Water damage assessment method and device, computer equipment and readable storage medium

Abstract

The invention relates to a water damage assessment method and device, computer equipment and a readable storage medium. The method comprises the following steps: acquiring an attribute data set and a radar data set of a railway ballast bed to be detected; simulating each simulated water-containing system value of the to-be-detected railway ballast bed under each relative dielectric constant according to the attribute data set, and constructing an evaluation standard of the to-be-detected railway ballast bed according to each simulated water-containing system value; extracting waveform features in the radar data set to obtain a feature data set, and determining a water-containing system value of the to-be-detected railway ballast bed according to the feature data set and a water-containing coefficient algorithm; and determining a water damage evaluation result of the to-be-detected railway ballast bed according to the evaluation standard and the water-containing system value of the to-be-detected railway ballast bed. The method can improve the accuracy of the water damage assessment method.

Description

Technical Field

[0001] The present application relates to the field of railway detection technology, and in particular to a water damage assessment method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art

[0002] Water damage to railway trackbeds refers to the accumulation of water within the trackbed structure due to factors such as rainwater infiltration, rising groundwater levels, or changes in ambient humidity. Water damage can reduce the trackbed's load-bearing capacity and structural stability, threatening driving safety. Therefore, it is necessary to assess the extent of water damage to ensure maintenance and ensure safe operation.

[0003] Traditionally, radar is used to survey the railway trackbed to generate a radar data set, which is then transferred to a computer and displayed on the computer as a radar map. Based on the radar map and their experience, inspectors determine the extent of water damage to the trackbed.

[0004] However, traditional technologies rely on manual determination of water damage extent based on experience and radar datasets, which is subject to many subjective factors. Consequently, current water damage assessment methods have low accuracy. Summary of the Invention

[0005] Based on this, it is necessary to provide a water damage assessment method, device, computer equipment, computer-readable storage medium and computer program product to address the above technical problems.

[0006] In a first aspect, the present application provides a water damage assessment method, comprising:

[0007] Obtain the attribute dataset and radar dataset of the railway track bed to be detected;

[0008] Simulating various simulated moisture coefficient values ​​of the railway track bed to be detected at various relative dielectric constants according to the attribute data set, and constructing an evaluation standard for the railway track bed to be detected according to the various simulated moisture coefficient values;

[0009] Extracting waveform features from the radar data set to obtain a feature data set, and determining a moisture coefficient value of the railway track bed to be detected based on the feature data set and a moisture coefficient algorithm;

[0010] The water damage assessment result of the railway track bed to be detected is determined according to the evaluation standard of the railway track bed to be detected and the water content coefficient value.

[0011] In one embodiment, simulating various simulated moisture coefficient values ​​of the railway track bed to be detected at various relative dielectric constants according to the attribute data set, and constructing an evaluation standard for the railway track bed to be detected according to the simulated moisture coefficient values, includes:

[0012] Establishing a concentric circle roadbed model of the railway roadbed to be detected according to the attribute data set;

[0013] Simulating a simulated radar data set of the railway roadbed to be detected at various relative dielectric constants according to the concentric circle roadbed model;

[0014] The simulated water content values ​​corresponding to the relative dielectric constants are determined based on the simulated radar data set, and an evaluation standard for the railway track bed to be detected is constructed according to the simulated water content values ​​and the damage degrees corresponding to the relative dielectric constants.

[0015] In one embodiment, the concentric circle roadbed model includes a healthy concentric circle roadbed model and a water-containing concentric circle roadbed model, and simulating a simulated radar data set of the railway roadbed to be detected at various relative dielectric constants based on the concentric circle roadbed model includes:

[0016] A healthy simulated radar data subset of the railway roadbed to be detected in a healthy state is simulated based on the healthy concentric circle roadbed model;

[0017] Simulating each subset of water-containing simulated radar data of the railway track bed to be detected in a water-containing state based on the water-containing concentric circle track bed model and each relative dielectric constant;

[0018] A simulated radar data set is constructed based on the healthy simulated radar data subset and each of the water-containing simulated radar data subsets.

[0019] In one embodiment, the simulated radar data set includes simulated radar data subsets corresponding to the relative dielectric constants, and determining the simulated water content values ​​corresponding to the relative dielectric constants based on the simulated radar data set includes:

[0020] For each of the simulated radar data subsets corresponding to the relative dielectric constant, the simulated radar data subsets are screened according to the water-bearing areas in the concentric circle roadbed model, and the screened simulated radar data subsets are averaged to obtain average simulated waveform data;

[0021] Performing wavelet transformation on the average analog waveform data, and extracting a simulation feature set of the average analog waveform data after the wavelet transformation;

[0022] Data calculation is performed on the simulation feature set according to a water content coefficient algorithm to obtain a simulated water content coefficient value corresponding to the relative dielectric constant.

[0023] In one embodiment, extracting waveform features from the radar data set to obtain a feature data set includes:

[0024] filtering waveform data of each initial time domain waveform in the radar data set to obtain waveform data of each time domain waveform;

[0025] Averaging the waveform data of each of the time domain waveforms to obtain average time domain waveform data of the inner area of ​​the railway track bed to be detected;

[0026] Performing wavelet transform on the average time-domain waveform data to obtain the transformed average time-domain waveform data;

[0027] The skewness feature, the kurtosis feature and the frequency centroid feature are extracted from the transformed average time domain waveform data to obtain a feature data set.

[0028] In one embodiment, the feature data set includes a skewness feature, a kurtosis feature, and a frequency centroid feature, and determining the moisture coefficient value of the railway track bed to be detected based on the feature data set and a moisture coefficient algorithm includes:

[0029] Data calculation is performed on the skewness feature, the kurtosis feature and the frequency centroid feature according to a water content coefficient algorithm to obtain a water content coefficient value of the railway track bed to be detected.

[0030] In one embodiment, the evaluation criteria include each water coefficient interval and the damage degree corresponding to each water coefficient interval, and determining the water damage assessment result of the railway track bed to be detected based on the evaluation criteria of the railway track bed to be detected and the water coefficient value includes:

[0031] Determining a target moisture coefficient interval within each moisture coefficient interval where the moisture coefficient value is located;

[0032] The damage degree corresponding to the target water coefficient interval is determined as the water damage assessment result of the railway track bed to be detected.

[0033] In a second aspect, the present application further provides a water damage assessment device, comprising:

[0034] An acquisition module is used to acquire the attribute dataset and radar dataset of the railway track bed to be detected;

[0035] a simulation module, configured to simulate various simulated moisture coefficient values ​​of the railway track bed to be detected at various relative dielectric constants according to the attribute data set, and to construct an evaluation standard for the railway track bed to be detected according to the various simulated moisture coefficient values;

[0036] an extraction module, configured to extract waveform features from the radar data set to obtain a feature data set, and determine the moisture coefficient value of the railway track bed to be detected based on the feature data set and a moisture coefficient algorithm;

[0037] The determination module is used to determine the water damage assessment result of the railway track bed to be detected according to the evaluation standard of the railway track bed to be detected and the water content coefficient value.

[0038] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0039] Obtain the attribute dataset and radar dataset of the railway track bed to be detected;

[0040] Simulating various simulated moisture coefficient values ​​of the railway track bed to be detected at various relative dielectric constants according to the attribute data set, and constructing an evaluation standard for the railway track bed to be detected according to the various simulated moisture coefficient values;

[0041] Extracting waveform features from the radar data set to obtain a feature data set, and determining a moisture coefficient value of the railway track bed to be detected based on the feature data set and a moisture coefficient algorithm;

[0042] The water damage assessment result of the railway track bed to be detected is determined according to the evaluation standard of the railway track bed to be detected and the water content coefficient value.

[0043] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0044] Obtain the attribute dataset and radar dataset of the railway track bed to be detected;

[0045] Simulating various simulated moisture coefficient values ​​of the railway track bed to be detected at various relative dielectric constants according to the attribute data set, and constructing an evaluation standard for the railway track bed to be detected according to the various simulated moisture coefficient values;

[0046] Extracting waveform features from the radar data set to obtain a feature data set, and determining a moisture coefficient value of the railway track bed to be detected based on the feature data set and a moisture coefficient algorithm;

[0047] The water damage assessment result of the railway track bed to be detected is determined according to the evaluation standard of the railway track bed to be detected and the water content coefficient value.

[0048] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the following steps:

[0049] Obtain the attribute dataset and radar dataset of the railway track bed to be detected;

[0050] Simulating various simulated moisture coefficient values ​​of the railway track bed to be detected at various relative dielectric constants according to the attribute data set, and constructing an evaluation standard for the railway track bed to be detected according to the various simulated moisture coefficient values;

[0051] Extracting waveform features from the radar data set to obtain a feature data set, and determining a moisture coefficient value of the railway track bed to be detected based on the feature data set and a moisture coefficient algorithm;

[0052] The water damage assessment result of the railway track bed to be detected is determined according to the evaluation standard of the railway track bed to be detected and the water content coefficient value.

[0053] The aforementioned water damage assessment method, apparatus, computer device, computer-readable storage medium, and computer program product acquire a radar dataset of a railway trackbed to be inspected and, based on the radar dataset, determine time-domain waveform data of the internal region of the trackbed to be inspected. Feature extraction is performed on the time-domain waveform data to obtain a feature dataset, and the water content value of the trackbed to be inspected is determined based on the feature dataset and a water content algorithm. Finally, a water damage assessment result for the trackbed to be inspected is determined based on the evaluation criteria for the trackbed to be inspected and the water content value. This method simulates simulated water content values ​​of the trackbed to be inspected at various relative permittivities using an attribute dataset, and constructs evaluation criteria for the trackbed to be inspected based on each simulated water content value, thus achieving automated and objective evaluation criteria. Subsequently, a feature dataset is extracted from the trackbed to be inspected, and a water content value representing the water content of the trackbed to be inspected is determined based on the water content algorithm and the data feature dataset. Then, the water damage assessment result of the railway track bed to be detected was determined by the water content coefficient value and objective evaluation criteria, realizing fully automated water damage assessment of the railway track bed to be detected, avoiding manual participation and improving the accuracy of the water damage assessment method. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments of the present application or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying any creative work.

