Dirty ballast damage identification method and device

Abstract

The invention discloses a dirty ballast damage identification method and device, and relates to the technical field of railway damage detection. Comprising the following steps: firstly, collecting a railway ballast surface near-infrared image with known hydrological characteristic parameters, and carrying out noise reduction and enhancement preprocessing to generate a training sample data set; a neural network prediction model is established based on the data set, a training sample image is used as input, and hydrological characteristic parameters are used as labels. A near-infrared image of a to-be-detected railway ballast is collected and preprocessed, and a hydrological characteristic parameter prediction value is obtained. And meanwhile, geometric parameters of the dirty particles are obtained. And calculating a railway ballast water seepage influence coefficient by utilizing the geometric parameters and the predicted values, and generating a railway ballast structure response coefficient by combining the average roughness and the vibration characteristic parameters of the railway ballast. And finally, acquiring environmental parameters, dynamically correcting a damage judgment threshold, comprehensively generating a railway ballast damage index, comparing the railway ballast damage index with the corrected threshold, and generating a railway ballast damage identification result. And efficient and accurate ballast damage identification is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of railway damage detection, and in particular to a dirty ballast damage identification method and device. Background Art

[0002] In railway transportation, ballasted track is one of the main track structures, and the health of its ballast layer plays a vital role in the safety and stability of the railway. The ballast layer not only bears the weight of the track structure, but also affects the running stability and comfort of the train. However, over time, environmental factors (such as rainfall, temperature changes) and vibrations caused by train operation will cause the ballast layer to become dirty and damaged. Especially under extreme climatic conditions, water accumulation and sludge accumulation will aggravate the deterioration of the ballast, resulting in a decrease in the strength and stiffness of the ballast, which in turn affects the service life and operational safety of the railway. Therefore, how to accurately identify and evaluate the damage status of ballast has become an important issue in railway maintenance and management.

[0003] Existing ballast damage detection technologies can be roughly divided into physical detection methods and image processing methods. Physical detection methods usually rely on sensors and test equipment, such as vibration meters, ultrasonic detectors, etc. Although they can provide relatively accurate data, their high equipment costs, complex operating procedures and dependence on environmental conditions limit their application. Image processing methods, especially the application of near-infrared images, provide a new idea for ballast damage identification. Near-infrared images can effectively penetrate some dirt and obtain characteristic information on the ballast surface, thereby overcoming the limitations of traditional methods to a certain extent. However, traditional image processing methods are usually difficult to effectively combine with environmental parameters, resulting in inaccurate assessment of damage status. Therefore, how to combine near-infrared image technology with environmental parameters and hydrological characteristics to form an efficient and accurate damage identification solution is an urgent problem to be solved in the current technical field.

[0004] In the prior art, publication number CN112199838A discloses a method and device for identifying damage of a ballast layer under a ballasted track sleeper, comprising: establishing a two-dimensional ballasted track system model for describing the response of the ballasted track to be tested under the action of an excitation force, and dividing the ballast layer under the sleeper in the model into n regions; performing modal identification on acceleration time history data collected from multiple data collection points on the sleeper of the ballasted track to be tested in a hammer test, and obtaining the frequency and vibration mode of the ballasted track to be tested in a damaged state; using the identified frequency and vibration mode as target data, and using a sparse Bayesian method to identify the ballast stiffness damage parameters of the model, and obtaining the maximum a posteriori estimate of the ballast stiffness damage parameters in each region under the damaged state as the corresponding damage degree, so as to determine the damaged area and damage degree of the ballast layer under the ballasted track sleeper to be tested. However, the acceleration time history data collected through the hammer test in this scheme may be affected by environmental noise or instrument errors. The data interfered by noise may lead to inaccurate modal identification results. At the same time, the damage area and damage degree are judged only by the acceleration time history data. The data is too single and does not fully consider external parameters, such as water accumulation caused by dirt and environmental parameters. Therefore, the accuracy and effectiveness of damage identification are reduced.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The object of the present invention is to provide a method and device for identifying dirty ballast damage to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions: A dirty ballast damage identification method, the specific steps comprising: Collecting several near-infrared images of the ballast surface with known hydrological characteristic parameters, preprocessing the near-infrared images of the ballast surface to obtain training sample images, mapping the training sample images with corresponding hydrological characteristic parameters one by one to generate a training sample data set, wherein the hydrological characteristic parameters include the ballast surface area, the water surface area and the water accumulation volume, and the preprocessing includes image denoising preprocessing and image enhancement preprocessing; Based on the data in the training sample data set, a neural network prediction model is established, the training sample images in the training sample data set are used as the input of the model, and the hydrological characteristic parameters in the training sample data set are used as labels to train the neural network prediction model to obtain a hydrological characteristic prediction model; Collect near-infrared images of the surface of the ballast to be inspected, pre-process the images and input them into the trained hydrological characteristic prediction model to obtain the predicted values ​​of the hydrological characteristic parameters of the ballast to be inspected, and simultaneously obtain the geometric parameters of the dirt particles on the surface of the ballast to be inspected, the geometric parameters including the average particle size of the dirt particles, the average perimeter of the dirt particles, the average surface area of ​​the dirt particles and the average projected area; According to the geometric parameters of the dirt particles on the ballast surface to be detected, combined with the predicted values ​​of the hydrological characteristic parameters, the ballast water seepage influence coefficient is calculated, the average roughness of the ballast surface and the vibration characteristic parameters of the ballast when the train passes by are obtained, and the ballast structure response coefficient is calculated based on the vibration characteristic parameters and the average roughness, and the vibration characteristic parameters include vibration frequency and amplitude; The environmental parameters of the location of the ballast to be inspected are collected, and the pre-set damage judgment threshold is dynamically corrected according to the obtained environmental parameters to obtain the damage judgment correction threshold. Based on the obtained ballast structure response coefficient and ballast water seepage influence coefficient, a ballast damage index is comprehensively generated. The ballast damage index is compared with the damage judgment correction threshold. According to different comparison results, a corresponding ballast damage identification result is generated. The environmental parameters include average rainfall, daily average temperature and daily average humidity.

