Rail train anti-collision identification device and method in unmanned state

By integrating multiple sensors and machine learning algorithms on rail trains, a panoramic perception system is built, which solves the problem that rail trains are difficult to detect and avoid foreign objects in real time, and realizes efficient and safe operation of rail trains.

CN118918552AActive Publication Date: 2024-11-08SOUTHWEST JIAOTONG UNIV

Patent Information

Application Number
CN202410961883.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-18
Publication Date
2024-11-08
Estimated Expiration
2044-07-18

AI Technical Summary

Technical Problem

The prior art is difficult to detect and avoid foreign objects in rail transit in real time, resulting in limited detection range, insufficient real-time performance, high false alarm rate, and manual monitoring cannot meet the high efficiency and reliability of modern railway systems. need.

Method used

Using the combination of data acquisition system, intelligent detection system and positioning system, the railway network big data detection cluster is built through vehicle-mounted sensors such as lidar, thermal imager, camera, acceleration sensor, etc., combined with machine learning artificial intelligence algorithms, to achieve the goal of Panoramic perception and automatic obstacle avoidance decisions for rail train operation scenarios.

Benefits of technology

It realizes the full-scene detection capability of rail trains at high speed operation, improves detection accuracy and response speed, enhances the automation and intelligence level of the system, and meets the needs of modern railway transportation for high safety and efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118918552A_ABST
    Figure CN118918552A_ABST
Patent Text Reader

Abstract

The invention provides an anti-collision recognition device and method for a rail train in an unmanned driving state, and relates to the technical field of rail traffic tests.The anti-collision recognition device comprises a data acquisition system, an intelligent detection system and a positioning system, a cab structure is installed on one side of a train body, and the positioning system is arranged on an underframe edge beam; three-dimensional imaging data around a train body is obtained and stored to form a complete railway network big data detection cluster; training the railway network big data detection cluster through an artificial intelligence machine learning algorithm to obtain a railway train invader judgment and prediction model; the real-time invader is detected, and the detection result is comprehensively judged; therefore, pre-judgment and response to the invader of the railway system are realized. The technical problem that an existing high-speed train is difficult to detect foreign matter invasion in real time and avoid obstacles autonomously is effectively solved, and an important technical support and an implementation approach are provided for the full-scene detection capability of the railway vehicle under high-speed operation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of rail transit test technology, and in particular to a device and method for anti-collision identification of a rail train in an unmanned driving state. Background Art

[0002] With the rapid development of railway transportation systems, train operation safety has become a focus of attention in the railway industry. Current technologies mainly rely on traditional sensors and manual monitoring methods to detect intrusions in train operation, which have problems such as limited detection range, lack of real-time performance, and high false alarm rate. For example, traditional track detection systems may use simple switch sensors to detect foreign objects on the track, but these systems can often only detect objects that directly contact the track, and are powerless against non-contact or smaller intrusions.

[0003] In addition, manual monitoring is limited by human resources, visual blind spots and response time, and cannot meet the needs of modern railway systems for high efficiency and high reliability. Existing technologies are also relatively backward in data processing and analysis, and usually lack effective algorithms to extract useful information from large amounts of sensor data, resulting in inaccurate classification, positioning and risk assessment of intrusions. Summary of the invention

[0004] The purpose of the present invention is to provide a device and method for anti-collision identification of railway trains in an unmanned driving state to improve the above-mentioned problem. In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is as follows:

[0005] In a first aspect, the present application provides a rail train anti-collision identification device in an unmanned driving state, comprising:

