Method and system for inspecting railway track based on unmanned aerial vehicle

Abstract

The invention discloses a method and a system for inspecting a railway track based on an unmanned aerial vehicle. The method comprises the following steps: acquiring a multi-dimensional data set of the railway track through the unmanned aerial vehicle for railway inspection; dividing the multi-dimensional data set into track data, roadbed data and surrounding environment data; inputting the track data, the roadbed data and the surrounding environment data into a pre-trained track detection model, a roadbed detection model and an environment detection model respectively, and outputting a corresponding track state, a roadbed state and an environment state in parallel; according to the track data, the roadbed data and the surrounding environment data, respectively calculating a track health index, a roadbed health index and an environment health index of the railway; and according to the track state and the track health index thereof, the roadbed state and the roadbed health index thereof, and the environment state and the environment health index thereof, determining the health degree and the fault degree of the railway so as to realize railway inspection. By utilizing the embodiment of the invention, the inspection efficiency and accuracy can be improved, the state of the railway facility can be updated and monitored in real time, the health degree and the fault degree of the railway can be better judged, and the safe operation of the railway is ensured.

Description

Technical Field

[0001] The present invention belongs to the field of railway technology, and in particular to a method and system for inspecting railway tracks based on unmanned aerial vehicles. Background Art

[0002] With the rapid development of railway transportation, the safety and reliability of railways have received more and more attention. Traditional railway inspection methods mainly rely on manual inspections. Although they have ensured the safety of railways to a certain extent, there are still many shortcomings, which are mainly reflected in the following aspects: Railway lines are usually long and the inspection range is wide. Manual inspections not only require a lot of manpower and material resources, but are also easily affected by factors such as weather, light and terrain. The inspection personnel have high work intensity and low work efficiency, resulting in a long inspection cycle, which is difficult to meet the real-time requirements of railway operation for safety monitoring; manual inspections are limited by the experience and technical level of maintenance personnel, which is prone to missed inspections and wrong inspections, especially in some hidden or difficult-to-reach areas. Traditional detection methods are difficult to detect potential safety hazards, which increases the safety risks of railway operations and may cause accidents; traditional inspection methods usually have high requirements for the timeliness and accuracy of data recording and analysis. However, the data recorded during manual inspections are often scattered and in non-uniform formats. The later sorting and analysis are relatively complicated, making it difficult to form effective data support, which affects the assessment of the status of railway facilities. Summary of the invention

[0003] The purpose of the present invention is to provide a method and system for inspecting railway tracks based on drones to address the deficiencies in the prior art, improve the efficiency and accuracy of inspections, and be able to update and monitor the status of railway facilities in real time, better judge the health and degree of failure of the railway, and ensure the safe operation of the railway.

[0004] An embodiment of the present application provides a method for inspecting railway tracks based on a drone, the method comprising:

[0005] Collecting multidimensional datasets of railway tracks through drones for railway inspection;

[0006] dividing the multidimensional data set into track data, roadbed data and surrounding environment data;

[0007] Input the track data, the roadbed data and the surrounding environment data into a pre-trained track detection model, a roadbed detection model and an environment detection model, respectively, and output the corresponding track status, roadbed status and environment status in parallel;

[0008] Calculating the track health index, roadbed health index and environmental health index of the railway respectively according to the track data, the roadbed data and the surrounding environment data;

[0009] According to the track state and its track health index, the roadbed state and its roadbed health index, and the environment state and its environment health index, the health degree and fault degree of the railway are determined to realize railway inspection.

[0010] Optionally, the calculation formula for calculating the railway track health index is:

[0011]

[0012] Among them, the RTHI is the track health index, the HD is the horizontal deviation of the inspection track, the TD is the vertical deviation of the inspection track, the FD is the track defect frequency of the inspection track, the L is the length of the inspection track, the AD is the parallelism difference between the two sides of the inspection track, the alpha, the beta, and the gamma are corresponding weight coefficients, and the delta and the epsilon are corresponding exponential coefficients.