[0055] Figure 1 1 is a flow chart of a water damage assessment method according to an embodiment;

[0056] Figure 2 A schematic diagram of a process for constructing evaluation criteria in one embodiment;

[0057] Figure 3 is a schematic diagram of a concentric circle trackbed model in an exemplary embodiment;

[0058] Figure 4 A schematic diagram of a process for simulating a radar data set in one embodiment;

[0059] Figure 5 is a schematic diagram of a model-time domain model-radar map in an exemplary embodiment;

[0060] Figure 6 A schematic flow chart of the steps for determining a simulated water content value in one embodiment;

[0061] Figure 7 A time-domain single-track waveform comparison diagram of five groups of internal areas of the corrugated track bed in an exemplary embodiment;

[0062] Figure 8 is a schematic diagram of averaged analog waveform data after wavelet transformation in an exemplary embodiment;

[0063] Figure 9 is a schematic diagram of a wavelet basis function in an exemplary embodiment;

[0064] Figure 10 is an energy distribution diagram of five sets of average simulation waveform data at different frequencies in an exemplary embodiment;

[0065] Figure 11 is a schematic diagram of power spectrum density of five sets of averaged simulation waveform data in an exemplary embodiment;

[0066] Figure 12 is a line graph of simulated water content values ​​in an exemplary embodiment;

[0067] Figure 13 A schematic diagram of a process for extracting a feature data set in one embodiment;

[0068] Figure 14 A schematic diagram of a process for determining a water damage assessment result in one embodiment;

[0069] Figure 15 is a flow chart of an algorithm for determining a water content in an exemplary embodiment;

[0070] Figure 16 is a structural block diagram of a water damage assessment device in one embodiment;

[0071] Figure 17 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0072] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0073] It should be noted that the terms "including" and "having" and any variations thereof used in this application are intended to cover non-exclusive inclusions. The term "plurality" used in this application refers to two or more. The term "and / or" used in this application refers to one or more solutions.

[0074] The trackbed of heavy-haul railways is a crucial component of the track structure. Its mechanical properties and structural stability are directly related to the safety of train operations and the service life of the track system. However, with the continued increase in railway transport loads and the impact of climate change, water damage to the trackbed is becoming an increasingly prominent problem, becoming a technical challenge that urgently needs to be addressed in the railway engineering field.

[0075] Water damage to railway trackbeds refers to the accumulation of water within the trackbed structure due to factors such as rainwater infiltration, rising groundwater levels, or changes in ambient humidity. Water damage can significantly reduce the trackbed's bearing capacity and structural stability, further leading to a series of problems such as roadbed settlement, slope deformation, and track irregularities. These problems not only increase vibration and noise during train operation, threatening driving safety, but also significantly increase the frequency and cost of railway line repairs, seriously affecting the economic benefits and service life of the railway system. Especially on heavy-duty railways, where train loads are high, the track system is more susceptible to water damage, which accelerates the degradation of the trackbed's performance. To address this issue, it is necessary to identify and assess the extent of water damage to the trackbed so that the trackbed can be maintained based on the water damage assessment results to ensure driving safety.

[0076] Traditionally, radar is used to survey the railway trackbed to generate a radar data set, which is then transferred to a computer and displayed on the computer as a radar map. Based on the radar map and their experience, inspectors determine the extent of water damage to the trackbed.

[0077] However, traditional technologies rely on manual determination of water damage extent based on experience and radar datasets, which is subject to many subjective factors. Consequently, current water damage assessment methods have low accuracy.

[0078] Furthermore, existing technologies still lack effective simulation models and comprehensive analysis methods for railway trackbed water damage. On the one hand, current simulation modeling methods often rely on simplified assumptions, making it difficult to truly reflect the distribution of trackbed water damage and its impact on its mechanical properties. On the other hand, existing analysis methods lack specificity and are unable to accurately extract the characteristic signals caused by water damage, limiting the guiding value of simulation models in actual engineering projects.

[0079] To address the above-mentioned issues, the present application provides a water damage assessment method, apparatus, computer device, computer-readable storage medium, and computer program product. This method simulates the simulated water content coefficient values ​​of the railway trackbed to be detected at various relative dielectric constants using an attribute data set, and constructs evaluation criteria for the detected railway trackbed based on each simulated water content coefficient value, thereby achieving automated construction of objective evaluation criteria. Subsequently, by extracting a feature data set from the railway trackbed to be detected and determining a water content coefficient value representing the moisture content of the railway trackbed to be detected based on a water content coefficient algorithm and a data feature data set, the water damage assessment result for the railway trackbed to be detected is then determined using the water content coefficient value and objective evaluation criteria. This achieves fully automated water damage assessment of the railway trackbed to be detected, avoids manual intervention, and improves the accuracy of the water damage assessment method.

[0080] This application also proposes a modeling approach based on a "concentric circle model" to simulate water damage to railway trackbeds in a more realistic manner. This "concentric circle model" is constructed using regular spherical elements. The upper layer simulates the effect of ballast being encased in cement-containing soil, using a "large circle encasing a small circle" pattern. The lower layer, however, uses water-containing blocks encasing the ballast to reflect the water content of the trackbed. This model not only features a simple structure and ease of adjustment, but also effectively amplifies the impact of water damage on the normal operation of the ballast, making the simulation results more representative and generalizable.

[0081] In addition, this application has also developed a complete set of analysis methods in combination with the "concentric circle model". From model construction to electromagnetic forward simulation, and then to wavelet transform to extract the time-frequency characteristics brought by water damage, this application can systematically describe the occurrence and development process of water damage. By extracting the three key indicators of kurtosis, skewness and frequency centroid, a scientific basis is provided for the diagnosis and treatment of water damage. Compared with traditional methods, the water damage assessment method of this application has the advantages of intuitive modeling, efficient calculation and strong adaptability. It can be applied to the maintenance and optimization of heavy-load railway trackbeds, and provide technical guarantees for the safety and economic benefits of railway transportation.

[0082] In one embodiment, Figure 1As shown, a water damage assessment method is provided. This embodiment of the application takes the method applied to a computer device as an example for description. This embodiment of the application does not limit the execution device of a data processing method. The method includes the following steps 102 to 108:

[0083] Step 102: Acquire an attribute dataset and a radar dataset of the railway track bed to be detected.

[0084] The railway track bed to be inspected is a heavy-load railway track bed that requires water damage inspection.

[0085] During implementation, the computer device obtains an attribute dataset of the railway trackbed to be detected, input by a detection user. The attribute dataset contains various attribute data of the railway trackbed to be detected. Simultaneously, the computer device detects the railway trackbed to be detected using a detection radar, thereby obtaining a radar dataset of the railway trackbed to be detected.

[0086] Specifically, the computer device obtains and displays the attribute template in response to the water damage assessment request. The detection user inputs the attribute data set of the railway bed to be detected based on the attribute template. The attribute data set includes the width, length and ballast particle size of the railway bed to be detected. The detection personnel submits the input attribute data set to the computer device. The computer device obtains the user data set input by the detection personnel based on the attribute template. At the same time, the detection radar detects the railway bed to be detected and obtains the radar data set of the railway bed to be detected. Then, the detection radar transmits the radar data set of the railway bed to be detected to the computer device via a network connection or a limited connection, so that the computer device obtains the radar data set of the railway bed to be detected. It should be noted that the frequency and time window of the detection radar need to be the same as the frequency and time window in the subsequent simulation.

[0087] Step 104 , simulating various simulated moisture coefficient values ​​of the railway track bed to be detected at various relative dielectric constants according to the attribute data set, and constructing an evaluation standard for the railway track bed to be detected according to the various simulated moisture coefficient values.

[0088] During implementation, the computer equipment creates concentric circle models of the railway trackbed to be inspected at various relative permittivities based on the attribute datasets. The computer equipment then simulates radar datasets of the railway trackbed to be inspected at various relative permittivities based on each concentric circle model. Based on the radar datasets, the computer equipment determines simulated moisture content values ​​for the railway trackbed to be inspected at each relative permittivity. The computer equipment then determines moisture content intervals based on the simulated moisture content values ​​and constructs evaluation criteria for the railway trackbed to be inspected based on the damage levels corresponding to each moisture content interval and relative permittivity.

[0089] Specifically, the computer device establishes a two-dimensional concentric circle roadbed model of the railway to be detected at various relative permittivity values ​​based on the attribute dataset. Then, for each relative permittivity, the computer device simulates a radar dataset of the railway to be detected at that relative permittivity using the concentric circle roadbed model corresponding to that relative permittivity. The computer device preprocesses the radar dataset and performs feature extraction on the preprocessed radar dataset to obtain a simulated feature set. The computer device then determines a simulated moisture content value for the railway to be detected at that relative permittivity value based on the simulated feature set. The computer device uses each simulated moisture content value as an interval boundary to determine each moisture content interval. The computer device then establishes a correlation between the moisture content interval and the degree of damage based on the relative permittivity corresponding to each moisture content interval. The computer device constructs an evaluation standard based on each moisture content interval and the degree of damage corresponding to that moisture content interval.

[0090] Step 106 , extracting waveform features from the radar data set to obtain a feature data set, and determining the moisture coefficient value of the railway track bed to be detected based on the feature data set and a moisture coefficient algorithm.

[0091] During implementation, the computer preprocesses the radar dataset to obtain a preprocessed radar dataset. The computer then performs a wavelet transform on the preprocessed radar dataset and extracts waveform features from the wavelet-transformed radar dataset to obtain a feature dataset. The computer then performs data operations on the feature dataset using a moisture coefficient algorithm to obtain a moisture coefficient value for the railway trackbed to be inspected.

[0092] Specifically, the radar data set includes waveform data for each initial time-domain waveform. A computer device filters the waveform data for each initial time-domain waveform and averages the filtered waveform data to obtain average time-domain waveform data. The computer device then extracts waveform features from the average time-domain waveform data to obtain a feature data set. The computer device then performs data operations on the feature data set based on a moisture coefficient algorithm to obtain a moisture coefficient value for the railway trackbed to be detected. A higher moisture coefficient value indicates a higher relative dielectric constant and a greater degree of damage to the trackbed to be detected.

[0093] Step 108 : determining a water damage assessment result of the railway track bed to be detected based on the evaluation criteria and the water content coefficient value of the railway track bed to be detected.