[0006] Data acquisition system, intelligent detection system and positioning system, wherein the data acquisition system includes a vehicle body, a driver's cab structure is installed on one side of the vehicle body, a positioning system is arranged on the side beam of the vehicle body chassis, and bogies are installed on the left and right ends of the lower side of the vehicle body respectively, wherein the positioning system includes a GPS positioning system and three-dimensional digital map imaging technology, and the intelligent detection system includes an artificial intelligence detection algorithm of machine learning; a zoom camera is installed inside the windshield of the driver's cab structure, a thermal imager is installed outside the windshield, and a pitot tube is installed on the lower side of the thermal imager, wherein the thermal imager is used to detect invading organisms with thermal effects, thereby realizing real-time imaging; wherein the pitot tube is used to measure the temperature, pressure and speed outside the vehicle and serves as a needle Early warning of extreme weather and as input for safe train operation; the zoom camera is used to detect imaging less than 500 meters, thereby realizing detection of intrusive objects and data transmission; a telephoto camera is installed above the thermal imager, a pressure sensor is installed on the skin plate of the coupler opening and closing mechanism, a snowplow is installed on the lower side of the opening and closing mechanism, and a laser radar is installed on the outside of the opening and closing mechanism and the skirt, wherein the laser radar is used to detect and scan the surrounding scenes of the rail vehicle, build a three-dimensional map, and realize detection and early warning of different objects approaching the train; the telephoto camera is used to achieve imaging greater than 500 meters, thereby realizing detection of intrusive objects and data transmission, wherein the GPS positioning system can also be a Beidou navigation and positioning system.

[0007] Preferably, the bogie comprises a wheelset, a frame, an axle box cover, a first tie spring and a second tie spring, a first acceleration sensor, a second acceleration sensor and an angular momentum encoder are arranged on the axle box cover, and a frame is arranged between the two wheelsets;

[0008] The first acceleration sensor, the second acceleration sensor and the angular momentum encoder are used to monitor and identify the foreign objects crushing the track, collect the vibration of the wheels in real time, and transmit the impact signal to slow down the train or perform emergency braking.

[0009] Preferably, a front camera and a rear camera are installed above the vehicle body; wherein the front camera and the rear camera are used for intrusion detection and imaging of falling objects from high altitudes.

[0010] Preferably, a storage exchange system is arranged below the driver's cab structure, wherein the storage exchange system includes a data exchanger, a data storage device, a port and a computer; the data exchanger is used to forward the data acquisition system to the correct port by identifying the destination address in the data acquisition system, thereby realizing rapid data exchange; the data storage device is used to store the equipment for collecting and saving data from unmanned rail trains; the port is used to distinguish different services or processes, wherein each network connection needs to specify a source port and a target port; the computer is used to perform computing tasks and process data.

[0011] Preferably, an opening and closing device is provided below the data exchanger and the data storage device.

[0012] In a second aspect, the present application also provides a method for anti-collision identification of a rail train in an unmanned driving state, comprising the following steps:

[0013] The laser radar is used to obtain three-dimensional imaging data around the train body, and other on-board sensors are used to detect and determine potential intrusions to obtain initial data on the train operation scene.

[0014] The initial data of the train operation scenario is accumulated as initial data, and combined with the collected historical detection data, it is stored to form a complete railway network big data detection cluster;

[0015] The railway network big data detection cluster is trained through artificial intelligence machine learning algorithms to obtain a railway train intrusion judgment prediction model, where the model is used to classify intrusions, determine their locations, and predict the degree of danger to train operation;

[0016] Based on the railway train intrusion judgment and prediction model, real-time intrusion detection is carried out and the detection results are comprehensively judged;

[0017] According to the comprehensive judgment results, the corresponding train operation instructions are output to achieve the prediction and response to the intrusions in the railway system. Specifically, the prediction and judgment accuracy of the intrusion judgment prediction model will increase significantly as the train mileage increases (in practice, trains on various lines across the country are always running, and the training data of the railway network big data detection cluster is always increasing).

[0018] In a third aspect, the present application also provides a rail train anti-collision identification device in an unmanned driving state, comprising:

[0019] Memory for storing computer programs;

[0020] A processor is used to implement the steps of the method for anti-collision identification of rail trains in an unmanned driving state when executing the computer program.