[0013] Optionally, the calculation formula of the roadbed health index is:

[0014]

[0015] Among them, the SHI is the roadbed health index, the G_s is the soil strength, the H_c is the settlement, the R_c is the drainage capacity score, the D is the roadbed defect frequency of the inspection track, the eta is the maximum health index representing the theoretical optimal health state of the roadbed, the w_1 is the positive influencing factor, and the w_2 is the negative influencing factor.

[0016] Optionally, the calculation formula of the environmental health index is:

[0017]

[0018] Among them, the EHI is the environmental health index, the C_s is the soil stability index, the S is the soil moisture content, the C_r is the degree of environmental desertification, the R is the environmental landslide risk, and the p_1 and p_2 are corresponding weight factors.

[0019] Another embodiment of the present application provides a system for inspecting railway tracks based on a drone, the system comprising:

[0020] A collection module for collecting multidimensional datasets of railway tracks via drones used for railway inspection;

[0021] A partitioning module, used for partitioning the multi-dimensional data set into track data, roadbed data and surrounding environment data;

[0022] An output module, used to input the track data, the roadbed data and the surrounding environment data into a pre-trained track detection model, a roadbed detection model and an environment detection model, respectively, and output the corresponding track status, roadbed status and environment status in parallel;

[0023] A calculation module, used to calculate the track health index, roadbed health index and environmental health index of the railway according to the track data, the roadbed data and the surrounding environment data;

[0024] The determination module is used to determine the health and fault degree of the railway according to the track state and its track health index, the roadbed state and its roadbed health index, and the environmental state and its environmental health index, so as to realize railway inspection.

[0025] Yet another embodiment of the present application provides a storage medium, wherein the storage medium stores a computer program, wherein the computer program is configured to execute any of the above methods when running.

[0026] Yet another embodiment of the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute any of the methods described above.

[0027] Compared with the prior art, the present invention provides a method for inspecting railway tracks based on unmanned aerial vehicles, which collects a multidimensional data set of railway tracks through a unmanned aerial vehicle used for railway inspection; divides the multidimensional data set into track data, roadbed data and surrounding environment data; inputs the track data, the roadbed data and the surrounding environment data into a pre-trained track detection model, a roadbed detection model and an environment detection model, respectively, and outputs the corresponding track status, roadbed status and environment status in parallel; calculates the track health index, roadbed health index and environment health index of the railway according to the track data, the roadbed data and the surrounding environment data; determines the health and fault degree of the railway according to the track status and its track health index, the roadbed status and its roadbed health index, and the environment status and its environment health index, so as to realize railway inspection, thereby improving the efficiency and accuracy of inspection, and can also update and monitor the status of railway facilities in real time, better judge the health and fault degree of the railway, and ensure the safe operation of the railway. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A hardware structure block diagram of a computer terminal for a method for inspecting railway tracks based on an unmanned aerial vehicle provided in an embodiment of the present invention;

[0029] Figure 2A schematic diagram of a flow chart of a method for inspecting railway tracks using a drone provided in an embodiment of the present invention;

[0030] Figure 3 A schematic diagram of the structure of a system for inspecting railway tracks based on drones provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, but should not be construed as limiting the present invention.

[0032] The embodiment of the present invention first provides a method for inspecting railway tracks based on a drone, which can be applied to electronic devices such as computer terminals, specifically ordinary computers, etc.

[0033] The following describes it in detail by taking running on a computer terminal as an example. Figure 1 The hardware structure block diagram of a computer terminal for a method of inspecting railway tracks based on an unmanned aerial vehicle is provided in an embodiment of the present invention. Figure 1 As shown, the computer terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. Optionally, the computer terminal may also include a transmission device 106 for communication functions and an input and output device 108. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0034] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / modules corresponding to the method for inspecting railway tracks based on drones in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, implementing the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0035] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of a computer terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0036] See also Figure 2 , an embodiment of the present invention provides a method for inspecting railway tracks based on a drone, which may include the following steps:

[0037] S201, Collecting multidimensional datasets of railway tracks via drones for railway inspection;

[0038] In this step, drones designed for railway inspection are used to collect multi-dimensional data of railway tracks in an efficient and accurate manner. Drones are equipped with a variety of sensors (such as high-definition cameras, lidar, infrared sensors, etc.), which can collect relevant data on the physical state of the track and the surrounding environment during flight. These data include but are not limited to the geometry of the track, the stability of the roadbed, environmental conditions, etc., forming a comprehensive multi-dimensional data set. This step is crucial to ensure the safety and reliability of railway operations. The use of drones not only improves inspection efficiency and reduces the time and labor costs required for manual inspections, but also can perform tasks in complex and dangerous working environments, effectively reducing personnel risks. Through comprehensive data collection, it can provide a basis for subsequent data analysis and health assessment, and timely identify potential failure risks, thereby providing a scientific basis for railway maintenance.