[0094] The evaluation criteria include each moisture coefficient interval and the degree of damage corresponding to each moisture coefficient interval.

[0095] During implementation, the computer determines a target moisture coefficient interval based on the moisture coefficient value and determines the damage level corresponding to the target moisture coefficient interval as the target damage level. The computer then determines the target damage level as the water damage assessment result for the railway trackbed to be inspected.

[0096] Specifically, the computer device determines a target moisture coefficient interval within each moisture coefficient interval. The computer device then determines the damage level corresponding to the target moisture coefficient interval as the target damage level, and determines the target damage level as the water damage assessment result for the railway trackbed to be inspected.

[0097] In the aforementioned water damage method, simulated water content values ​​for the trackbed under various relative dielectric constants are simulated using an attribute dataset. Evaluation criteria for the trackbed under investigation are then constructed based on these simulated water content values, achieving automated and objective evaluation criteria. Subsequently, a feature dataset is extracted from the trackbed under investigation, and a water content value representing the water content of the trackbed under investigation is determined based on a water content algorithm and the data feature dataset. The water damage assessment results for the trackbed under investigation are then determined using the water content value and objective evaluation criteria. This fully automated water damage assessment of the trackbed under investigation is achieved, eliminating manual intervention and improving the accuracy of the water damage assessment method.

[0098] In an exemplary embodiment, Figure 2 As shown, the specific processing process of step 104 includes steps 202 to 206. Among them:

[0099] Step 202: establishing a concentric circle trackbed model of the railway trackbed to be detected based on the attribute data set.

[0100] During implementation, the computer device obtains a detection parameter set of the detection radar and then establishes a concentric circle trackbed model of the railway trackbed to be detected based on the attribute data set and the detection parameter set.

[0101] Specifically, the computer device obtains the detection parameter set of the detection radar. This detection parameter set includes the antenna frequency and time window of the detection radar. Then, based on the detection parameter set and the attribute data set, the computer device determines the basic parameter set and specific setting parameter set of the concentric circle roadbed model. The computer device also determines the water content parameter set of the concentric circle roadbed model. Then, based on the basic parameter set, the specific setting parameter set, and the water content parameter set, the computer device establishes a concentric circle roadbed model of the railway trackbed to be detected. To set evaluation criteria for the railway trackbed to be detected, it is necessary to determine the simulated water content coefficient values ​​of the railway trackbed to be detected under different relative dielectric constants. Therefore, the computer device needs to establish concentric circle roadbed models under different conditions. To simplify the model establishment process, only two models are required: a healthy concentric circle roadbed model (a water-free concentric circle roadbed model) and a water-containing concentric circle roadbed model. It should be noted that the concentric circle roadbed model is a concentric circle model. The concentric circle model consists of multiple parallel spheres and a base plate. In the healthy concentric circle roadbed model, the spheres are solid. In the water-containing concentric circle roadbed model, the spheres are special spheres, which are composed of a first sphere with a slightly larger diameter and a second sphere with a slightly smaller diameter. The first sphere and the second sphere have the same center.

[0102] In an exemplary embodiment, Figure 3 FIG. 1 is a schematic diagram of a concentric circle track bed model in an exemplary embodiment. Figure 3As shown, the concentric circle model on the left is a healthy concentric circle model, while the concentric circle model on the right is a water-damaged concentric circle model. The concentric circle model (concentric circle model) is based on the structural and material properties of an actual water-damaged railway trackbed. The concentric circle model measures 1.6m x 1.0m and is generally divided into two layers from top to bottom. The first layer of the healthy concentric circle model consists of two rows of 0.08m diameter spheres. The water-damaged concentric circle model has a 2x5 special sphere in the center. This is a concentric circle model, representing a 0.08m diameter sphere "enveloping" a normal 0.06m diameter sphere. This simulates the situation after water damage, when wet mud and other dirt envelop the previously healthy ballast, affecting the normal operation of the railway trackbed. The second layer of the healthy concentric circle roadbed model is composed of 6 rows of small balls with a smaller diameter (0.04m), which are used to indicate the ballast that has received a certain degree of wear and is located deeper in the roadbed. The water-containing concentric circle roadbed model that simulates water damage also has a 0.56m×0.24m rectangular block in the center position to represent water damage in the deeper layer of the roadbed. This is also in line with the fact that when the ballast is wrapped in water and dirt on the upper layer of the roadbed, the disease in the lower layer of the roadbed is often very serious. The long rectangular block at the bottom of the concentric circle roadbed model is used to provide an accurate reflection surface for the lower surface of the concentric circle roadbed model, so as to provide more accurate simulation conditions and subsequent analysis and processing. The existence of the bottom plate here has no effect on the subsequent simulation results. In the subsequent processing, this application will intercept the simulated radar data set by region, that is, select the data value inside the railway roadbed, and the bottom plate part will not actually participate in the subsequent analysis and calculation. And because the relative dielectric constant of the base plate is selected as 15, the simulation error is almost negligible compared to direct contact with the air. For the sake of more visual results, the long rectangular block is retained here.

[0103] During the modeling of the concentric circle trackbed model, different relative dielectric constants were set for the sections simulating water damage to simulate varying degrees of water damage severity. The relative dielectric constant of the spheres simulating normal ballast was set to 5, a lower bound based on the relative dielectric constant of granite, which ranges from 5 to 8. For the sections simulating water-damaged areas of the railway trackbed (the outer concentric circles in the first layer and the rectangular blocks in the second layer), the relative dielectric constants were set to 30, 50, 70, and 81, respectively, depending on the severity of the damage. The severity of the water damage ranged from mild to extremely severe, providing a comprehensive model foundation for subsequent analysis and processing.

[0104] The concentric circle model of this application simulates the structure of the actual roadbed in terms of geometric dimensions and component composition, and accurately simulates the physical and dielectric properties, providing an effective tool for the detection, maintenance and research of water damage to heavy-load railway roadbeds.

[0105] In one exemplary embodiment, during the modeling process of the concentric circle trackbed model, a computer device sets the basic parameters of the concentric circle trackbed model on a simulation platform based on a basic parameter set, and sets the specific parameters of the concentric circle trackbed model based on a specific parameter set. Specifically, the computer device sets each specific parameter in the specific parameter set as a model parameter in a .in file (a file format), thereby performing corresponding model settings for the concentric circle trackbed model. The computer device then imports the aquifer parameter set into the simulation platform to provide the relative dielectric constant and sphere dimensions for the concentric circle trackbed model. Table 1 is a specific example of a basic parameter set, Table 2 is a specific example of a specific parameter set, and Table 3 is a specific example of a water parameter set.

[0106] Table 1

[0107]

[0108] Table 2

[0109]

[0110] Table 3

[0111]

[0112] Step 204 , simulating a radar data set of the railway track bed to be detected at various relative dielectric constants according to the concentric circle track bed model.

[0113] Among them, the concentric circle roadbed model includes a healthy concentric circle roadbed model and a water-containing concentric circle roadbed model.

[0114] During implementation, the computer device simulates subsets of simulated radar data of the railway trackbed to be inspected in both healthy and wet states using a concentric circle trackbed model and various relative permittivities. The computer device then constructs a simulated radar data set based on the subsets of simulated radar data.

[0115] Specifically, the simulated radar data subsets are divided into two categories based on the status of the railway to be detected: a healthy simulated radar data subset and a water-containing simulated radar data subset. The concentric circle roadbed model includes a healthy roadbed and a water-containing roadbed. Based on the healthy concentric circle roadbed model and the relative dielectric constant corresponding to the healthy concentric circle roadbed model, the computer device simulates the healthy simulated radar data subset for the railway roadbed to be detected in a healthy state. Then, based on the water-containing concentric circle roadbed model and the relative dielectric constants corresponding to the water-containing concentric circle roadbed model, the computer device simulates each water-containing simulated radar data subset for the railway roadbed to be detected in a water-containing state. The computer device constructs a simulated radar data set based on the healthy simulated radar data subset and each water-containing simulated radar data subset.

[0116] Step 206 , determining each simulated water content value corresponding to each relative dielectric constant based on the simulated radar data set, and constructing an evaluation standard for the railway track bed to be detected according to each simulated water content value and the damage degree corresponding to each relative dielectric constant.

[0117] The simulated radar data set includes simulated radar data subsets corresponding to various relative dielectric constants.

[0118] During implementation, the computer preprocesses the simulated radar data subset corresponding to each relative permittivity and determines the simulated water content value corresponding to that relative permittivity based on the simulated radar data subset. The computer then determines water content intervals based on each simulated water content value and constructs an evaluation standard based on the damage level corresponding to each water content interval and relative permittivity.

[0119] Specifically, the computer simulates radar data subsets for each relative dielectric constant based on the water-bearing areas in the concentric circle roadbed model. The selected simulated radar data subsets are averaged to obtain average simulated waveform data. The computer then performs a wavelet transform on the average simulated waveform data and extracts a simulated feature set from the wavelet-transformed average simulated waveform data. The computer then performs data operations on the simulated feature set using a relative dielectric constant coefficient algorithm to obtain simulated water content values ​​corresponding to the relative dielectric constants. The computer uses each simulated water content value as an interval boundary to determine each water content interval. The computer then establishes a correlation between the water content interval and the degree of damage based on the relative dielectric constant corresponding to each water content interval. The computer then constructs an evaluation standard based on each water content interval and the degree of damage corresponding to that water content interval.

[0120] In one exemplary embodiment, the relative permittivity values ​​are 0, 30, 50, 70, and 81. The correlation between the relative permittivity and the degree of damage is as follows: 0 (no water) corresponds to no damage, 0 to 30 corresponds to mild damage, 30 to 50 corresponds to moderate damage, 50 to 70 corresponds to severe damage, 70 to 81 corresponds to extreme damage, and 81 or above corresponds to complete damage. The simulation coefficient values ​​corresponding to the relative permittivity values ​​are shown in Table 4 below:

[0121] Table 4

[0122]

[0123] The computer device uses each simulation coefficient value as the interval boundary to determine each coefficient interval. The coefficient intervals are (0, 8.35×10^-4], (8.35×10^-4, 0.285], (0.285, 0.475], (0.475, 0.828], (0.828, 0.939], (0.939, 1]. Then, the computer device establishes the correlation between each coefficient interval and each damage degree based on each relative dielectric constant. Specifically: (0, 8.35×10^-4] is associated with no damage, (8.35×10^-4, 0.285] is associated with mild damage, (0.285, 0.475] is associated with moderate damage, (0.475, 0.828] is associated with severe damage, (0.828, 0.939] is associated with extreme damage, and (0.939, 1] is associated with complete damage. Then, the computer device constructs an evaluation standard based on each coefficient interval, each damage degree, and the correlation between each coefficient interval and each damage degree.