[0021] In a fourth aspect, the present application further provides a readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned method for anti-collision identification of rail trains in an unmanned driving state are implemented.

[0022] The beneficial effects of the present invention are:

[0023] The present invention combines a data acquisition system, an intelligent detection system and a positioning system to realize the intercommunication between data and the early warning of train intrusions and the automatic control of trains. The method and system for anti-collision identification of rail trains in an unmanned driving state have active and semi-active environmental perception capabilities, which are of great significance for detecting static and dynamic obstacles on and beside the track, realizing hazard assessment, and fully realizing autonomous operation of rail trains, and play an indispensable key role in ensuring the safe operation of trains.

[0024] The purpose of the present invention is to provide a method and system for anti-collision identification of rail trains in an unmanned driving state, which can effectively solve the technical problem that existing high-speed trains are difficult to detect foreign object intrusion and autonomously avoid obstacles in real time, and provide an important technical support and implementation path for the full-scene detection capability of rail vehicles under high-speed operation.

[0025] The method and system used in the present invention are one of the key technologies for rail vehicles to achieve full-scene perception during operation. When a train is invaded by foreign objects during operation, machine learning artificial intelligence technology is used to perform data training, identification and analysis through on-board sensors and various detection equipment, so as to improve the rail train's environmental perception ability and issue instructions for the train to avoid collisions, thereby protecting the train's safety during operation. Therefore, the rail vehicle data acquisition system, intelligent detection system and rail vehicle positioning system are integrated, and artificial intelligence and big data are introduced to achieve full-scene perception capabilities during rail vehicle operation, and obtain automatic driving functions for track foreign object intrusion detection and train automatic obstacle avoidance decision-making.

[0026] The system in the present invention collects three-dimensional data of train operation scenes in real time through on-board laser radar and other advanced sensors, builds a comprehensive railway network big data detection cluster, and uses advanced data processing technology and machine learning algorithms to accurately classify, locate and assess the risks of detected intrusions. In addition, the present application also includes an adaptive control strategy that can dynamically adjust train operation instructions based on real-time analysis results, such as slowing down, emergency braking or opening a snowplow to ensure the safety of train operation. In this way, the present application not only improves the accuracy and response speed of detection, but also enhances the automation and intelligence level of the system, meeting the needs of modern railway transportation for high safety and high efficiency.

[0027] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by implementing the embodiments of the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0029] Figure 1 It is a schematic structural diagram of a rail train anti-collision identification device in an unmanned driving state according to an embodiment of the present invention;

[0030] Figure 2 It is a schematic flow chart of a method for anti-collision identification of a rail train in an unmanned driving state described in an embodiment of the present invention;

[0031] Figure 3 It is a schematic diagram of the structure of the anti-collision identification device for a rail train in an unmanned driving state described in an embodiment of the present invention.

[0032] In the figure: 1. track; 2. intrusion; 3. pressure sensor; 4. airspeed tube; 5. thermal imager; 6. zoom camera; 7. telephoto camera; 8. vehicle body; 9. camera on the front of the vehicle; 10. camera on the rear of the vehicle; 11. GPS positioning system; 12. laser radar; 13. angular momentum encoder; 14. first acceleration sensor; 15. second acceleration sensor; 16. opening and closing device; 17. obstacle remover; 18. wheelset; 19. axle box cover; 20. first spring; 21. frame; 22. second spring; 23. data exchanger; 24. data storage device; 25. port; 26. computer; 27. bogie; 28. skirt; 29. ​​driver's cab structure; 800. anti-collision identification equipment for rail trains in unmanned driving state; 801. processor; 802. memory; 803. multimedia component; 804. I / O interface; 805. communication component. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0034] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0035] Embodiment 1:

[0036] like Figure 1 As shown, this embodiment provides a rail train anti-collision identification device in an unmanned driving state, see Figure 1 The device comprises:

[0037] A data acquisition system, an intelligent detection system and a positioning system, wherein the data acquisition system includes a vehicle body 8, a driver's cab structure 29 is installed on one side of the vehicle body 8, a positioning system is arranged on the side beam of the chassis of the vehicle body 8, and bogies 27 are installed on the left and right ends of the lower side of the vehicle body 8, wherein the positioning system includes a GPS positioning system 11 and a three-dimensional digital map imaging technology, and the intelligent detection system includes an artificial intelligence detection algorithm of machine learning; a zoom camera 6 is installed inside the windshield of the driver's cab structure 29, a thermal imager 5 is installed outside the windshield, and a pitot tube 4 is installed on the lower side of the thermal imager 5, wherein the thermal imager 5 is used to detect invading organisms with thermal effects, thereby realizing real-time imaging; wherein the pitot tube 4 is used to measure the temperature outside the vehicle The temperature, pressure and speed are measured and used as early warning for extreme weather and as input for safe operation of trains; the zoom camera 6 is used to detect images less than 500 meters, thereby realizing detection and data transmission of intrusions 2; a telephoto camera 7 is installed above the thermal imager 5, a pressure sensor 3 is installed on the skin plate of the coupler opening and closing mechanism, a snow remover 17 is installed on the lower side of the opening and closing mechanism, and a laser radar 12 is installed on the outside of the opening and closing mechanism and the skirt 28, wherein the laser radar 12 is used to detect and scan the surrounding scenes of the vehicle on the track 1 and to construct a three-dimensional map and realize detection and early warning of different objects approaching the train; the telephoto camera 7 is used to realize imaging greater than 500 meters, thereby realizing detection and data transmission of intrusions 2.

[0038] Specifically, the bogie 27 includes a wheelset 18, a frame 21, an axle box cover 19, a first tie spring 20 and a second tie spring 22. A first acceleration sensor 14, a second acceleration sensor 15 and an angular momentum encoder 13 are arranged on the axle box cover 19, and a frame 21 is arranged between the two wheelsets 18;

[0039] It can be understood that in this step, a first acceleration sensor, a second acceleration sensor and an angular momentum encoder are arranged on the axle box cover to collect the vibration of the wheelset, especially for the monitoring and identification of the foreign objects crushing the track. The pressure sensor in the closed structure of the train driver's cab structure can detect the damage caused by foreign objects (stones, sticks, obstacles, animals, people, etc. that invade the track line) when they hit the vehicle end. Minor collisions do not affect the normal form of the vehicle. In the event of a more serious collision (the impact acceleration or loss of the pressure sensor can be used as the trigger alarm threshold), an impact signal is transmitted to slow down the train or emergency brake.

[0040] Among them, the first acceleration sensor 14, the second acceleration sensor 15 and the angular momentum encoder 13 are used to monitor and identify the foreign objects crushing the track 1, collect the vibration of the wheelset 18 in real time, and transmit the impact signal to slow down the train or perform emergency braking.

[0041] Specifically, a front camera 9 and a rear camera 10 are installed above the vehicle body 8;

[0042] Among them, the front camera 9 and the rear camera 10 are used for intrusion detection and imaging of falling objects.

[0043] Specifically, a storage exchange system is provided below the driver's cab structure 29, wherein the storage exchange system includes a data exchanger 23, a data storage 24, a port 25 and a computer 26;

[0044] Among them, the data exchanger 23 is used to forward the data acquisition system to the correct port by identifying the destination address in the data acquisition system, thereby realizing rapid data exchange; the data storage device 24 is used to store the equipment for collecting and saving data of the rail train in the unmanned state; the port 25 is used to distinguish different services or processes, in which each network connection needs to specify a source port and a target port; the computer 26 is used to perform computing tasks and process data.

[0045] Specifically, an opening and closing device 16 is provided below the data exchanger 23 and the data storage device 24 .