[0039] Specifically, before data collection, the optimal drone inspection route can be generated by using algorithms through geographic information system (GIS) and track data modeling. This route should include all key parts of the track and take into account flight altitude and speed to maximize the coverage and accuracy of data collection. Integrate multiple sensors onto drones and calibrate them before flight to ensure the consistency and accuracy of sensor data. The combination of multiple sensors can effectively make up for the shortcomings of a single sensor. For example, the combination of visual sensors and lidar can obtain robust data under different lighting and climatic conditions. During the inspection process, the drone records the collected data in real time and stores it in a local or cloud database. This process involves the management of data streams, including data compression, encryption and classification. Through advanced storage technology, drones can ensure the integrity and security of data.

[0040] S202, dividing the multidimensional data set into track data, roadbed data and surrounding environment data;

[0041] In this step, the collected multidimensional data sets need to be carefully processed and analyzed to effectively divide them into three categories: track data, roadbed data, and surrounding environment data. Each type of data represents a different focus in railway inspection, ensuring that the subsequent detection model can analyze and evaluate the health status of the track and possible failure risks in a targeted manner. This process involves the use of data classification algorithms to establish the categories of data and label the corresponding features.

[0042] This classification is extremely important for subsequent analysis and evaluation, because it helps technicians and systems quickly locate problems and take appropriate maintenance measures. Through clear classification, data processing efficiency can be improved, ensuring that the detection model can make full use of dedicated data during analysis, thereby improving accuracy and reliability. In addition, this step also lays a good foundation for data visualization and decision support.

[0043] Specifically, we can first implement a preliminary analysis of the data, including an understanding of the data dimensions, formats, scope, and types. This step will involve statistical analysis techniques (such as pivot tables and descriptive statistics) to obtain an overview of the data and identify possible features and indicators. Based on the results of the data analysis, a specific feature template is constructed, including the key features of track data, roadbed data, and surrounding environmental data. For example, the features of track data may include the geometry of the track, horizontal and vertical deviations, etc.; roadbed data may focus on soil strength, settlement, etc.; and surrounding environmental data include soil moisture, meteorological conditions, etc. Use classification algorithms in machine learning (such as random forests and support vector machines) to label and classify data, and create models to distinguish between track, roadbed, and environmental data. This process requires a large amount of labeled data for training to improve the accuracy of the classification model. For each data item, the model will automatically determine its category based on the feature template.

[0044] S203, inputting the track data, the roadbed data and the surrounding environment data into a pre-trained track detection model, a roadbed detection model and an environment detection model, respectively, and outputting the corresponding track status, roadbed status and environment status in parallel;

[0045] In this step, different categories of data (i.e. track data, roadbed data, and surrounding environment data) are input into targeted training models for analysis. These models are built based on a large amount of historical data and features to identify and evaluate different states. The track detection model focuses on evaluating the geometry and stability of the track itself; the roadbed detection model focuses on the stability and bearing capacity of the soil; and the environmental detection model analyzes various risk factors in the surrounding environment. Through parallel processing, efficient data analysis can be achieved and the state results corresponding to each data category can be quickly obtained.

[0046] The implementation of this step has greatly improved the automation and intelligence level of railway inspection, and reduced the subjectivity and errors of manual evaluation. By using deep learning and data mining technology, the model can quickly identify potential problems in complex environments and provide timely status feedback. This not only improves inspection efficiency, but also significantly reduces safety hazards and accident risks, providing a scientific basis for the safety management of railway operations.