[0008] Furthermore, each collected near-infrared image of the ballast surface is preprocessed to obtain a training sample image, wherein the preprocessing includes: unifying the image size, image enhancement and denoising preprocessing, wherein a wavelet transform denoising method is used to denoise each near-infrared image of the ballast surface, and a bilateral filter is used to perform image enhancement preprocessing on each near-infrared image of the ballast surface; The method for generating the training sample data set is as follows: mapping the training sample images to the corresponding hydrological characteristic parameters one by one to form a corresponding grid, and recording the formed grid as the training sample data set; Based on the long short-term memory network model LSTM model, a neural network model is established, and the activation function and optimization algorithm are selected. The Tanh function is selected as the activation function, and Adam is selected as the optimization algorithm of the LSTM model; the formula of the Tanh function is: ; In the formula, Represents the Tanh function, independent variable Represents the weighted sum of the neuron's input, that is, the result of weighted summation of the input received by the neuron from the previous layer; At the same time, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of training times, the batch processing number, and the number of hidden layer neurons; The number of network layers is set to 4 layers, the number of iterations is set to 200, the learning rate is set to 0.01, the batch size is set to 32, the number of training times is set to 200, the batch size is set to 256, and the number of hidden layer neurons is 32; The trained hydrological characteristic prediction model takes as input the preprocessed near-infrared image data of the ballast surface, and outputs the predicted values ​​of hydrological characteristic parameters, including the predicted values ​​of the ballast surface area, the water surface area, and the accumulated water volume.

[0009] Furthermore, according to the geometric parameters of the dirt particles on the surface of the ballast to be detected and the predicted values ​​of the hydrological characteristic parameters, the ballast water seepage influence coefficient is calculated, wherein the specific formula for calculating the ballast water seepage influence coefficient is: ; In the formula, is the influence coefficient of water seepage of the ballast to be tested, and are the predicted value of the surface area and volume of water on the ballast to be tested, respectively. is the shape factor of the dirt particle, is the average particle size of the dirt particles, and the shape factor of the dirt particles The calculation is based on the geometric parameters of the dirt particles, where the shape factor of the dirt particles is The specific formula on which the calculation is based is: ; In the formula, is the average surface area of ​​the dirt particles, is the average circumference of the dirt particles, is the average projected area of ​​the dirt particles.

[0010] Furthermore, the average roughness of the ballast surface and the vibration characteristic parameters of the ballast when the train passes by are obtained, and the ballast structure response coefficient is calculated based on the vibration characteristic parameters and the average roughness, wherein the formula for calculating the ballast structure response coefficient is: ; In the formula, is the ballast structure response coefficient, is the average roughness of the ballast surface to be tested, is the vibration frequency, is the vibration amplitude.

[0011] Furthermore, the average roughness of the ballast surface to be tested is The specific method of obtaining is: The ballast surface roughness measurement equipment uses a stylus profilometer. Multiple sampling areas are randomly selected from the ballast area to be tested. The stylus profilometer is used to measure and analyze the sampling areas to obtain the surface roughness of each sampling area, and the surface roughness average of all sampling areas is calculated. This average is used as the average roughness of the ballast surface to be tested. The calculation formula is as follows: ; ; In the formula, represents the sampling length of the stylus profilometer in the kth sampling area, k is the index of the sampling area, and , K is the number of sampling areas, is the height of the qth sampling point within the sampling length of the kth sampling area from the center line, q is the coordinate of the sampling point within the sampling length, is the surface roughness of the kth sampling area, is the average roughness of the ballast surface to be tested, the center line and The data is obtained through the built-in software of the stylus profilometer.

[0012] Furthermore, based on the obtained ballast structure response coefficient and ballast water seepage influence coefficient, the ballast damage index is comprehensively generated, wherein the formula for calculating the ballast damage index is: ; In the formula, is the ballast damage index, and are the weight coefficients of the ballast structure response coefficient and the ballast water seepage influence coefficient to be tested, respectively. , and and Both are greater than 0.

[0013] Furthermore, the environmental parameters of the location of the ballast to be inspected are collected, and the pre-set damage judgment threshold is dynamically corrected according to the obtained environmental parameters to obtain the damage judgment correction threshold, wherein the damage judgment correction threshold is calculated based on the formula: ; In the formula, Correct the threshold for damage judgment, is the pre-set damage judgment threshold. is the average rainfall, specifically the average monthly rainfall, is the average daily humidity, is the daily average temperature, For reference rainfall, is the reference humidity, is the reference temperature; The ballast damage index is compared with the damage judgment correction threshold, and the corresponding ballast damage identification result is generated according to different comparison results. The specific judgment logic is as follows: when When the ballast to be tested is in poor health, it is judged that the ballast to be tested is damaged and should be stopped from use in time for repair; when When the ballast to be inspected is judged to be in good health, it means that the ballast to be inspected is at risk of urban damage and should be inspected and reinforced; when , the health status of the ballast to be tested is judged to be excellent, indicating that there is no damage to the ballast to be tested.