[0124] In this example, a concentric circle trackbed model was established using the target railway trackbed attribute dataset. Based on this concentric circle trackbed model, precise simulations of its physical and dielectric characteristics were performed, enabling automated generation of a simulated radar dataset. Evaluation criteria for the target railway trackbed were then determined based on the simulated radar dataset. This automated and objective evaluation criteria was constructed, eliminating the need for empirical evaluation and improving the accuracy of the water quality assessment method.

[0125] In an exemplary embodiment, the concentric circle roadbed model includes a healthy concentric circle roadbed model and a water-containing concentric circle roadbed model, such as Figure 4 As shown, the specific processing process of step 204 includes steps 402 to 406. Among them:

[0126] Step 402 : Simulate a healthy simulated radar data subset of the railway track bed to be detected in a healthy state based on a healthy concentric circle track bed model.

[0127] During implementation, the computer equipment simulates the time domain waveform of the railway track bed to be detected in a healthy state based on the healthy concentric circle track bed model and simulation platform to obtain a healthy simulation radar data subset.

[0128] Specifically, the computer system runs a 300-channel gprMax (a ground-penetrating radar 3D full-waveform simulator) simulation based on the healthy concentric circle trackbed model and simulation platform, generating 300 A-scan single-channel time-domain waveforms. These 300 A-scan (the basic scanning method for ground-penetrating radar data processing) single-channel time-domain waveforms represent a subset of the healthy simulated radar data for the target railway. The computer system then uses the "tools.outputfiles_merge" command to obtain a B-scan (the basic scanning method for ground-penetrating radar data processing) image of the target railway trackbed in its healthy state.

[0129] Optionally, the simulation platform may be, but is not limited to, gprMax. This embodiment of the present application does not limit the simulation platform.

[0130] Step 404 , simulating each subset of water-containing simulated radar data of the railway track bed to be detected in a water-containing state based on the water-containing concentric circle track bed model and each relative dielectric constant.

[0131] During implementation, the computer device simulates a subset of water-containing simulated radar data of the railway to be detected in a water-containing state corresponding to each relative dielectric constant based on the water-containing concentric circle roadbed model and each relative dielectric constant.

[0132] Specifically, the computer sets the water parameters for the water-bearing concentric circle trackbed model based on each relative permittivity. Then, using the water-bearing concentric circle trackbed model and the simulation platform, the computer runs a 300-channel gprMax (a three-dimensional full-waveform simulator for ground-penetrating radar) simulation, generating 300 A-scan single-channel time-domain waveforms. These 300 A-scan (the basic scanning method for ground-penetrating radar data processing) single-channel time-domain waveforms represent a subset of simulated radar data for the water-bearing railway under investigation. The computer then uses the "tools.outputfiles_merge" command to obtain a B-scan (the basic scanning method for ground-penetrating radar data processing) image of the water-bearing railway trackbed under investigation.

[0133] In one exemplary embodiment, the relative dielectric constants under water conditions are 30, 50, 70, and 81, respectively. Therefore, based on the water-containing concentric circle trackbed model and the relative dielectric constants, the computer simulates subsets of simulated water-containing radar data for the railway to be inspected under simulated water-containing conditions corresponding to the respective relative dielectric constants, resulting in four sets of simulated water-containing radar data subsets.

[0134] Step 406 : constructing a simulated radar data set based on the healthy simulated radar data subset and each water-containing simulated radar data subset.

[0135] In implementation, the computer device combines the healthy simulated radar data subset and each water-containing simulated radar data subset to obtain a simulated radar data set.

[0136] In one exemplary embodiment, after obtaining the healthy simulated radar data subset and each of the water-containing simulated radar data subsets, the computer generates a healthy B-scan image based on the healthy radar data subset. Simultaneously, the computer generates a water-containing B-scan image based on each of the water-containing simulated radar data subsets. Figure 5 Schematic diagram of model-time domain model-radar map in an exemplary embodiment. Figure 5 As shown in the figure, the waveform on the left is a single-channel waveform in the time domain of a water-free area, while the waveform on the right is a single-channel waveform in the time domain of a water-containing area. The figure in the middle is a B-scan image of the water-containing area.

[0137] In this embodiment, the railway track bed to be detected in the water-containing state and the healthy state is simulated by using a concentric circle track bed model to obtain a radar data set, which facilitates the subsequent generation of evaluation criteria corresponding to the railway track bed to be detected based on the radar data set.

[0138] In an exemplary embodiment, the simulated radar data set includes simulated radar data subsets corresponding to respective relative dielectric constants, such as Figure 6 As shown, the specific processing process of determining the simulated water content values ​​corresponding to the relative dielectric constants based on the simulated radar data set in step 206 includes steps 602 to 606.

[0139] Step 602 : for each simulated radar data subset corresponding to a relative dielectric constant, the simulated radar data subset is screened according to the water-bearing area in the concentric circle roadbed model, and the screened simulated radar data subset is averaged to obtain average simulated waveform data.

[0140] During implementation, the computer device reduces the water-bearing area in the concentric circle roadbed model to obtain a reduced water-bearing area. The computer device then filters the simulated radar data subset corresponding to each relative dielectric constant based on the reduced water-bearing area to obtain a filtered simulated radar data subset. The computer device then averages the filtered simulated radar data subsets to obtain average simulated waveform data.

[0141] Specifically, to reduce the impact of edge effects, the computer device reduces the water-bearing area in the concentric circle roadbed model to obtain a reduced water-bearing area. Optionally, the computer device may also demarcate a water-bearing area within the water-bearing area and, based on the new water-bearing area, determine a new water-free area within the concentric circle roadbed model. The simulated radar data subset contains waveform data for each initial simulated time-domain waveform. For each water-bearing simulated radar data subset, the computer device filters the waveform data for each simulated time-domain waveform from the waveform data for each initial simulated time-domain waveform within the water-bearing simulated radar data subset based on the new water-bearing area (the reduced water-bearing area). For example, if the simulated radar data subset contains waveform data for 300 initial simulated time-domain waveforms, the computer device will obtain waveform data for 160 simulated time-domain waveforms after filtering. The computer device then averages the waveform data for each simulated time-domain waveform to obtain average simulated waveform data.

[0142] In an exemplary embodiment, since the relative dielectric constants are 0, 30, 50, 70, and 81, respectively, the computer device will screen and average the five sets of simulated radar data subsets to obtain five sets of average simulated waveform data. The five sets of average simulated waveform data are five sets of waveforms, such as Figure 7 shown. Figure 7 This is a comparison diagram of time-domain single-track waveforms in the internal area of ​​five groups of corrugated track beds in an exemplary embodiment.

[0143] In an optional embodiment, the simulated radar data subset is based on the ground-penetrating radar echo data obtained by scanning the entire model from left to right. However, for the sake of analysis efficiency and accuracy, due to the influence of edge effects, analyzing the entire model data will have a large error. Therefore, in this application, some preprocessing operations are required before subsequent analysis of the ground-penetrating radar data. The specific steps are as follows:

[0144] Regional Delineation: In the concentric circle model of water-bearing roadbeds, the location of water-bearing areas can be clearly identified. This area corresponds to the roadbed damage area defined in the concentric circle model. Therefore, regional delineation should focus on this water-bearing area. Since both the left and right sides of the water-bearing area touch the unaffected "large ball" area, to reduce the impact of edge effects, the delineated area should be appropriately narrowed to a width slightly narrower than the entire water-bearing area. The computer then filters each simulated radar data subset based on the newly delineated water-bearing area, generating a filtered simulated radar data subset. Furthermore, to ensure the rigor and accuracy of subsequent analysis, a control area of ​​the same width as the water-bearing area should be defined within the unwatered area for subsequent calculations and comparative analysis. To accurately assess the impact of moisture within the roadbed, data from specific sections within the roadbed (e.g., signals within the roadbed depth) are selected to eliminate the influence of irrelevant areas (such as surface or edge effect areas). These steps yield more accurate signals within the roadbed, providing a foundation for subsequent time-frequency analysis.

[0145] Waveform Averaging: Based on the aforementioned regional divisions, water-bearing and water-free areas are clearly distinguished, and each area contains several A-scan waveforms. Using data analysis tools, the waveform data of the n initial simulated time-domain waveforms in the filtered simulated radar data subset are time-domain averaged to obtain a single waveform, the average simulated waveform data. This single waveform effectively reflects the typical characteristics of the region and is therefore selected as a representative for subsequent comparative analysis to improve the convenience and operability of data processing. In radar signal processing, the average signal within a specific time window generally represents the typical waveform within that time window, reflecting the impact of different water contents on radar wave propagation characteristics. By averaging the data under each water content condition (water-containing model data with relative dielectric constants of 30, 50, 70, and 81), random errors in individual measurements can be eliminated and more representative signal characteristics can be extracted.

[0146] In an optional embodiment, B-scan images play an irreplaceable role in analyzing overall target characteristics. Therefore, a comprehensive analysis combining the "model, time-domain waveform, and radar map" is essential. The computer performs a spatiotemporal coordinate analysis on each B-scan image, preliminarily determining the location of the water-damaged area within the map and annotating it to provide a reference for subsequent analysis. During B-scan map processing, the computer, using data analysis tools, also performs a blank calibration operation. This operation involves creating a blank area of ​​the same size as the concentric circle roadbed model to be analyzed and running a simulation with the same number of tracks as in the actual situation, thereby obtaining the echo signal from the ground-penetrating radar antenna into the blank area. The computer then performs a differential analysis of the resulting "concentric circle model" ground-penetrating radar data matrix with this echo signal, effectively removing interference from non-target signals such as noise and direct waves. This process significantly improves the quality of the B-scan images and is crucial for subsequent analysis. Taking the case of extreme water content as an example, the coordinated analysis of the model, the time domain single channel average waveform and the B-scan spectrum can generate the following Figure 5 The image shown is of great reference value for feature extraction and subsequent analysis of electromagnetic forward modeling results.