[0047] Specifically, first, the divided track data, roadbed data, and surrounding environment data can be processed and standardized to ensure that the characteristics of different data sets are within the same order of magnitude for model analysis. This process usually includes data cleaning, missing value filling, and normalization, which can effectively improve the training effect and prediction accuracy of the model.

[0048] Using the ensemble learning method, three sub-models (track detection model, roadbed detection model, and environment detection model) are designed. Each model can use different algorithms. For example, the track detection model can use a convolutional neural network (CNN) to focus on feature extraction of image data; the roadbed detection model can use a random forest to process numerical roadbed features; and the environment detection model can use a long short-term memory network (LSTM) for time series data analysis. The parallel architecture design allows the three models to run independently, thereby speeding up the overall data processing speed.

[0049] Use the labeled training data set to train the three models separately. Through cross-validation and hyperparameter tuning methods, the performance of each model is continuously optimized. The overall performance of the integrated model can be achieved through weighted voting or averaging to ensure that the final output state is more accurate. This process can also introduce transfer learning to utilize existing pre-trained models to accelerate training speed and improve results.

[0050] In the data processing stage, parallel reasoning of data is realized, that is, the standardized data is input into each model for status assessment. Each model runs simultaneously and generates corresponding output results (track status, roadbed status and environmental status). After completion, data fusion technology (such as weighted average or Bayesian fusion method) can also be used to integrate model outputs to eliminate possible model bias and obtain a more accurate comprehensive status assessment.

[0051] S204, respectively calculating a track health index, a roadbed health index, and an environmental health index of the railway according to the track data, the roadbed data, and the surrounding environment data;

[0052] In this step, three health indexes are calculated according to the pre-set calculation formula using the previously collected track data, roadbed data and surrounding environment data. These health indices are designed to quantify the health status of each aspect and help technicians intuitively understand the overall status of the railway. The track health index focuses on the geometry and structural integrity of the track; the roadbed health index evaluates the bearing capacity and stability of the soil; and the environmental health index reflects the impact of the surrounding environment on railway safety. This quantitative method can provide a scientific basis for subsequent inspection and maintenance decisions.

[0053] Specifically, a calculation formula for calculating the track health index of a railway may be:

[0054]

[0055] Among them, the RTHI is the track health index, and the HD is the horizontal deviation of the inspection track, which reflects the geometric integrity of the track. The larger the value, the more the track deviates from the ideal state. The TD is the vertical deviation of the inspection track, which plays an important indicative role especially when the track is deformed or settled. The FD is the track defect frequency of the inspection track, which indicates the number of track defects per unit length, and directly affects riding comfort and driving safety. The L is the length of the inspection track, which is used as a normalization factor to avoid the influence of the track defect frequency on the calculation of the health index due to the different track lengths. The AD is the difference in parallelism on both sides of the inspection track, which affects the stability of the train. The alpha, beta, and gamma are corresponding weight coefficients, which represent the importance factors of different parameters in the calculation of the health index, and can be determined by analyzing historical data or expert experience. The delta and epsilon are corresponding exponential coefficients, which are adjustment factors for nonlinear responses, making them more sensitive to the influence of certain features.

[0056] The Track Health Index (RTHI) is designed to assess the overall health of the track by taking into account the geometric and physical characteristics of the track. The design of the formula takes into account multiple influencing factors, so that the health index can accurately reflect the track condition. HD and TD can be measured and analyzed in real time by laser ranging sensors, high-definition cameras and image processing technology carried by drones; FD can be calculated by recording and counting the track defects found during inspections; L is the known length of the inspection path, which can be obtained through the navigation system of the drone; AD can be obtained by measuring both sides of the track and calculating the difference.