[0014] The present invention also provides a dirty ballast damage identification device, which is used to execute the above-mentioned dirty ballast damage identification method, comprising: A training sample acquisition module is used to acquire several near-infrared images of the ballast surface with known hydrological characteristic parameters, pre-process the near-infrared images of the ballast surface to obtain training sample images, map the training sample images with the corresponding hydrological characteristic parameters one by one, and generate a training sample data set. The hydrological characteristic parameters include the ballast surface area, the water surface area, and the water accumulation volume. The pre-processing includes image noise reduction pre-processing and image enhancement pre-processing; The prediction model training module is used to establish a neural network prediction model based on the data in the training sample data set, take the training sample images in the training sample data set as the input of the model, and take the hydrological characteristic parameters in the training sample data set as labels to train the neural network prediction model to obtain a hydrological characteristic prediction model; A dirt parameter acquisition module is used to acquire near-infrared images of the surface of the ballast to be detected, and input the images into the trained hydrological characteristic prediction model after preprocessing to obtain the predicted values ​​of the hydrological characteristic parameters of the ballast to be detected, and at the same time obtain the geometric parameters of the dirt particles on the surface of the ballast to be detected, wherein the geometric parameters include the average particle size of the dirt particles, the average perimeter of the dirt particles, the average surface area of ​​the dirt particles, and the average projected area; The structural response analysis module is used to calculate the ballast water seepage influence coefficient according to the geometric parameters of the dirt particles on the ballast surface to be detected and the predicted values ​​of the hydrological characteristic parameters, obtain the average roughness of the ballast surface and the vibration characteristic parameters of the ballast when the train passes by, and calculate and generate the ballast structural response coefficient according to the vibration characteristic parameters and the average roughness, wherein the vibration characteristic parameters include vibration frequency and amplitude; The comprehensive damage judgment module is used to collect the environmental parameters of the location of the ballast to be detected, dynamically correct the pre-set damage judgment threshold according to the obtained environmental parameters, obtain the damage judgment correction threshold, and comprehensively generate the ballast damage index based on the obtained ballast structure response coefficient and ballast water seepage influence coefficient. The ballast damage index is compared with the damage judgment correction threshold, and the corresponding ballast damage identification result is generated according to different comparison results. The environmental parameters include average rainfall, daily average temperature and daily average humidity.

[0015] Compared with the prior art, the present invention has the following beneficial effects: First, by preprocessing the near-infrared image of the ballast surface, such as noise reduction and image enhancement, the image quality can be effectively improved, making the subsequent feature extraction more accurate. Secondly, the established neural network prediction model combines the training sample image with the hydrological characteristic parameters using a mapping relationship. The model can automatically extract the hydrological characteristic parameters of the ballast, such as the volume and area of ​​water accumulation, and then accurately predict the state of the ballast to be tested. Compared with traditional methods, the deep learning model has stronger adaptability and self-optimization capabilities, which significantly improves the accuracy and efficiency of ballast damage identification. In addition, the ballast water seepage influence coefficient is generated by combining the geometric parameters of the dirt particles on the surface of the ballast to be tested with the predicted values ​​of the hydrological characteristics, providing a more comprehensive perspective for the health assessment of the ballast. Through this coefficient, the specific impact of moisture and dirt on the performance of the ballast can be deeply analyzed. Finally, the dynamic correction of the collected environmental parameters (such as average rainfall, daily average temperature and humidity) in damage judgment effectively improves the accuracy of the damage judgment threshold. By updating the damage judgment correction threshold in real time, it can adapt to different climatic conditions and significantly reduce the risk of misjudgment caused by environmental changes. This dynamic adaptability enhances the reliability of ballast damage identification, making railway maintenance management more efficient and scientific. It achieves efficient and accurate identification of dirty ballast and improves the monitoring and maintenance level of railway ballast. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the overall method flow of the present invention; Figure 2 It is a schematic diagram of the overall structure of the device of the present invention. DETAILED DESCRIPTION

[0017] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0018] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0019] Example: See also Figure 1 , the present invention provides a technical solution: A dirty ballast damage identification method, the specific steps comprising: Step 1: Collect several near-infrared images of the ballast surface with known hydrological characteristic parameters, pre-process the near-infrared images of the ballast surface to obtain training sample images, map the training sample images with the corresponding hydrological characteristic parameters one by one, and generate a training sample data set. The hydrological characteristic parameters include the ballast surface area, water surface area and water accumulation volume. The pre-processing includes image denoising pre-processing and image enhancement pre-processing.

[0020] Preprocessing each collected near-infrared image of the ballast surface to obtain a training sample image, wherein the preprocessing includes: unifying the image size, image enhancement and denoising preprocessing, wherein the wavelet transform denoising method is used to denoise each near-infrared image of the ballast surface, and bilateral filtering is used to perform image enhancement preprocessing on each near-infrared image of the ballast surface; The method for generating the training sample data set is as follows: mapping the training sample images to the corresponding hydrological characteristic parameters one by one to form a corresponding grid, and recording the formed grid as the training sample data set; The specific method for unifying the image size is to choose a unified target size, such as width and height (for example, 512x512 pixels or 256x256 pixels). This size should be selected based on the subsequent processing requirements and the availability of computing resources. Use a programming language (such as Python) or image processing software (such as OpenCV, PIL, etc.) to import the images to be processed, obtain the original width and height of each image for scaling, and choose the appropriate scaling method according to specific needs. Common scaling methods include: Nearest Neighbor Interpolation: simple and fast, but may cause jagged edges; Bilinear Interpolation: considers the weighted average of the surrounding four pixels, which has better results; Bicubic Interpolation: considers the nearby 16 pixels, which usually provides smoother results.