[0147] Step 604 : performing wavelet transform on the average analog waveform data, and extracting a simulation feature set of the average analog waveform data after the wavelet transform.

[0148] During implementation, the computer device performs wavelet transform processing on the average analog waveform data to obtain the average analog waveform data after wavelet transform. Then, the computer device extracts the simulated skewness characteristics, simulated kurtosis characteristics and simulated frequency centroid characteristics of the average analog waveform data after wavelet transform to obtain a simulation feature set.

[0149] Specifically, the computer device performs wavelet transform processing on each averaged analog waveform data. Through the wavelet transform processing, the frequency components of the signal are effectively extracted to obtain the time-varying patterns of the averaged analog waveform data after the wavelet transform. The computer device extracts the simulated skewness characteristics of the averaged analog waveform data after the wavelet transform using a skewness algorithm, and extracts the simulated kurtosis characteristics of the averaged analog waveform data after the wavelet transform using a kurtosis algorithm. Then, the computer device extracts the simulated frequency centroid characteristics of the averaged analog waveform data after the wavelet transform using a frequency centroid algorithm.

[0150] In one exemplary embodiment, the relative dielectric constants are 0, 30, 50, 70, and 81, respectively. The computer device then contains five sets of averaged simulated waveform data. Wavelet transform is a time-frequency analysis method that simultaneously provides both time and frequency information about a signal. Unlike traditional Fourier transforms, wavelet transforms have superior time-frequency localization properties and are suitable for analyzing instantaneous frequency characteristics in time-varying signals. In this application, wavelet transforms are used to analyze radar signals under different water content conditions, effectively extracting how the signal's frequency components vary over time.

[0151] By performing wavelet transform based on Morlet wavelet basis on each group of average analog waveform data, the energy distribution of these signals in the time-frequency domain is obtained. Figure 8 A schematic diagram of averaged simulated waveform data after wavelet transformation in an exemplary embodiment. This analysis helps reveal the specific impact of moisture content on the frequency and propagation characteristics of ground-penetrating radar waveforms, such as whether the signal's frequency distribution shifts or whether frequency components become more concentrated or expanded.

[0152] like Figure 8 The figure shows the time-frequency analysis results for different moisture contents (from zero to extremely watery). The wavelet transform is used to visualize the frequency and time distribution characteristics. In the absence of water, the frequency range is primarily concentrated between 1 and 3 GHz, and the signal energy distribution is relatively flat. The signal intensity fluctuates significantly over time. As the moisture content increases, the frequency distribution gradually shifts toward higher frequencies. At a moisture content of 50%, the signal's frequency components significantly expand, and the signal intensity increases, primarily concentrating between 2 and 4 GHz. At a moisture content of 81%, the frequency components further broaden, particularly between 2 and 5 GHz. The signal's intensity and temporal distribution become more complex, revealing more high-frequency details. Overall, increasing moisture content leads to an upward shift in the frequency distribution and a concentration of energy. In particular, under high moisture conditions, the signal's frequency characteristics undergo significant changes, indicating that moisture significantly affects the propagation characteristics of ground penetrating radar signals.

[0153] The Morlet wavelet basis function is shown in the following formula (1):

[0154] (1)

[0155] In the above formula (1), is the independent variable, is the exponential constant, is a trigonometric function. is the wavelet basis function. Figure 9 Schematic diagram of wavelet basis functions in an exemplary embodiment.

[0156] The reasons for choosing this wavelet basis are:

[0157] ① Good time-frequency localization: The Morlet wavelet is a Gaussian function modulated by a complex exponential. Its expression contains a Gaussian window and sine (or cosine) wave components. This gives the Morlet wavelet good localization performance in both the time and frequency domains, making it suitable for capturing instantaneous changes in signals and features at different scales.

[0158] ② Applicable to continuous wavelet transform: Wavelet basis is particularly suitable for non-stationary signal analysis and can accurately locate the distribution of different frequency components of the signal in time. Continuous wavelet transform can achieve smooth time-frequency representation through Morlet wavelet.

[0159] ③ Advantages of combining with Fourier transform: The frequency domain characteristics of Morlet wavelet are close to the Fourier basis, making it easier to understand and apply when performing frequency analysis. Unlike pure Fourier transform, Morlet wavelet can also provide time resolution, making it suitable for analyzing non-stationary signals.

[0160] ④ The results of the wavelet transform may contain some noise and detail fluctuations, so smoothing is needed to eliminate these high-frequency noises and highlight the main signal features. Smoothing the wavelet transform results by moving average can reduce local fluctuations and obtain more stable time-frequency characteristics.

[0161] Furthermore, considering the impact of frequency range on signal analysis, this application limits the frequency range to 0 to 5 GHz. This is because the simulated antenna frequency is set to 1.5 GHz, which is consistent with the frequency range of interest in actual ground-penetrating radar applications. Frequencies outside this range contribute less to target detection. Therefore, by limiting the frequency range, we can more accurately analyze the main frequency components of the signal and avoid unnecessary frequency interference.

[0162] Then, the computer device extracts the simulated skewness feature of the average analog waveform data after the simulated wavelet transform according to the skewness algorithm. And extracts the simulated kurtosis feature of the average analog waveform data after the wavelet transform according to the kurtosis algorithm. Then, the computer device extracts the simulated frequency centroid feature of the average analog waveform data after the wavelet transform according to the frequency centroid algorithm. The simulated skewness feature of the average analog waveform data is the skewness of the average analog waveform. The simulated kurtosis feature of the average analog waveform data is the kurtosis of the average analog waveform, and the simulated frequency centroid feature of the average analog waveform data is the frequency centroid of the average analog waveform. Specifically:

[0163] Skewness is a parameter that measures the symmetry of the data distribution, especially whether there is a shift in the data distribution. Specifically, the skewness of the wavelet coefficients can help determine the symmetry of the signal, whether there are some extreme outliers or whether the signal is stronger in certain areas. For example, if the skewness of the water signal is greater than that of the water-free signal, it may indicate that some parts of the water signal are more biased towards high values. A higher skewness value may indicate that the electromagnetic wave is more strongly reflected or attenuated, which may be related to the increase in water content. The skewness algorithm is shown in the following formula (2):

[0164] (2)

[0165] Among them, in the above formula (2), is the data point in the averaged simulated waveform data, is the mean value of the averaged simulated waveform data, is the standard deviation of the average simulated waveform data. is the total number of averaged simulation waveform data.

[0166] Kurtosis is a parameter that measures the kurtosis of the data distribution, that is, whether the tail of the distribution is heavier or lighter. The kurtosis of the wavelet coefficient can help understand the concentration of the signal and whether the energy of certain frequency components is concentrated in certain frequency bands. In the case of high water content, the signal may show a higher kurtosis, indicating that the energy of the electromagnetic wave is concentrated in certain frequency bands, which may be closely related to the moisture status and disease characteristics of the roadbed. The kurtosis algorithm is shown in the following formula (3):

[0167] (3)

[0168] In the above formula (3), is the data point in the averaged simulated waveform data, is the mean value of the averaged simulated waveform data, is the standard deviation of the average simulated waveform data. is the total number of averaged simulation waveform data.

[0169] Frequency centroid: This parameter is used to calculate the center of the signal's energy distribution in the frequency domain. It is used to characterize the overall trend of frequency offset under the influence of moisture content. The frequency centroid algorithm is shown in the following formula (4):

[0170] (4)

[0171] In the above formula (4), is the frequency, is the energy of the corresponding frequency.

[0172] Table 5 is a schematic table of various simulation feature sets in an exemplary embodiment.

[0173] Table 5

[0174]

[0175] From the results shown in the table, it can be seen that the railway trackbed under different moisture content conditions (different relative dielectric constants) shows significant differences in skewness, kurtosis and frequency centroid, each of which has the following characteristics:

[0176] Simulated skewness characteristics: An increase in skewness indicates a decrease in the symmetry of the signal distribution, possibly indicating the presence of extreme values ​​or abnormal signal enhancement in certain areas. The data shows that as the water content (relative dielectric constant) increases, the skewness value gradually increases from 2.6713 in the water-free model to 4.4785 in the extremely water-rich (81) model. This indicates that the reflection or attenuation of water-rich signals in high-value areas is more significant, demonstrating the severity of the impact of moisture on the roadbed structure.

[0177] Simulated kurtosis characteristics: An increase in kurtosis indicates a higher degree of skewness in the signal distribution, meaning that the tail of the distribution becomes heavier. The data shows that the kurtosis value increases significantly with increasing moisture content, from 10.5126 in the model without moisture to 27.1406 in the model with extreme moisture (81). This indicates that under high moisture content conditions, the signal energy is more concentrated in certain frequency bands, which is closely related to the moisture status and disease characteristics of the roadbed.

[0178] Simulated frequency centroid characteristics: The upward shift of the frequency centroid indicates that the distribution center of the signal energy in the frequency domain shifts toward higher frequencies. The data shows that the frequency centroid gradually increases from 1.7252 GHz in the water-free model to 1.8243 GHz in the extremely water-containing (81) model. This trend indicates that as the moisture content increases, the frequency distribution center of the signal shifts toward higher frequencies, which may be related to the effect of moisture on the propagation characteristics of electromagnetic waves. This trend indicates that the change in the frequency centroid can be used to monitor and evaluate the moisture changes in the roadbed.

[0179] In summary, the changes in these parameters can more accurately identify and quantify the moisture content in the trackbed, providing a scientific basis for railway maintenance and disease prevention. Furthermore, the calculation results of these statistical parameters provide a quantitative method for detecting water damage in the trackbed, which can be used to assess and monitor the moisture status and disease characteristics of heavy-load railway trackbeds.