[0057] Specifically, a calculation formula for the roadbed health index can be:

[0058]

[0059] Among them, the SHI is the roadbed health index, the G_s is the soil strength, which reflects the bearing capacity of the roadbed soil and affects the stability and safety of the roadbed. The H_c is the settlement, which refers to the vertical settlement of the roadbed. Excessive settlement will cause the track position to change, affecting driving safety. The R_c is the drainage capacity score, which is an indicator of the drainage capacity of the roadbed. Good drainage capacity can reduce soil softening and settlement. The D is the roadbed defect frequency of the inspection track, which is similar to the track defect. It describes the number of defects in the roadbed and affects the overall safety. The eta is the maximum health index representing the theoretical optimal health state of the roadbed, which represents the theoretical optimal health state of the roadbed and is usually defined by the historical best data. The w_1 is a positive influencing factor, and the w_2 is a negative influencing factor, which respectively describes the positive and negative influence of the roadbed characteristics on the health index, and can be adjusted through historical data. The roadbed health index (SHI) aims to comprehensively evaluate the bearing capacity and stability of the roadbed to ensure the safety of railway operation.

[0060] Specifically, a calculation formula for an environmental health index can be:

[0061]

[0062] Among them, the EHI is the environmental health index, and the C_s represents the soil stability index, which reflects the stability of the soil under the railway structure and affects the safety of the roadbed. The S is the soil moisture content. High moisture content may cause soil softening, slippage and other problems, affecting stability. The C_r is the degree of environmental desertification, which describes the degree of desertification in the surrounding environment. Severe desertification may affect the surrounding soil structure. The R is the environmental landslide risk, an indicator for assessing the possibility of landslides, which poses a direct threat to railway safety. The p_1 and p_2 are corresponding weight factors. The relative impact of different environmental factors on the health index needs to be set through historical data and expert knowledge. The Environmental Health Index (EHI) is used to quantify the impact of the surrounding environment on railway operations and comprehensively consider the potential threat of environmental factors to railway safety.

[0063] The process of calculating the health index can not only help railway operation managers identify potential risks, but also monitor the long-term health trend of the railway. By comparing historical health index data, maintenance effects can be evaluated and more effective maintenance strategies can be formulated, thereby improving the safety and stability of railway operations. In addition, the health index provides a clear basis for data-driven decision-making and promotes the in-depth integration of drone inspection technology and traditional railway maintenance management.

[0064] S205, determining the health and fault degree of the railway according to the track state and its track health index, the roadbed state and its roadbed health index, and the environment state and its environment health index, so as to realize railway inspection.

[0065] In the railway inspection method, drones are used to collect and analyze the status and health index of tracks, roadbeds and the environment, providing a systematic evaluation tool for the safe operation of railways. Specifically, the steps to determine the health and fault level of railways mainly include the following aspects:

[0066] Comparing the track status, roadbed status and environmental status with the corresponding health index can intuitively reflect the overall health status of the railway. This includes a detailed analysis of the specific conditions of each part, such as whether the track is straight, whether the roadbed has settled, whether there is a landslide risk in the environment, etc.; by combining the track health index (RTHI), roadbed health index (SHI) and environmental health index (EHI), a comprehensive health assessment model is formed. This model can not only evaluate the health status of each part, but also integrate the influence of different parts through weighted average or other algorithms to obtain an overall health index; based on the above assessment, the system can identify possible fault areas. For example, if the track health index is low and the roadbed health index shows settlement, the system will prompt the relevant departments to conduct further inspections and maintenance; based on the analysis results, railway managers can make data-driven decisions, including the priority of repair plans, the adjustment of inspection frequency, and the reallocation of resources to ensure the safety and stability of the railway system.

[0067] Through regular and real-time health assessments, potential safety hazards can be discovered and resolved early, effectively reducing the risk of accidents; quantitative analysis of health status and fault level can help railway managers rationally allocate funds and human resources and avoid unnecessary expenditures; timely maintenance and inspections can prevent small faults from developing into big problems, thereby extending the service life of tracks and roadbeds and reducing overall maintenance costs; through drone inspections and data analysis, the efficiency and accuracy of inspections are improved in a data-driven manner, which greatly saves time and manpower compared to traditional manual inspections.

[0068] Exemplarily, a track status classification and track health index range may include:

[0069] - Health status: RTHI ≥ 0.8;

[0070] -Minor fault state: 0.6≤RTHI<0.8;

[0071] - Moderate fault state: 0.4≤RTHI<0.6;

[0072] - Severe fault condition: RTHI < 0.4.