[0046] It is understandable that in this step, the images and data obtained by the vehicle-mounted sensors are stored, including the three-dimensional scanning imaging data of the laser radar, the threshold judgment data of the pressure sensor, the biological detection data of the thermal imager (shape size and position judgment), the detection and scene imaging data of the intrusion in front of the vehicle by the telephoto camera and the zoom camera, the extreme weather detection and warning data of the pitot tube, and the intrusion detection data of the high-altitude falling objects by the cameras at the front and rear of the vehicle. The accuracy and robustness of the system are ensured by the multi-sensor data fusion technology.

[0047] Embodiment 2:

[0048] This embodiment provides a method for anti-collision identification of a rail train in an unmanned driving state.

[0049] See also Figure 2 , the figure shows that the method includes step S100, step S200, step S300, step S400 and step S500.

[0050] S100, obtaining three-dimensional imaging data of the periphery of the train body according to the laser radar (12), and detecting and judging potential intrusions using other on-board sensors, thereby obtaining initial data of the train operation scene.

[0051] It can be understood that in this step, the train data factory is a systematic and automated data management and processing framework for efficiently training sensor-based data acquisition systems, and in active and semi-active autonomous driving, it can detect obstacles by identifying and classifying them, and realize the observation and analysis of railway environmental conditions. The train data factory consists of three core components: train-side data sensors, trackside data contacts, and high-performance data centers. The train-side data sensors are responsible for collecting real-time data, and the trackside data contacts serve as nodes for data transmission to ensure that the data is uploaded to the high-performance data center, which is responsible for storing, processing, and analyzing the data. Developers can access the data in the data center for system debugging, algorithm testing, and model training to improve the reliability of train data. The train data factory continuously collects new data to train and optimize the model through a data circulation mechanism to adapt to changes in the line environment.

[0052] S200: The initial data of the train operation scenario is accumulated as initial data, and combined with the collected historical detection data, the data is stored to form a complete railway network big data detection cluster.

[0053] It is understandable that in this step, in step S200, the initial data of the train operation scene is first obtained from the train sensor. These data include but are not limited to the three-dimensional imaging data of the vehicle body surrounding captured by the on-board laser radar, and the real-time detection information of potential intruders by other sensors. The collection of initial data is the basis for building a big data detection cluster, and provides raw materials for subsequent data analysis and machine learning model training. In order to form a comprehensive and in-depth railway network big data detection cluster, it is necessary to combine these initial data with historical detection data, where the historical detection data may come from previous train operation records, maintenance logs, accident reports and other relevant data archives. The accumulation of these data is crucial to understanding long-term trends and abnormal patterns in the railway environment.

[0054] Therefore, when storing this data, an efficient database management system is needed to ensure data accessibility, consistency and security. The design of the database needs to take into account query efficiency and the convenience of data update to support fast data retrieval and real-time data analysis needs. This cluster not only provides a large number of training samples for the machine learning model, but also increases the generalization ability of the model for intrusion detection under different operating conditions. The construction and management of the big data detection cluster may also involve data fusion technology, which integrates information from different sensors and data sources to obtain a more comprehensive and accurate understanding of the train operation scene. The data fusion algorithm can improve the robustness of the detection system and reduce the risks caused by failures or errors in a single data source.

[0055] S300. Train the railway network big data detection cluster through artificial intelligence machine learning algorithms to obtain a railway train intrusion judgment prediction model, where the model is used to classify intrusions, determine their locations, and predict the degree of danger to train operation.

[0056] Specifically, in this step, the contact downloads a large amount of sensor data such as lidar collected by the train during its travel through the wireless connection range of the train, and preprocesses and reduces the data so that only the relevant part of the data needs to be transmitted to the data center. The collected data is then labeled to provide the supervisory signal required to train the machine learning model. Features that help the model learn, such as texture, shape, and color in the image, are selected, and machine learning models such as deep neural networks are trained using the selected features and labeled data. High-performance computers are distributed in the data center, GPUs are used for model training and reasoning, and efficient optimization algorithms such as the Adam optimization algorithm are applied to speed up model training and improve recognition accuracy.