[0180] In an optional embodiment, in order to calculate the corresponding sampling frequency from the original signal, the computer device determines a time interval based on the number of sampling points and the total time length, and determines the sampling frequency based on this time interval. The optimal sampling rate setting value is calculated by the following formula (5):

[0181] (5)

[0182] In the above formula (5), is the sampling frequency in Hz, is the number of data points, which depends on the simulation parameter settings of the simulation platform and can be regarded as a constant here. It can be easily obtained by looking at the imported simulated radar dataset; is the total time, i.e., the time window size set during simulation on the simulation platform, in nanoseconds. An accurate sampling frequency is the basis for subsequent time-frequency analysis, wavelet transform, and power spectral density (PSD) analysis, as the sampling frequency determines the frequency range and spectral resolution of the signal.

[0183] Secondly, because the data for water-damaged and undamaged roadbeds will inevitably have different numbers of sampling points, a zero-padding operation is performed to ensure that they have the same length and facilitate subsequent processing. Zero-padding supplements the end of the average simulated waveform data corresponding to the healthy roadbed with the required number of points, thereby ensuring that the signals under different conditions have a uniform time resolution and length during the processing process. In this operation, the total number of zero padding is selected to be a power of 2 greater than the signal sampling points (for the example of extremely water-damaged conditions, 2^12 = 4096 is selected), which helps to improve the speed of computer equipment when performing FFT operations.

[0184] The computer extracts the energy distribution and power spectral density characteristics of the averaged simulated waveform data and uses these characteristics to analyze the damage to the railway trackbed under different relative dielectric constants. Specifically, the computer calculates the energy distribution of different frequency components to obtain the energy characteristics of each water-containing concentric circular trackbed model at different frequencies. This is the most direct and critical step in analyzing GPR signal data in the frequency domain.

[0185] Because the wavelet coefficients provide information in three dimensions simultaneously: time, frequency, and amplitude, the energy of each frequency component can be integrated based on the above wavelet transform results to obtain the total energy distribution of the signal in the 0 to 5 GHz frequency band. Figure 10 : is an energy distribution diagram of five sets of average simulation waveform data at different frequencies in an exemplary embodiment. Figure 10This figure shows the energy distribution of five different sets of averaged simulated waveform data within the frequency range of 0 to 5 GHz. The horizontal axis represents frequency (GHz), and the vertical axis represents energy value. The energy distribution curve for the averaged simulated waveform data corresponding to the healthy concentric circle roadbed model in the figure peaks at approximately 1.5 GHz and then rapidly decreases with increasing frequency. This indicates that in the absence of water, energy is primarily concentrated in the lower frequency range. The energy distribution curve for the averaged simulated waveform data of the water-containing concentric circle roadbed model shows that the energy peak shifts to higher frequencies and the peak energy increases with increasing relative permittivity. For example, the energy curve for the water-containing concentric circle roadbed model with a relative permittivity of 30 peaks at approximately 2 GHz, while the energy curve for the water-containing concentric circle roadbed model with a relative permittivity of 81 peaks at around 2.5 GHz. This indicates that as water damage increases, the peak frequency of the energy distribution increases and the energy distribution curve becomes broader, indicating that energy is distributed over a wider frequency range. This analysis helps reveal the impact of moisture content changes on the spectral characteristics of radar signals, such as whether increased moisture content causes the spectrum to shift toward low or high frequencies, or whether the energy distribution of the signal changes.

[0186] In addition, the dominant frequency analysis can show the temporal changes in the dominant frequency of the signal, which reflects the effect of moisture content on the propagation speed and attenuation characteristics of radar waves. Under different concentric circle roadbed models, as the moisture content increases, the propagation characteristics of the radar waves change, resulting in changes in the dominant frequency.

[0187] Power spectral density analysis is also crucial in ground-penetrating radar signal processing. It reveals the distribution characteristics of signal energy at different frequencies, helping to distinguish target signals from noise and optimize target detection accuracy. Furthermore, power spectral density analysis is used to study the impact of changes in the dielectric properties of underground media on the signal spectrum, providing a basis for inverting underground structure and material properties. It also helps optimize detection depth and resolution, improving detection effectiveness and data processing quality.

[0188] The computer equipment calculates the power spectral density of the averaged analog waveform data using the Welch method (a modern spectral analysis technique for estimating signal power spectral density). The PSD (Power Spectral Density) results and corresponding frequencies of the averaged analog waveform data are extracted using the pwelch function (a tool for quickly estimating signal power spectral density). The frequency range is limited to 0-5 GHz, and PSD results with amplitudes above -100 dB (decibels) are considered valid results. The frequency range is then re-extracted based on the set limits. Figure 11 FIG. 4 is a schematic diagram of power spectrum density of five sets of averaged simulation waveform data in an exemplary embodiment.

[0189] like Figure 11 As shown in the figure, the power spectrum density of five sets of average simulation waveform data in the frequency range of 0 to 5 GHz is shown. The horizontal axis also represents frequency (GHz) and the vertical axis represents power spectrum density (dB / Hz, decibel / Hertz). Figure 11 As can be seen, the power spectral density of the average simulated waveform data for the healthy concentric circle roadbed model reaches its maximum value near 1.5 GHz and then rapidly decreases, indicating that in the absence of water, the signal power is primarily concentrated at lower frequencies. The power spectral density curve for the average simulated waveform data for the water-damaged concentric circle roadbed model similarly shows that the peak of the power spectral density shifts to higher frequencies as the relative permittivity increases. The power spectral density curve for the water-damaged concentric circle roadbed model with a relative permittivity of 30 reaches its peak at approximately 2 GHz, while the power spectral density curve for the water-damaged concentric circle roadbed model with a relative permittivity of 81 reaches its peak at approximately 2.5 GHz. Furthermore, the power spectral density curve for the water-damaged concentric circle roadbed model decreases more rapidly after the peak, indicating that in cases of severe water damage, the signal attenuation in the high-frequency band is more significant.

[0190] Figure 10 and Figure 11 The analysis of is of great significance for ground penetrating radar data signal processing. First, Figure 10 and Figure 11 The study revealed the impact of water damage on the signal’s frequency distribution and power spectral density, helping to identify and quantify the moisture content in the roadbed. By analyzing the signal’s frequency response and power spectral density, the water damage condition of the roadbed can be more accurately assessed, providing a scientific basis for maintenance and repair work.

[0191] Step 606 , performing data operations on the simulation feature set according to the water content coefficient algorithm to obtain a simulated water content coefficient value corresponding to the relative dielectric constant.

[0192] The simulated feature set includes a simulated skewness feature, a simulated kurtosis feature, and a simulated frequency centroid feature.

[0193] During implementation, the computer device performs data operations on the simulated skewness characteristics, simulated kurtosis characteristics and simulated frequency centroid characteristics according to the water coefficient algorithm to obtain the simulated water coefficient value corresponding to the relative dielectric constant.

[0194] Specifically, the computer device constructs a water content coefficient calculation formula based on the sigmoid function form. The basic form of the sigmoid function is shown in the following formula (6):

[0195] (6)

[0196] In the above formula (6), is the independent variable of the sigmoid function, is an exponential constant. Then, the computer device constructs a water content coefficient algorithm based on the three characterization values ​​of the sigmoid function, the simulated skewness characteristic, the simulated kurtosis characteristic, and the simulated frequency centroid characteristic. The water content coefficient algorithm is shown in the following formula (7):

[0197] (7)

[0198] In the above formula (7), To simulate the frequency centroid characteristics, To simulate the skewness characteristic, The center frequency is usually the frequency of the ground penetrating radar's transmitting and receiving antenna. To simulate the kurtosis characteristic. is the simulated water content value.

[0199] The computer device performs data operations on the simulated skewness feature, the simulated kurtosis feature and the simulated frequency centroid feature in each simulated feature set according to the water coefficient algorithm to obtain the simulated water coefficient value corresponding to the relative dielectric constant.

[0200] In an exemplary embodiment, the computer device contains a set of simulated features with relative dielectric constants of 0, 30, 50, 70, and 81. The reason and necessity for proposing the water content coefficient algorithm is that the three characterization values ​​of skewness, kurtosis, and frequency centroid well describe the differences in the wavelet transform results of different water damage models. However, in order to guide the actual disease situation, an analytical method is still needed to specifically quantify the water content. Similarly, using such a water content coefficient algorithm to specifically calculate the water content coefficient of the model is also conducive to more flexible use and scheduling by relevant professionals, providing convenience for specific use. The computer device constructs a water content coefficient algorithm based on the three characterization values ​​of sigmoid function, simulated skewness feature, simulated kurtosis feature, and simulated frequency centroid feature. The water content coefficient algorithm is shown in the above formula (7). The sigmoid function form is selected because the value range of the sigmoid function is (0,1), which is an ideal range for the calculation result of the evaluation coefficient, and the function graph is smooth, which can also have a good effect when there are many eigenvalues. Therefore, this form is also used here. The computer device performs data operations on the simulated feature sets with relative permittivities of 0, 30, 50, 70, and 81 using the water content algorithm to obtain simulated water content values ​​corresponding to the relative permittivities. Table 6 is a schematic table of simulated water content values ​​in an exemplary embodiment.

[0201] Table 6

[0202]

[0203] The computer equipment draws a line graph of the simulated water content coefficient value according to Table 6 above, and obtains Figure 12 . Figure 12 FIG. 4 is a line graph of simulated water content values ​​in an exemplary embodiment.

[0204] Table 6 and Figure 12 The following analysis results were obtained:

[0205] Positive correlation between water content coefficient and severity of water damage: Figure 12 The figure shows that as the relative permittivity of water damage increases (from no water to extremely water-rich with a relative permittivity of 81), the corresponding simulated water content coefficient values ​​show a monotonically increasing trend. This indicates that there is a strong positive correlation between the severity of water damage (represented by the relative permittivity) and the water content coefficient.

[0206] The changing trend of the water-containing model: Water-containing (30): The corresponding relative dielectric constant is 30 (the subsequent data all indicate this meaning and will not be repeated here), and its simulated water coefficient value is 0.285. This value is significantly higher than the non-water-containing model, indicating that mild water damage has a significant impact on the dielectric properties of the roadbed. Water-containing (50): Its simulated water coefficient value further increases to 0.475, an increase of about 67% compared with the value of water-containing (30). This shows that when the degree of water damage intensifies, the growth rate of the simulated water coefficient value also accelerates. Water-containing (70): Its simulated water coefficient value is 0.828. The increase in the simulated water coefficient value of water-containing (70) is further increased than the previous two levels, which shows that in the higher relative dielectric constant range, the severity of water damage has a more significant impact on the simulated water coefficient value. Water-containing (81): Its simulated water coefficient value reaches 0.939, close to the saturation level of 1. This indicates that under extremely high water damage conditions, the simulated water content of the concentric circle roadbed model approaches a certain limit, and the change in the water content gradually stabilizes. This may be because the concentric circle roadbed model approaches saturation under high water damage conditions, reflecting the nonlinear growth trend of the simulated water content.