[0021] The denoising method of wavelet transform is used to denoise the near-infrared image of the ballast surface. The specific steps of the wavelet transform denoising method include: decomposing the near-infrared image of the ballast surface through wavelet transform to obtain the wavelet coefficients of the image at different scales and directions; thresholding the wavelet coefficients to set the low-amplitude wavelet coefficients to zero and retain the high-amplitude wavelet coefficients; inversely transforming the wavelet coefficients after thresholding, reconstructing the processed coefficients into an image, and completing the image denoising; Bilateral filtering is used to enhance the details of the near-infrared image of the ballast surface. The formula for the specific filtering transformation is: ; In the formula, is the coordinate vector in the image coordinate system, is the coordinate vector The gray value at Gray value The gray value after bilateral filtering transformation is are all Gaussian functions, where The formula is: ; ; In the formula, is the coordinate vector in the image coordinate system, is the coordinate vector The gray value at and They are The standard deviation of .

[0022] Step 2: Based on the data in the training sample data set, a neural network prediction model is established. The training sample images in the training sample data set are used as the input of the model, and the hydrological characteristic parameters in the training sample data set are used as labels to train the neural network prediction model to obtain a hydrological characteristic prediction model.

[0023] Based on the long short-term memory network model LSTM model, a neural network model is established, and the activation function and optimization algorithm are selected. The Tanh function is selected as the activation function, and Adam is selected as the optimization algorithm of the LSTM model; the formula of the Tanh function is: ; In the formula, Represents the Tanh function, independent variable Represents the weighted sum of the neuron's input, that is, the result of weighted summation of the input received by the neuron from the previous layer; At the same time, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of training times, the batch processing number, and the number of hidden layer neurons; The number of network layers is set to 4 layers, the number of iterations is set to 200, the learning rate is set to 0.01, the batch size is set to 32, the number of training times is set to 200, the batch size is set to 256, and the number of hidden layer neurons is 32; The trained hydrological characteristic prediction model takes as input the preprocessed near-infrared image data of the ballast surface, and outputs the predicted values ​​of hydrological characteristic parameters, including the predicted values ​​of the ballast surface area, the water surface area, and the accumulated water volume.

[0024] In the near-infrared band (usually ranging from 700nm to 2500nm), the reflectivity of water is very low, especially near 1400nm and 1900nm, where water exhibits strong absorption characteristics. In contrast, the reflectivity of surrounding materials (such as soil or ballast) in the same band is relatively high, so it can be used to detect the surface water area of ​​ballast. The volume of accumulated water can be obtained by manually extracting the surface water and conducting detection.

[0025] LSTM is particularly suitable for processing time series data, such as the prediction of hydrological characteristics, because time series has time dependence, and the data at the previous moment may have a significant impact on the data at the current moment. LSTM can capture these time correlations and improve the accuracy of predictions. LSTM can learn effective features with fewer data samples, and its structure enables the model to maintain good performance when facing sparse data and noise. Due to its complex network structure, the LSTM model can better adapt to data changes when facing variable input features.

[0026] Step 3: Collect a near-infrared image of the surface of the ballast to be inspected, pre-process the image and input it into the trained hydrological characteristic prediction model to obtain the predicted value of the hydrological characteristic parameters of the ballast to be inspected, and at the same time obtain the geometric parameters of the dirt particles on the surface of the ballast to be inspected, which include the average particle size of the dirt particles, the average circumference of the dirt particles, the average surface area of ​​the dirt particles and the average projected area.

[0027] The specific method for obtaining the geometric parameters of the dirt particles on the surface of the ballast to be tested is as follows: several sections of the ballast to be tested are selected for analysis, several sampling points are randomly selected in each section for multiple sampling to obtain the dirt particle samples, and the geometric parameters of the dirt particle samples are determined experimentally, and the mean value of the geometric parameters of the dirt particle samples in the same section is used as the geometric parameters of the section, and finally the geometric parameters of the dirt particles on the entire ballast to be tested are characterized by the average value of the geometric parameters of each section.

[0028] The sampling points selected randomly select dirt particle samples to ensure that the collected samples can reflect the size and shape distribution of dirt particles in the area. Standard sieves are used to physically screen particles of different sizes, and the mass of particles passing through each sieve is measured. The distribution of particle diameters is calculated based on the aperture of the sieve, and the average value is calculated based on the dirt particle size data obtained at each sampling point, and this average value is used as the average particle size of dirt particles in this section.

[0029] The circumference of the dirty particles can be obtained by extracting the contours of the particles, calculating the circumference of the particle contours through software such as OpenCV or MATLAB, and averaging the circumference data of the particle contours at each sampling point, and using this average value as the average circumference of the dirty particles in the section.

[0030] The projection area of ​​dirty particles is obtained by extracting a binary image of the particles (background and particles separated) in the image processing software. The software automatically calculates the projection area of ​​the particles on the plane, or covers the particles with transparent paper, draws the outline and then calculates the area (suitable for small sample particles). The average projection area of ​​the particles at each sampling point is calculated and this average is used as the average projection area of ​​the dirty particles in the section.

[0031] The surface area of ​​the dirt particles can also be calculated using image processing software. The actual surface area of ​​the particles can be calculated by averaging the surface area data of the dirt particles at each sampling point, and the average value can be used as the surface area of ​​the dirt particles in the section.

[0032] The preprocessing of the near-infrared image of the ballast surface to be inspected is the same as the above method and will not be repeated here.

[0033] Step 4: According to the geometric parameters of the dirt particles on the ballast surface to be detected and the predicted values ​​of the hydrological characteristic parameters, the ballast water seepage influence coefficient is calculated, the average roughness of the ballast surface and the vibration characteristic parameters of the ballast when the train passes by are obtained, and the ballast structure response coefficient is calculated based on the vibration characteristic parameters and the average roughness. The vibration characteristic parameters include vibration frequency and amplitude.