[0073] A classification of roadbed conditions and roadbed health index range may include:

[0074] -Health status: SHI>>0.8;

[0075] -Minor fault state: 0.6<<SHI<0.8;

[0076] - Moderate fault state: 0.4<<SHI<0.6;

[0077] - Severe fault state: SHI < 0.4;

[0078] A classification of environmental conditions and range of environmental health indices may include:

[0079] -Health status: EHI>>0.8;

[0080] -Minor fault state: 0.6<<EHI<0.8;

[0081] - Moderate fault state: 0.4<<EHI<0.6;

[0082] - Severe fault status: EHI < 0.4.

[0083] Collect the health index of the track, roadbed and environment. Determine the health of the track, roadbed and environment based on the classification of each state. If any part is in a "serious fault state", the overall health level should be marked as "serious fault"; if there is no "serious fault state" and any part is in a "moderate fault state", the overall health level should be marked as "moderate fault"; if there is no "serious fault state" and "moderate fault state", any part is in a "mild fault state", the overall health level should be marked as "mild fault"; if all parts are in a "healthy state", the overall health level is marked as "healthy".

[0084] Assume that after a drone inspection, the following health index is obtained:

[0085] - Track Health Index (RTH I) = 0.55;

[0086] - Roadbed health index (SHI) = 0.75;

[0087] - Environmental Health Index (EHI) = 0.65.

[0088] 1. Track status assessment: RTH I = 0.55, which is a “moderate fault status”;

[0089] 2. Roadbed condition assessment: SHI = 0.75, which is a "minor fault condition";

[0090] 3. Environmental status assessment: EH I = 0.65, which is a “minor fault state”.

[0091] Result integration: The track is in "moderate fault state", the roadbed is in "minor fault state", and the environment is in "minor fault state". Since the health state of the track is moderate fault, the overall health state is "moderate fault".

[0092] In this case, although the health of the roadbed and environment is good, the overall railway health is judged to be "moderate failure" due to the "moderate failure state" of the track. This prompts the railway management department to conduct further inspection and maintenance on this section of track to ensure safe operation, and should not ignore the comprehensive monitoring and maintenance of other parts. This classification and assessment method provides clear guidance for railway operation management and helps to prioritize areas at risk.

[0093] It can be seen that a multidimensional data set of railway tracks is collected by using drones for railway inspection; the multidimensional data set is divided into track data, roadbed data and surrounding environment data; the track data, the roadbed data and the surrounding environment data are respectively input into a pre-trained track detection model, a roadbed detection model and an environment detection model, and the corresponding track status, roadbed status and environment status are output in parallel; the track health index, roadbed health index and environment health index of the railway are respectively calculated based on the track data, the roadbed data and the surrounding environment data; the health and fault degree of the railway are determined based on the track status and its track health index, the roadbed status and its roadbed health index, and the environment status and its environment health index, so as to realize railway inspection, thereby improving the efficiency and accuracy of inspection, and being able to update and monitor the status of railway facilities in real time, better judge the health and fault degree of the railway, and ensure the safe operation of the railway.

[0094] Another embodiment of the present invention provides a system for inspecting railway tracks based on drones, see Figure 3 , the system may include:

[0095] A collection module 301 is used to collect a multi-dimensional data set of railway tracks by using a drone used for railway inspection;

[0096] A division module 302, for dividing the multi-dimensional data set into track data, roadbed data and surrounding environment data;

[0097] An output module 303 is used to input the track data, the roadbed data and the surrounding environment data into a pre-trained track detection model, a roadbed detection model and an environment detection model, respectively, and output the corresponding track status, roadbed status and environment status in parallel;

[0098] A calculation module 304 is used to calculate the track health index, the roadbed health index and the environmental health index of the railway according to the track data, the roadbed data and the surrounding environment data;

[0099] The determination module 305 is used to determine the health and fault degree of the railway according to the track state and its track health index, the roadbed state and its roadbed health index, and the environment state and its environment health index, so as to realize railway inspection.

