posterior Statistics and the posterior The statistic is converted into conditional probability, and the expression is: ； ； ； ； In the formula, represents the deformation data to be analyzed of the i-th sub-block to be analyzed, F represents the fault state, N represents the normal state, represents the SPE confidence limit of the prior sub-block corresponding to the i-th sub-block to be analyzed, Indicates the prior sub-block corresponding to the i-th sub-block to be analyzed Confidence limits; represents the posterior of the i-th sub-block to be analyzed Statistics, represents the posterior of the i-th sub-block to be analyzed Statistics; Indicates based on Statistical indicators Normal probability, Indicates based on Statistical indicators Prior probability of anomaly; Indicates the SPE statistic indicator Normal probability, Indicates the SPE statistic indicator Prior probability of anomaly; but The posterior probability of being abnormal can be expressed as: ； ； ； ； in, Indicates based on Statistical indicators The posterior probability of anomaly, express The probability of abnormality, based on Statistical indicators The probability of anomaly; Indicates the SPE statistic indicator The posterior probability of anomaly, Indicates the SPE statistic indicator Probability of abnormality; Based on the preset significance level ,definition for ,definition for , then the Bayesian comprehensive statistic expression is: ； Wherein, B represents the number of sub-blocks to be analyzed.

8. The railway roadbed deformation risk identification method according to claim 1 is characterized in that: Combine the first type of control limit and the second type of control limit to compare the posterior of each sub-block to be analyzed Statistics and Posterior Statistics monitor local deformation risks and when they occur, including: The posterior of each sub-block to be analyzed The statistical quantity above the first type of control limit is marked as having a local deformation risk and the corresponding risk period is marked; And, the posterior of each sub-block to be analyzed The statistical quantity above the second type of control limit is marked as having local deformation risk and the corresponding risk period is marked; The Bayesian comprehensive statistic is compared with a preset significance level value to monitor the global deformation risk and its occurrence time, including: marking the location of the Bayesian comprehensive statistic that is higher than the significance level value as having a global deformation risk and marking the corresponding risk period.

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 according to any one of claims 1 to 8 are implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images