[0034] According to the geometric parameters of the dirt particles on the ballast surface to be detected, combined with the predicted values ​​of the hydrological characteristic parameters, the ballast seepage influence coefficient is calculated. The specific formula for calculating the ballast seepage influence coefficient is: ; In the formula, is the influence coefficient of water seepage of the ballast to be tested, and are the predicted value of the surface area and volume of water on the ballast to be tested, respectively. is the shape factor of the dirt particle, is the average particle size of dirt particles.

[0035] It should be noted that the water seepage influence coefficient of the ballast to be tested is It is used to indicate the water permeability of ballast. The larger the value, the worse the water permeability of the ballast to be tested, resulting in water accumulation and a greater probability of damage.

[0036] Among them, the predicted value of the water accumulation surface area of ​​the ballast to be tested is The larger the surface area of ​​water accumulation, the more obvious the presence of water and its impact on the ballast. The more water accumulates on the ballast surface, which in turn reflects the decrease in water permeability and the predicted value of water accumulation volume of the ballast to be tested. The increase in the volume of water accumulation also means that the impact of water on the permeability and water accumulation of the ballast increases. Therefore, both are related to the water seepage influence coefficient of the ballast to be tested. By adding these two parameters, the total impact of water accumulation is emphasized, reflecting the relationship between the water permeability of the ballast and the water accumulation condition. The more water accumulates, the worse the water permeability may be, through the exponential function It means that the predicted value of water accumulation surface area and water accumulation volume of the ballast to be tested obviously reflect the water seepage performance of the ballast. When they increase, the water seepage influence coefficient of the ballast to be tested will be significantly increased.

[0037] Shape factor of dirt particles , a larger shape factor generally indicates an increase in surface irregularity and porosity of the particle, which may improve water flow, so the shape factor of the dirty particle is The influence coefficient of water seepage of the ballast to be tested Inversely proportional, through the square form The influence of shape on water permeability is emphasized and nonlinear relationships are added.

[0038] Average particle size of dirt particles The larger the particle size, the larger the gap between particles, which increases the fluidity of water and leads to less water accumulation. Therefore, the average particle size of the dirt particles is The influence coefficient of water seepage of the ballast to be tested Inversely proportional, through the logarithmic function It represents the change in the effect of smoothed particle size on water seepage, reduces over-sensitivity in the case of small particle sizes, and allows changes in particle size to be reflected in the influence coefficient in a smoother form.

[0039] Shape factor of dirt particles The calculation is based on the geometric parameters of the dirt particles, where the shape factor of the dirt particles is The specific formula on which the calculation is based is: ; In the formula, is the average surface area of ​​the dirt particles, is the average circumference of the dirt particles, is the average projected area of ​​the dirt particles.

[0040] Shape factor of dirt particles The larger the size, the more complex and irregular the shape of the dirt particles. Indicates the sphericity of the dirt particles. The larger the value is and the closer it is to 1, the closer the dirt particles are to the standard sphere. Therefore, the first part Shape factor of dirt particles Inversely proportional.

[0041] Part 2 Represents the structure of the dirty particles, the average perimeter of the dirty particles. The perimeter is related to the boundary characteristics of the particles. A larger perimeter may mean the complexity and irregularity of the particles. The average projected area of ​​the dirty particles. The projected area is the "shadow" of the particle in a specific direction, which is directly related to the shape characteristics of the particle. By normalizing the ratio of perimeter to projected area, the morphological characteristics of the particles in the flow state can be further reflected. The larger the value, the more complex and irregular the structure of the dirty particles. Therefore, the second part Shape factor of dirt particles Directly proportional.

[0042] The average roughness of the ballast surface and the vibration characteristic parameters of the ballast when the train passes by are obtained, and the ballast structure response coefficient is calculated based on the vibration characteristic parameters and the average roughness. The formula for calculating the ballast structure response coefficient is: ; In the formula, is the ballast structure response coefficient, is the average roughness of the ballast surface to be tested, is the vibration frequency, is the vibration amplitude.

[0043] It should be noted that the ballast structure response coefficient It is used to characterize the vibration condition and average surface roughness of the ballast to be tested during normal operation. The larger the value, the rougher the surface of the ballast to be tested, resulting in more uneven force, greater vibration, greater risk of accidents, and greater probability of damage to the ballast to be tested.

[0044] The average roughness of the ballast surface to be tested is Roughness is an important parameter that determines the contact characteristics between ballast and train wheel rail, reflecting the irregularity and friction characteristics of the ballast surface. The average roughness of the ballast surface to be tested is The larger the value, the more uneven the stress on the ballast to be tested is, and the greater the probability of damage. Therefore, the average roughness of the surface of the ballast to be tested is and ballast structure response coefficient Proportional, through The square root emphasizes the effect of roughness on the response coefficient of the ballast structure, where greater roughness will result in a greater response.

[0045] The vibration frequency refers to the frequency of vibration generated when a train passes over the ballast. The higher the vibration frequency, the greater the dynamic response may be, affecting the stability of the ballast. The vibration amplitude indicates the intensity or displacement of the vibration. A larger vibration amplitude will also lead to an increase in the response of the ballast structure. Therefore, both the vibration frequency and the vibration amplitude are related to the ballast structure response coefficient. is proportional to It is a comprehensive indicator that combines vibration frequency and vibration amplitude, reflecting the total intensity of vibration. By taking the square root, we ensure that the effects of frequency and amplitude can be combined, rather than linearly superimposed, through the exponential function It represents the attenuation characteristics of the structural response coefficient as the vibration frequency and amplitude increase. As the vibration intensity increases, the factors affecting the stability and response of the ballast will increase significantly.