[0207] Physical Meaning: The relative dielectric constant of water damage directly reflects the changes in the dielectric properties of the roadbed due to moisture. A larger value indicates a greater severity of the damage. The simulated water content coefficient value is a characteristic parameter calculated using the water content coefficient algorithm. This parameter maps different relative dielectric constants into quantitative indicators of water damage, providing a direct reflection of the severity of the damage. The gradual increase in the simulated water content coefficient value from zero to extremely water-rich conditions demonstrates the sensitivity and accuracy of the water content coefficient algorithm in distinguishing the severity of water damage, particularly for medium to high severity water damage.

[0208] In summary, the positive correlation between the simulated water content coefficient and the relative dielectric constant indicates that as water damage intensifies (increases in dielectric constant), the simulated water content coefficient increases significantly, demonstrating the reliability of the water content coefficient algorithm in water damage monitoring. For water damage with extremely high dielectric constants, the water content coefficient approaches saturation, indicating that the formula has good stability and limit value constraints, providing a theoretical basis and guidance for engineering disease diagnosis.

[0209] In this embodiment, by performing a wavelet transform on a simulated radar dataset and extracting a simulated feature dataset from the wavelet-transformed radar dataset, the characteristic performance of the railway trackbed under different water content conditions was obtained. Simulated water coefficient values ​​were then determined using the simulated feature dataset and a water coefficient algorithm. Evaluation criteria were then constructed based on the simulated water coefficient values, and a quantitative assessment method for water damage to heavy-duty railway trackbeds was established, providing a scientific basis for the diagnosis and treatment of water damage. Compared to traditional methods, the simulation model and analysis method proposed in this application have the advantages of intuitive modeling, efficient computation, and strong adaptability. They can be applied to the maintenance and optimization of heavy-duty railway trackbeds, providing technical support for the safety and economic benefits of railway transportation.

[0210] In an exemplary embodiment, Figure 13 As shown, the specific process of extracting waveform features from the radar data set in step 106 to obtain the feature data set includes steps 1302 to 1308. Among them:

[0211] Step 1302 : Filter the waveform data of each initial time domain waveform in the radar data set to obtain the waveform data of each time domain waveform.

[0212] The radar data set includes waveform data of each initial time domain waveform.

[0213] During implementation, the computer generates a radar map based on the radar data set and divides the radar map into water-bearing areas and non-water-bearing areas. The computer screens the waveform data of each initial time-domain waveform based on the water-bearing areas to obtain waveform data of each time-domain waveform.

[0214] Specifically, since the radar dataset consists of waveform data from multiple A-scan waveforms, which are time-domain waveforms, the computer generates a B-scan map (graph) of the railway trackbed to be inspected based on the radar dataset. The computer then divides the B-scan map into water-bearing and non-water-bearing areas. Based on the water-bearing areas, the computer selects the waveform data of the time-domain waveforms of the water-bearing areas from the waveform data of each initial time-domain waveform.

[0215] Step 1304 , performing averaging processing on the waveform data of each time domain waveform to obtain average time domain waveform data of the inner area of ​​the railway track bed to be detected.

[0216] During implementation, the computer device averages the waveform data of each time domain waveform to obtain average time domain waveform data of the average time domain waveform of the internal area of ​​the railway track bed to be detected. The computer device averages the waveform data of each time domain waveform at the same sampling time, thereby integrating the time domain waveforms into an average time domain waveform, and obtaining average time domain waveform data of the average time domain waveform.

[0217] Step 1306: Perform wavelet transform on the average time-domain waveform data to obtain transformed average time-domain waveform data.

[0218] During implementation, the computer device performs wavelet transform processing on the average time domain waveform data, thereby extracting the time-varying pattern of the frequency components of the radar signal in the average time domain waveform, and obtaining the transformed average time domain waveform data.

[0219] Specifically, the computer device performs a wavelet transform based on a Morlet wavelet basis on the average time domain waveform data to obtain the energy distribution of the radar signal in the time-frequency domain in the average time domain waveform, that is, the transformed average time domain waveform data.

[0220] Step 1308 : extracting the skewness feature, the kurtosis feature, and the frequency centroid feature from the transformed average time-domain waveform data to obtain a feature data set.

[0221] In implementation, the computer device extracts features from the transformed average time-domain waveform data according to a skewness algorithm to obtain a skewness feature, and extracts features from the transformed average time-domain waveform data according to a kurtosis algorithm to obtain a kurtosis feature. Then, the computer device extracts features from the transformed average time-domain waveform data according to a frequency centroid algorithm to obtain a frequency centroid feature. The skewness algorithm is shown in the above formula (2), the kurtosis algorithm is shown in the above formula (3), and the frequency centroid algorithm is shown in the above formula (4).

[0222] In this embodiment, by performing wavelet transform and feature extraction on the radar data set, a feature set of the current railway track bed to be detected is obtained, which facilitates the subsequent determination of the water damage degree of the railway track bed to be detected based on the evaluation criteria.

[0223] In an exemplary embodiment, the feature data set includes a skewness feature, a kurtosis feature, and a frequency centroid feature. The specific processing of determining the moisture coefficient value of the railway track bed to be detected based on the feature data set and the moisture coefficient algorithm in step 106 includes:

[0224] According to the moisture coefficient algorithm, data calculation is performed on the skewness feature, kurtosis feature and frequency centroid feature to obtain the moisture coefficient value of the railway track bed to be detected.

[0225] During implementation, the computer equipment performs data operations on the skewness characteristics, kurtosis characteristics, and frequency centroid characteristics of the railway track bed to be detected based on the water content coefficient algorithm to obtain the water content coefficient value of the railway track bed to be detected. The water content coefficient algorithm is shown in the following formula (8):

[0226] (8)

[0227] In the above formula (8), is the frequency centroid feature, is the skewness characteristic, The center frequency is usually the frequency of the detection radar's transmitting and receiving antenna. is the kurtosis characteristic. is the moisture coefficient value.

[0228] In this embodiment, the moisture coefficient value is determined by using a moisture coefficient algorithm and a characteristic data set. The moisture coefficient value can intuitively reflect the current degree of water damage to the railway to be detected, facilitating the subsequent determination of the degree of water damage to the railway track bed to be detected based on the evaluation criteria. This realizes the automated and precise assessment of the degree of water damage to the railway track bed to be detected, thereby improving the efficiency of the water damage assessment method.

[0229] In an exemplary embodiment, the evaluation criteria include each moisture coefficient interval and the damage degree corresponding to each moisture coefficient interval, such as Figure 14 As shown, the specific processing process of step 108 includes steps 1402 to 1404. Among them:

[0230] Step 1402: Determine the target moisture coefficient interval in each moisture coefficient interval.

[0231] During implementation, the computer device determines the moisture coefficient interval in which the moisture coefficient value is located in each moisture coefficient interval, and determines the moisture coefficient interval as the target moisture coefficient interval.

[0232] In an exemplary embodiment, the moisture coefficient intervals are (0, 8.35×10^-4], (8.35×10^-4, 0.285], (0.285, 0.475], (0.475, 0.828], (0.828, 0.939], and (0.939, 1]. The moisture coefficient value is 0.3. The computer device determines that (0.285, 0.475] where 0.3 is located is the target moisture coefficient interval.

[0233] Step 1404 : Determine the damage degree corresponding to the target water content coefficient interval as the water damage assessment result of the railway track bed to be detected.

[0234] During implementation, the computer device determines the damage degree corresponding to the target water content coefficient interval as the target damage degree. Then, the computer device determines the target damage degree as the water damage assessment result of the railway track bed to be inspected.

[0235] In an exemplary embodiment, (0, 8.35×10^-4] corresponds to no damage, (8.35×10^-4, 0.285] corresponds to mild damage, (0.285, 0.475] corresponds to moderate damage, (0.475, 0.828] corresponds to severe damage, (0.828, 0.939] corresponds to extreme damage, and (0.939, 1] corresponds to complete damage. The target water content coefficient interval is (0.285, 0.475]. The computer device determines that the moderate damage corresponding to (0.285, 0.475] is determined as the water damage assessment result of the railway track bed to be detected.

[0236] In this embodiment, the water damage assessment result of the railway track bed to be detected is determined by the water coefficient value and objective evaluation criteria, thereby realizing a fully automated water damage assessment of the railway track bed to be detected, avoiding manual participation, and improving the accuracy of the water damage assessment method.

[0237] In an exemplary embodiment, Figure 15 FIG. 1 is a flow chart of an algorithm for determining the water content in an exemplary embodiment. Figure 15 As shown, the steps of the algorithm for determining the water content include:

[0238] Step 1501 : establishing a concentric circle roadbed model based on the attribute data set of the railway roadbed to be detected, and simulating a simulated radar data set of the railway roadbed to be detected under various relative dielectric constants based on the concentric circle roadbed model.

[0239] Step 1502 : pre-process and air-calibrate the simulated radar data set, and perform wavelet transform on the simulated radar data set after air-calibration to extract time-frequency features, thereby obtaining a simulated radar data set after wavelet transform.

[0240] Step 1503 : Analyze the frequency shift characteristics caused by different water content states of the simulated radar data set after wavelet transformation.

[0241] Step 1504 , extracting features from the simulated radar data set after wavelet transformation to obtain simulated skewness features, simulated kurtosis features, and simulated frequency centroid features, and determining the simulated skewness features, simulated kurtosis features, and simulated frequency centroid features as characterization parameters.

[0242] Step 1505: construct a moisture coefficient algorithm based on the correlation between the characterization parameters and the moisture content of the roadbed.