[0100] It can be seen that a multidimensional data set of railway tracks is collected by using drones for railway inspection; the multidimensional data set is divided into track data, roadbed data and surrounding environment data; the track data, the roadbed data and the surrounding environment data are respectively input into a pre-trained track detection model, a roadbed detection model and an environment detection model, and the corresponding track status, roadbed status and environment status are output in parallel; the track health index, roadbed health index and environment health index of the railway are respectively calculated based on the track data, the roadbed data and the surrounding environment data; the health and fault degree of the railway are determined based on the track status and its track health index, the roadbed status and its roadbed health index, and the environment status and its environment health index, so as to realize railway inspection, thereby improving the efficiency and accuracy of inspection, and being able to update and monitor the status of railway facilities in real time, better judge the health and fault degree of the railway, and ensure the safe operation of the railway.

[0101] An embodiment of the present invention further provides a storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when running.

[0102] Specifically, in this embodiment, the above storage medium may be configured to store a computer program for performing the following steps:

[0103] S201, Collecting multidimensional datasets of railway tracks through drones for railway inspection;

[0104] S202, dividing the multidimensional data set into track data, roadbed data and surrounding environment data;

[0105] S203, inputting the track data, the roadbed data and the surrounding environment data into a pre-trained track detection model, a roadbed detection model and an environment detection model, respectively, and outputting the corresponding track status, roadbed status and environment status in parallel;

[0106] S204, respectively calculating a track health index, a roadbed health index, and an environmental health index of the railway according to the track data, the roadbed data, and the surrounding environment data;

[0107] S205, determining the health and fault degree of the railway according to the track state and its track health index, the roadbed state and its roadbed health index, and the environment state and its environment health index, so as to realize railway inspection.

[0108] It can be seen that a multidimensional data set of railway tracks is collected by using drones for railway inspection; the multidimensional data set is divided into track data, roadbed data and surrounding environment data; the track data, the roadbed data and the surrounding environment data are respectively input into a pre-trained track detection model, a roadbed detection model and an environment detection model, and the corresponding track status, roadbed status and environment status are output in parallel; the track health index, roadbed health index and environment health index of the railway are respectively calculated based on the track data, the roadbed data and the surrounding environment data; the health and fault degree of the railway are determined based on the track status and its track health index, the roadbed status and its roadbed health index, and the environment status and its environment health index, so as to realize railway inspection, thereby improving the efficiency and accuracy of inspection, and being able to update and monitor the status of railway facilities in real time, better judge the health and fault degree of the railway, and ensure the safe operation of the railway.

[0109] An embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0110] Specifically, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0111] Specifically, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0112] S201, Collecting multidimensional datasets of railway tracks through drones for railway inspection;

[0113] S202, dividing the multidimensional data set into track data, roadbed data and surrounding environment data;

[0114] S203, inputting the track data, the roadbed data and the surrounding environment data into a pre-trained track detection model, a roadbed detection model and an environment detection model, respectively, and outputting the corresponding track status, roadbed status and environment status in parallel;

[0115] S204, respectively calculating a track health index, a roadbed health index, and an environmental health index of the railway according to the track data, the roadbed data, and the surrounding environment data;

[0116] S205, determining the health and fault degree of the railway according to the track state and its track health index, the roadbed state and its roadbed health index, and the environment state and its environment health index, so as to realize railway inspection.

[0117] It can be seen that a multidimensional data set of railway tracks is collected by using drones for railway inspection; the multidimensional data set is divided into track data, roadbed data and surrounding environment data; the track data, the roadbed data and the surrounding environment data are respectively input into a pre-trained track detection model, a roadbed detection model and an environment detection model, and the corresponding track status, roadbed status and environment status are output in parallel; the track health index, roadbed health index and environment health index of the railway are respectively calculated based on the track data, the roadbed data and the surrounding environment data; the health and fault degree of the railway are determined based on the track status and its track health index, the roadbed status and its roadbed health index, and the environment status and its environment health index, so as to realize railway inspection, thereby improving the efficiency and accuracy of inspection, and being able to update and monitor the status of railway facilities in real time, better judge the health and fault degree of the railway, and ensure the safe operation of the railway.