[0046] The vibration frequency and vibration amplitude are detected by vibration sensors set on both sides of the track, and data is obtained. The data collection time period is specifically from the time when the front of the train arrives at the vibration sensor to the time when the rear of the train leaves the vibration sensor. The collection interval is set to 0.3 seconds. The average value of the collected vibration frequency data is calculated and the average value is used as the vibration frequency. , the maximum value of the vibration frequency data during the acquisition process is taken as the vibration amplitude .

[0047] The average roughness of the ballast surface to be tested is The specific method of obtaining is: The ballast surface roughness measurement equipment uses a stylus profilometer. Multiple sampling areas are randomly selected from the ballast area to be tested. The stylus profilometer is used to measure and analyze the sampling areas to obtain the surface roughness of each sampling area, and the surface roughness average of all sampling areas is calculated. This average is used as the average roughness of the ballast surface to be tested. The calculation formula is as follows: ; ; In the formula, represents the sampling length of the stylus profilometer in the kth sampling area, k is the index of the sampling area, and , K is the number of sampling areas, is the height of the qth sampling point within the sampling length of the kth sampling area from the center line, q is the coordinate of the sampling point within the sampling length, is the surface roughness of the kth sampling area, is the average roughness of the ballast surface to be tested, the center line and The data is obtained through the built-in software of the stylus profilometer.

[0048] Step 5: Collect environmental parameters of the location of the ballast to be tested, dynamically correct the pre-set damage judgment threshold according to the obtained environmental parameters, obtain the damage judgment correction threshold, comprehensively generate the ballast damage index based on the obtained ballast structure response coefficient and ballast water seepage influence coefficient, compare the ballast damage index with the damage judgment correction threshold, and generate corresponding ballast damage identification results according to different comparison results. The environmental parameters include average rainfall, daily average temperature and daily average humidity.

[0049] Based on the obtained ballast structure response coefficient and ballast water seepage influence coefficient, the ballast damage index is comprehensively generated, and the formula for calculating the ballast damage index is: ; In the formula, is the ballast damage index, and are the weight coefficients of the ballast structure response coefficient and the ballast water seepage influence coefficient to be tested, respectively. , and and Both are greater than 0.

[0050] Among them, the ballast damage index The larger the value, the greater the probability that the ballast to be tested is damaged. As mentioned above, the ballast structure response coefficient and the water seepage influence coefficient of the ballast to be tested are related to the ballast damage index. The proportional relationship is not described here. Through the exponential function It means that the increase of ballast structure response coefficient will significantly increase the ballast damage index. The growth of water seepage effect can be effectively handled, especially when the water seepage effect is small, the increase of the logarithmic function is relatively gentle, which is suitable for reflecting the impact of small changes on the overall damage.

[0051] Since the ballast structural response has a higher importance in the damage index, reflecting the dominant role of dynamic load on ballast damage in actual engineering, and the water seepage effect may be affected by the environment, , and and Both are greater than 0.

[0052] Collect the environmental parameters of the location of the ballast to be inspected, and dynamically correct the pre-set damage judgment threshold according to the obtained environmental parameters to obtain the damage judgment correction threshold. The damage judgment correction threshold is calculated based on the formula: ; In the formula, Correct the threshold for damage judgment, is the pre-set damage judgment threshold. is the average rainfall, specifically the average monthly rainfall, is the average daily humidity, is the daily average temperature, For reference rainfall, is the reference humidity, is the reference temperature; The daily average humidity is specifically the average value of the humidity detected every hour within 24 hours before the test day. The same is true for the daily average temperature. The monthly average rainfall is the average rainfall in the month before the test day and the average rainfall in the days before the test day.

[0053] The greater the rainfall and humidity, the higher the probability of water accumulation on the ballast, rather than water accumulation caused by damage to the non-detected ballast. Therefore, the system will judge that the probability of damage to the ballast to be detected increases, leading to misjudgment. Therefore, rainfall and humidity are related to the damage judgment correction threshold. The greater the rainfall and humidity, the higher the threshold should be to avoid misjudgment.

[0054] At the same time, the higher the temperature, the lower the strength of the track material may be. Therefore, under the same vibration intensity, the probability of detecting damage to the ballast increases, so the threshold should be lowered to avoid missed judgments.

[0055] The monthly average rainfall and reference rainfall can be obtained and set through the local meteorological station, and the reference humidity Generally set to to Reference temperature Generally set to .

[0056] The ballast damage index is compared with the damage judgment correction threshold, and the corresponding ballast damage identification result is generated according to different comparison results. The specific judgment logic is as follows: when When the ballast to be tested is in poor health, it is judged that the ballast to be tested is damaged and should be stopped from use in time for repair; when When the ballast to be inspected is judged to be in good health, it means that the ballast to be inspected is at risk of urban damage and should be inspected and reinforced; when , the health status of the ballast to be tested is judged to be excellent, indicating that there is no damage to the ballast to be tested.