[0243] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily performed in sequence in the order indicated by the arrows. Unless clearly stated herein, the execution of these steps is not strictly limited in order, and these steps can be performed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above may include multiple steps or multiple stages, and these steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of the steps or stages in other steps or other steps. It is understandable that the various steps in different embodiments can be freely combined as needed, and the various non-contradictory schemes formed by the combination all fall within the scope of protection of this application.

[0244] Based on the same inventive concept, embodiments of the present application also provide a water damage assessment device for implementing the aforementioned water damage assessment method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more water damage assessment device embodiments provided below can be found in the above-described limitations of the water damage assessment method and will not be further elaborated here.

[0245] In an exemplary embodiment, Figure 16 As shown, a water damage assessment device 1600 is provided, comprising: an acquisition module 1601, a simulation module 1602, an extraction module 1603 and a determination module 1604, wherein:

[0246] The acquisition module 1601 is used to acquire the attribute dataset and radar dataset of the railway track bed to be detected.

[0247] The simulation module 1602 is used to simulate various simulated water content coefficient values ​​of the railway track bed to be detected under various relative dielectric constants according to the attribute data set, and to construct an evaluation standard for the railway track bed to be detected according to the various simulated water content coefficient values.

[0248] The extraction module 1603 is used to extract waveform features from the radar data set to obtain a feature data set, and determine the moisture coefficient value of the railway track bed to be detected based on the feature data set and the moisture coefficient algorithm.

[0249] The determination module 1604 is configured to determine a water damage assessment result of the railway track bed to be detected based on the evaluation criteria and the water content coefficient value of the railway track bed to be detected.

[0250] In an exemplary embodiment, the simulation module 1602 includes:

[0251] The first establishment submodule is used to establish a concentric circle roadbed model of the railway roadbed to be detected based on the attribute data set.

[0252] The first simulation submodule is used to simulate the simulated radar data set of the railway roadbed to be detected under various relative dielectric constants according to the concentric circle roadbed model.

[0253] The first determination submodule is used to determine each simulated water coefficient value corresponding to each relative dielectric constant based on the simulated radar data set, and to construct an evaluation standard for the railway track bed to be detected according to each simulated water coefficient value and the damage degree corresponding to each relative dielectric constant.

[0254] In an exemplary embodiment, the concentric circle roadbed model includes a healthy concentric circle roadbed model and a water-containing concentric circle roadbed model. The first simulation submodule is specifically used to simulate a healthy simulation radar data subset of the railway roadbed to be detected in a healthy state based on the healthy concentric circle roadbed model; simulate each water-containing simulation radar data subset of the railway roadbed to be detected in a water-containing state based on the water-containing concentric circle roadbed model and each relative dielectric constant; and construct a simulated radar data set based on the healthy simulation radar data subset and each water-containing simulation radar data subset.

[0255] In an exemplary embodiment, the simulated radar data set includes simulated radar data subsets corresponding to respective relative permittivities, and the first determination submodule includes a second determination submodule and a first construction submodule. The second determination submodule is specifically configured to, for each simulated radar data subset corresponding to a relative permittivity, filter the simulated radar data subsets based on the water-bearing regions in the concentric circle roadbed model, average the filtered simulated radar data subsets to obtain average simulated waveform data, perform a wavelet transform on the average simulated waveform data, and extract a simulated feature set from the average simulated waveform data after the wavelet transform, and perform data operations on the simulated feature set according to a water coefficient algorithm to obtain a simulated water coefficient value corresponding to the relative permittivity.

[0256] In an exemplary embodiment, the extraction module includes a first extraction submodule and a third determination submodule. The first extraction submodule is specifically configured to filter the waveform data of each initial time-domain waveform in the radar data set to obtain waveform data of each time-domain waveform; average the waveform data of each time-domain waveform to obtain average time-domain waveform data of the internal area of ​​the railway track bed to be detected; perform a wavelet transform on the average time-domain waveform data to obtain transformed average time-domain waveform data; and extract skewness, kurtosis, and frequency centroid features from the transformed average time-domain waveform data to obtain a feature dataset.

[0257] In one exemplary embodiment, the feature dataset includes skewness, kurtosis, and frequency centroid features, and the extraction module includes a first extraction submodule and a third determination submodule. The third determination submodule is specifically configured to perform data operations on the skewness, kurtosis, and frequency centroid features based on a moisture coefficient algorithm to obtain a moisture coefficient value for the railway trackbed to be detected.

[0258] In an exemplary embodiment, the evaluation criteria include each moisture coefficient interval and the degree of damage corresponding to each moisture coefficient interval. The determination module 1604 is specifically used to determine the target moisture coefficient interval in which the moisture coefficient value is located in each moisture coefficient interval; and determine the degree of damage corresponding to the target moisture coefficient interval as the water damage assessment result of the railway track bed to be detected.

[0259] Each module in the water damage assessment device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in hardware form, or stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0260] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 17 As shown. The computer device includes a processor, memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals via wired or wireless means, and the wireless means can be implemented via Wi-Fi, a mobile cellular network, near-field communication (NFC), or other technologies. When executed by the processor, the computer program implements a water damage assessment method. The display unit of the computer device is used to produce a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse.

[0261] Those skilled in the art will understand that Figure 17 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0262] In one embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0263] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0264] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0265] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), quantum computing-based data processing logic devices, artificial intelligence (AI) processors, and the like.

[0266] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0267] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A water damage assessment method, characterized in that: The method comprises: Obtain the attribute dataset and radar dataset of the railway track bed to be detected; Simulating various simulated moisture coefficient values ​​of the railway track bed to be detected at various relative dielectric constants according to the attribute data set, and constructing an evaluation standard for the railway track bed to be detected according to the various simulated moisture coefficient values; Extracting waveform features from the radar data set to obtain a feature data set, and determining a moisture coefficient value of the railway track bed to be detected based on the feature data set and a moisture coefficient algorithm; The water damage assessment result of the railway track bed to be detected is determined according to the evaluation standard of the railway track bed to be detected and the water content coefficient value.

2. The method according to claim 1, characterized in that The simulating each simulated moisture coefficient value of the railway track bed to be detected at each relative dielectric constant according to the attribute data set, and constructing an evaluation standard for the railway track bed to be detected according to each simulated moisture coefficient value, includes: Establishing a concentric circle roadbed model of the railway roadbed to be detected according to the attribute data set; Simulating a simulated radar data set of the railway roadbed to be detected at various relative dielectric constants according to the concentric circle roadbed model; The simulated water content values ​​corresponding to the relative dielectric constants are determined based on the simulated radar data set, and an evaluation standard for the railway track bed to be detected is constructed according to the simulated water content values ​​and the damage degrees corresponding to the relative dielectric constants.

3. The method according to claim 2, characterized in that The concentric circle roadbed model includes a healthy concentric circle roadbed model and a water-containing concentric circle roadbed model. The simulated radar data set of the railway roadbed to be detected under various relative dielectric constants is simulated according to the concentric circle roadbed model, including: A healthy simulated radar data subset of the railway roadbed to be detected in a healthy state is simulated based on the healthy concentric circle roadbed model; Simulating each subset of water-containing simulated radar data of the railway track bed to be detected in a water-containing state based on the water-containing concentric circle track bed model and each relative dielectric constant; A simulated radar data set is constructed based on the healthy simulated radar data subset and each of the water-containing simulated radar data subsets.

4. The method according to claim 2, characterized in that The simulated radar data set includes a simulated radar data subset corresponding to each relative dielectric constant, and determining each simulated water coefficient value corresponding to each relative dielectric constant based on the simulated radar data set includes: For each of the simulated radar data subsets corresponding to the relative dielectric constant, the simulated radar data subsets are screened according to the water-bearing areas in the concentric circle roadbed model, and the screened simulated radar data subsets are averaged to obtain average simulated waveform data; Performing wavelet transformation on the average analog waveform data, and extracting a simulation feature set of the average analog waveform data after the wavelet transformation; Data calculation is performed on the simulation feature set according to a water content coefficient algorithm to obtain a simulated water content coefficient value corresponding to the relative dielectric constant.

5. The method according to claim 1, wherein The step of extracting waveform features from the radar data set to obtain a feature data set includes: filtering waveform data of each initial time domain waveform in the radar data set to obtain waveform data of each time domain waveform; Averaging the waveform data of each of the time domain waveforms to obtain average time domain waveform data of the inner area of ​​the railway track bed to be detected; Performing wavelet transform on the average time-domain waveform data to obtain the transformed average time-domain waveform data; The skewness feature, the kurtosis feature and the frequency centroid feature are extracted from the transformed average time domain waveform data to obtain a feature data set.

6. The method according to claim 1, characterized in that The characteristic data set includes a skewness feature, a kurtosis feature, and a frequency centroid feature. Determining the moisture coefficient value of the railway track bed to be detected based on the characteristic data set and a moisture coefficient algorithm includes: Data calculation is performed on the skewness feature, the kurtosis feature and the frequency centroid feature according to a water content coefficient algorithm to obtain a water content coefficient value of the railway track bed to be detected.

7. The method according to claim 1, characterized in that The evaluation criteria include each water coefficient interval and the damage degree corresponding to each water coefficient interval. Determining the water damage assessment result of the railway track bed to be detected based on the evaluation criteria of the railway track bed to be detected and the water coefficient value includes: Determining a target moisture coefficient interval within each moisture coefficient interval where the moisture coefficient value is located; The damage degree corresponding to the target water coefficient interval is determined as the water damage assessment result of the railway track bed to be detected.

8. A water damage assessment device, characterized in that: The device comprises: An acquisition module is used to acquire the attribute dataset and radar dataset of the railway track bed to be detected; a simulation module, configured to simulate various simulated moisture coefficient values ​​of the railway track bed to be detected at various relative dielectric constants according to the attribute data set, and to construct an evaluation standard for the railway track bed to be detected according to the various simulated moisture coefficient values; an extraction module, configured to extract waveform features from the radar data set to obtain a feature data set, and determine the moisture coefficient value of the railway track bed to be detected based on the feature data set and a moisture coefficient algorithm; The determination module is used to determine the water damage assessment result of the railway track bed to be detected according to the evaluation standard of the railway track bed to be detected and the water content coefficient value.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are 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 steps of the method according to any one of claims 1 to 7 are implemented.

Images