[0118] The above describes in detail the structure, features and effects of the present invention based on the embodiments shown in the drawings. The above is only a preferred embodiment of the present invention, but the present invention is not limited to the scope of implementation shown in the drawings. Any changes made according to the concept of the present invention, or modifications to equivalent embodiments with equivalent changes, which still do not exceed the spirit covered by the description and drawings, should be within the protection scope of the present invention.

Claims

1. A method for inspecting railway tracks based on drones, characterized in that: The method comprises: Collecting multidimensional datasets of railway tracks through drones for railway inspection; dividing the multidimensional data set into track data, roadbed data and surrounding environment data; Input the track data, the roadbed data and the surrounding environment data into a pre-trained track detection model, a roadbed detection model and an environment detection model, respectively, and output the corresponding track status, roadbed status and environment status in parallel; Calculating the track health index, roadbed health index and environmental health index of the railway respectively according to the track data, the roadbed data and the surrounding environment data; According to the track state and its track health index, the roadbed state and its roadbed health index, and the environment state and its environment health index, the health degree and fault degree of the railway are determined to realize railway inspection.

2. The method according to claim 1, characterized in that The calculation formula for calculating the railway track health index is: Among them, the RTHI is the track health index, the HD is the horizontal deviation of the inspection track, the TD is the vertical deviation of the inspection track, the FD is the track defect frequency of the inspection track, the L is the length of the inspection track, the AD is the parallelism difference between the two sides of the inspection track, the alpha, the beta, and the gamma are corresponding weight coefficients, and the delta and the epsilon are corresponding exponential coefficients.

3. The method according to claim 2, characterized in that The calculation formula of the roadbed health index is: Among them, the SHI is the roadbed health index, the G_s is the soil strength, the H_c is the settlement, the R_c is the drainage capacity score, the D is the roadbed defect frequency of the inspection track, the eta is the maximum health index representing the theoretical optimal health state of the roadbed, the w_1 is the positive influencing factor, and the w_2 is the negative influencing factor.

4. The method according to claim 3, characterized in that The calculation formula of the environmental health index is: Among them, the EHI is the environmental health index, the C_s is the soil stability index, the S is the soil moisture content, the C_r is the degree of environmental desertification, the R is the environmental landslide risk, and the p_1 and p_2 are corresponding weight factors.

5. A system for inspecting railway tracks based on drones, characterized in that: The system comprises: A collection module for collecting multidimensional datasets of railway tracks via drones used for railway inspection; A partitioning module, used for partitioning the multi-dimensional data set into track data, roadbed data and surrounding environment data; An output module, used to input the track data, the roadbed data and the surrounding environment data into a pre-trained track detection model, a roadbed detection model and an environment detection model, respectively, and output the corresponding track status, roadbed status and environment status in parallel; A calculation module, used to calculate the track health index, roadbed health index and environmental health index of the railway according to the track data, the roadbed data and the surrounding environment data; The determination module is used to determine the health and fault degree of the railway according to the track state and its track health index, the roadbed state and its roadbed health index, and the environmental state and its environmental health index, so as to realize railway inspection.

6. The system according to claim 5, characterized in that The calculation formula for calculating the railway track health index is: Among them, the RTHI is the track health index, the HD is the horizontal deviation of the inspection track, the TD is the vertical deviation of the inspection track, the FD is the track defect frequency of the inspection track, the L is the length of the inspection track, the AD is the parallelism difference between the two sides of the inspection track, the alpha, the beta, and the gamma are corresponding weight coefficients, and the delta and the epsilon are corresponding exponential coefficients.

7. The system according to claim 6, characterized in that The calculation formula of the roadbed health index is: Among them, the SHI is the roadbed health index, the G_s is the soil strength, the H_c is the settlement, the R_c is the drainage capacity score, the D is the roadbed defect frequency of the inspection track, the eta is the maximum health index representing the theoretical optimal health state of the roadbed, the w_1 is the positive influencing factor, and the w_2 is the negative influencing factor.

8. The system according to claim 7, characterized in that The calculation formula of the environmental health index is: Among them, the EHI is the environmental health index, the C_s is the soil stability index, the S is the soil moisture content, the C_r is the degree of environmental desertification, the R is the environmental landslide risk, and the p_1 and p_2 are corresponding weight factors.

9. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 4 when executed.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 4.

Images