[0057] See also Figure 2 The present invention further provides a dirty ballast damage identification device, the dirty ballast damage identification device is used to execute the above-mentioned dirty ballast damage identification method, comprising: A training sample acquisition module is used to acquire several near-infrared images of the ballast surface with known hydrological characteristic parameters, pre-process the near-infrared images of the ballast surface to obtain training sample images, map the training sample images with the corresponding hydrological characteristic parameters one by one, and generate a training sample data set. The hydrological characteristic parameters include the ballast surface area, the water surface area, and the water accumulation volume. The pre-processing includes image noise reduction pre-processing and image enhancement pre-processing; The prediction model training module is used to establish a neural network prediction model based on the data in the training sample data set, take the training sample images in the training sample data set as the input of the model, and take the hydrological characteristic parameters in the training sample data set as labels to train the neural network prediction model to obtain a hydrological characteristic prediction model; A dirt parameter acquisition module is used to acquire near-infrared images of the surface of the ballast to be detected, and input the images into the trained hydrological characteristic prediction model after preprocessing to obtain the predicted values ​​of the hydrological characteristic parameters of the ballast to be detected, and at the same time obtain the geometric parameters of the dirt particles on the surface of the ballast to be detected, wherein the geometric parameters include the average particle size of the dirt particles, the average perimeter of the dirt particles, the average surface area of ​​the dirt particles, and the average projected area; The structural response analysis module is used to calculate the ballast water seepage influence coefficient according to the geometric parameters of the dirt particles on the ballast surface to be detected and the predicted values ​​of the hydrological characteristic parameters, obtain the average roughness of the ballast surface and the vibration characteristic parameters of the ballast when the train passes by, and calculate and generate the ballast structural response coefficient according to the vibration characteristic parameters and the average roughness, wherein the vibration characteristic parameters include vibration frequency and amplitude; The comprehensive damage judgment module is used to collect the environmental parameters of the location of the ballast to be detected, dynamically correct the pre-set damage judgment threshold according to the obtained environmental parameters, obtain the damage judgment correction threshold, and comprehensively generate the ballast damage index based on the obtained ballast structure response coefficient and ballast water seepage influence coefficient. The ballast damage index is compared with the damage judgment correction threshold, and the corresponding ballast damage identification result is generated according to different comparison results. The environmental parameters include average rainfall, daily average temperature and daily average humidity.

[0058] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0059] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0060] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0061] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for identifying dirty ballast damage, characterized in that: The specific steps include: Collecting several near-infrared images of the ballast surface with known hydrological characteristic parameters, preprocessing the near-infrared images of the ballast surface to obtain training sample images, mapping the training sample images with corresponding hydrological characteristic parameters one by one to generate a training sample data set, wherein the hydrological characteristic parameters include the ballast surface area, the water surface area and the water accumulation volume, and the preprocessing includes image denoising preprocessing and image enhancement preprocessing; Based on the data in the training sample data set, a neural network prediction model is established, the training sample images in the training sample data set are used as the input of the model, and the hydrological characteristic parameters in the training sample data set are used as labels to train the neural network prediction model to obtain a hydrological characteristic prediction model; Collect near-infrared images of the surface of the ballast to be inspected, pre-process the images and input them into the trained hydrological characteristic prediction model to obtain the predicted values ​​of the hydrological characteristic parameters of the ballast to be inspected, and simultaneously obtain the geometric parameters of the dirt particles on the surface of the ballast to be inspected, the geometric parameters including the average particle size of the dirt particles, the average perimeter of the dirt particles, the average surface area of ​​the dirt particles and the average projected area; According to the geometric parameters of the dirt particles on the ballast surface to be detected, combined with the predicted values ​​of the hydrological characteristic parameters, the ballast water seepage influence coefficient is calculated, the average roughness of the ballast surface and the vibration characteristic parameters of the ballast when the train passes by are obtained, and the ballast structure response coefficient is calculated based on the vibration characteristic parameters and the average roughness, and the vibration characteristic parameters include vibration frequency and amplitude; The environmental parameters of the location of the ballast to be inspected are collected, and the pre-set damage judgment threshold is dynamically corrected according to the obtained environmental parameters to obtain the damage judgment correction threshold. Based on the obtained ballast structure response coefficient and ballast water seepage influence coefficient, a ballast damage index is comprehensively generated. The ballast damage index is compared with the damage judgment correction threshold. According to different comparison results, a corresponding ballast damage identification result is generated. The environmental parameters include average rainfall, daily average temperature and daily average humidity.

2. The dirty ballast damage identification method according to claim 1, characterized in that: Preprocessing each collected near-infrared image of the ballast surface to obtain a training sample image, wherein the preprocessing includes: unifying the image size, image enhancement and denoising preprocessing, wherein the wavelet transform denoising method is used to denoise each near-infrared image of the ballast surface, and bilateral filtering is used to perform image enhancement preprocessing on each near-infrared image of the ballast surface; The method for generating the training sample data set is as follows: mapping the training sample images to the corresponding hydrological characteristic parameters one by one to form a corresponding grid, and recording the formed grid as the training sample data set; Based on the long short-term memory network model LSTM model, a neural network model is established, and the activation function and optimization algorithm are selected. The Tanh function is selected as the activation function, and Adam is selected as the optimization algorithm of the LSTM model; the formula of the Tanh function is: ； In the formula, Represents the Tanh function, independent variable Represents the weighted sum of the neuron's input, that is, the result of weighted summation of the input received by the neuron from the previous layer; At the same time, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of training times, the batch processing number, and the number of hidden layer neurons; The number of network layers is set to 4 layers, the number of iterations is set to 200, the learning rate is set to 0.01, the batch size is set to 32, the number of training times is set to 200, the batch size is set to 256, and the number of hidden layer neurons is 32; The trained hydrological characteristic prediction model takes as input the preprocessed near-infrared image data of the ballast surface, and outputs the predicted values ​​of hydrological characteristic parameters, including the predicted values ​​of the ballast surface area, the water surface area, and the accumulated water volume.

3. The dirty ballast damage identification method according to claim 1, characterized in that: According to the geometric parameters of the dirt particles on the ballast surface to be detected, combined with the predicted values ​​of the hydrological characteristic parameters, the ballast seepage influence coefficient is calculated. The specific formula for calculating the ballast seepage influence coefficient is: ； In the formula, is the influence coefficient of water seepage of the ballast to be tested, and are the predicted values ​​of the surface area and volume of water on the ballast to be tested, respectively. is the shape factor of the dirt particle, is the average particle size of the dirt particles, and the shape factor of the dirt particles The calculation is based on the geometric parameters of the dirt particles, where the shape factor of the dirt particles is The specific formula on which the calculation is based is: ； In the formula, is the average surface area of ​​the dirt particles, is the average circumference of the dirt particles, is the average projected area of ​​the dirt particles.

4. A dirty ballast damage identification method according to claim 3, characterized in that: The average roughness of the ballast surface and the vibration characteristic parameters of the ballast when the train passes by are obtained, and the ballast structure response coefficient is calculated based on the vibration characteristic parameters and the average roughness. The formula for calculating the ballast structure response coefficient is: ； In the formula, is the ballast structure response coefficient, is the average roughness of the ballast surface to be tested, is the vibration frequency, is the vibration amplitude.

5. A dirty ballast damage identification method according to claim 4, characterized in that: The average roughness of the ballast surface to be tested is The specific method of obtaining is: The ballast surface roughness measurement equipment uses a stylus profilometer. Multiple sampling areas are randomly selected from the ballast area to be tested. The stylus profilometer is used to measure and analyze the sampling areas to obtain the surface roughness of each sampling area, and the surface roughness average of all sampling areas is calculated. This average is used as the average roughness of the ballast surface to be tested. The calculation formula is as follows: ； ； In the formula, represents the sampling length of the stylus profilometer in the kth sampling area, k is the index of the sampling area, and , K is the number of sampling areas, is the height of the qth sampling point within the sampling length of the kth sampling area from the center line, q is the coordinate of the sampling point within the sampling length, is the surface roughness of the kth sampling area, is the average roughness of the ballast surface to be tested, the center line and The data is obtained through the built-in software of the stylus profilometer.

6. A dirty ballast damage identification method according to claim 4, characterized in that: Based on the obtained ballast structure response coefficient and ballast water seepage influence coefficient, the ballast damage index is comprehensively generated, and the formula for calculating the ballast damage index is: ； In the formula, is the ballast damage index, and are the weight coefficients of the ballast structure response coefficient and the ballast water seepage influence coefficient to be tested, respectively. , and and Both are greater than 0.

7. A dirty ballast damage identification method according to claim 6, characterized in that: Collect the environmental parameters of the location of the ballast to be inspected, and dynamically correct the pre-set damage judgment threshold according to the obtained environmental parameters to obtain the damage judgment correction threshold. The damage judgment correction threshold is calculated based on the formula: ； In the formula, Correct the threshold for damage judgment, is the pre-set damage judgment threshold. is the average rainfall, specifically the average monthly rainfall, is the average daily humidity, is the daily average temperature, For reference rainfall, is the reference humidity, is the reference temperature; The ballast damage index is compared with the damage judgment correction threshold, and the corresponding ballast damage identification result is generated according to different comparison results. The specific judgment logic is as follows: when When the ballast to be tested is in poor health, it is judged that the ballast to be tested is damaged and should be stopped from use in time for repair; when When the ballast to be inspected is judged to be in good health, it means that the ballast to be inspected is at risk of urban damage and should be inspected and reinforced; when , the health status of the ballast to be tested is judged to be excellent, indicating that there is no damage to the ballast to be tested.

8. A dirty ballast damage identification device, characterized in that: The dirty ballast damage identification device is used to execute the dirty ballast damage identification method according to any one of claims 1 to 7, comprising: A training sample acquisition module is used to acquire several near-infrared images of the ballast surface with known hydrological characteristic parameters, pre-process the near-infrared images of the ballast surface to obtain training sample images, map the training sample images with the corresponding hydrological characteristic parameters one by one, and generate a training sample data set. The hydrological characteristic parameters include the ballast surface area, the water surface area, and the water accumulation volume. The pre-processing includes image noise reduction pre-processing and image enhancement pre-processing; The prediction model training module is used to establish a neural network prediction model based on the data in the training sample data set, take the training sample images in the training sample data set as the input of the model, and take the hydrological characteristic parameters in the training sample data set as labels to train the neural network prediction model to obtain a hydrological characteristic prediction model; A dirt parameter acquisition module is used to acquire near-infrared images of the surface of the ballast to be detected, and input the images into the trained hydrological characteristic prediction model after preprocessing to obtain the predicted values ​​of the hydrological characteristic parameters of the ballast to be detected, and at the same time obtain the geometric parameters of the dirt particles on the surface of the ballast to be detected, wherein the geometric parameters include the average particle size of the dirt particles, the average perimeter of the dirt particles, the average surface area of ​​the dirt particles, and the average projected area; The structural response analysis module is used to calculate the ballast water seepage influence coefficient according to the geometric parameters of the dirt particles on the ballast surface to be detected and the predicted values ​​of the hydrological characteristic parameters, obtain the average roughness of the ballast surface and the vibration characteristic parameters of the ballast when the train passes by, and calculate and generate the ballast structural response coefficient according to the vibration characteristic parameters and the average roughness, wherein the vibration characteristic parameters include vibration frequency and amplitude; The comprehensive damage judgment module is used to collect the environmental parameters of the location of the ballast to be detected, dynamically correct the pre-set damage judgment threshold according to the obtained environmental parameters, obtain the damage judgment correction threshold, and comprehensively generate the ballast damage index based on the obtained ballast structure response coefficient and ballast water seepage influence coefficient. The ballast damage index is compared with the damage judgment correction threshold, and the corresponding ballast damage identification result is generated according to different comparison results. The environmental parameters include average rainfall, daily average temperature and daily average humidity.

Images