[0057] It can be understood that step S300 includes S301, S302 and S303, wherein:

[0058] S301, performing data cleaning and formatting processing on the data in the railway network big data detection cluster, removing noise and outliers, and obtaining the final railway network big data detection cluster;

[0059] S302, annotating the final railway network big data detection cluster to distinguish different types of intrusions and different degrees of harm of each intrusion to train operation, and obtaining annotated railway network big data detection cluster;

[0060] S303. The labeled railway network big data detection cluster is trained based on the random forest algorithm. During the training process, the labeled railway network big data detection cluster is divided into a test set and a validation set. The test set is used for training, and the remaining validation set is used for validation. This cycle is repeated multiple times to obtain a railway train intrusion judgment prediction model, and the railway train intrusion judgment prediction model is adjusted and optimized to obtain a final railway train intrusion judgment prediction model, wherein the final railway train intrusion judgment prediction model includes adjusting model parameters and selecting different feature subsets.

[0061] It should be noted that in step S301, the railway network big data detection cluster is first thoroughly cleaned. This process involves identifying and removing noise and outliers in the data set, which may be caused by sensor errors, data transmission errors, or environmental interference. The purpose of data cleaning is to ensure the quality and consistency of the data set and provide an accurate data basis for subsequent analysis and model training. Next, data formatting is performed, which means converting the data into a unified format and structure for algorithm processing. Formatting may include standardizing data types, unifying timestamp formats, adjusting data with inconsistent dimensions, etc.

[0062] Specifically, in step S302, the cleaned and formatted data is labeled. Labeling is a process of classification and classification level, which requires professional knowledge to identify different types of intrusive objects (such as animals, vegetation, thrown objects, etc.) and evaluate their potential degree of harm to train operation (such as low, medium, and high).

[0063] Specifically, in step S303, the labeled data set is trained using the random forest algorithm. Random forest is an integrated learning method that improves the accuracy and robustness of the model by constructing multiple decision trees and combining their prediction results. During the training process, the data set is divided into a test set and a validation set. The test set is used for the actual training of the model, while the validation set is used to evaluate the performance of the model to ensure that the model is not overfitted. Through cross-validation, that is, repeating this process multiple times, each time using a different data subset as a validation set, the performance of the model can be more comprehensively evaluated. After the training is completed, the model is adjusted and optimized. This may include adjusting model parameters (such as the number of trees, maximum depth, etc.), selecting different feature subsets, or trying different algorithm configurations. Therefore, in actual research, the training and optimization of the model may also involve more advanced technologies, such as feature selection algorithms, model fusion technologies, hyperparameter optimization algorithms (such as Bayesian optimization), etc. In addition, with the development of deep learning technology, it is possible to consider using convolutional neural networks (CNN) or recurrent neural networks (RNN) to process image data or time series data to further improve the performance of the model.

[0064] By training, learning and processing massive amounts of scanned data through artificial intelligence (AI) software in the field of machine learning, the system can quickly locate and identify the surrounding environment and objects, such as train tracks, telephone poles, fences, etc., from the established massive data. At the same time, it uses lidar data to create accurate three-dimensional imaging modeling of the tracks and surrounding environment, and establish a digital map of the railway environment.

[0065] S400: Based on the railway train intrusion judgment prediction model, real-time intrusions are detected and comprehensive judgment is made on the detection results.

[0066] It can be understood that step S400 includes S401 and S402, wherein:

[0067] S401, based on the railway train intrusion judgment prediction model, use the real-time collected data to match the historical training data, and use the judgment formula to identify whether an intrusion has appeared. The judgment formula is as follows:

[0068]

[0069] Where d is the standardized value of the data difference, Znew is the mean of the newly collected data, Zhistory is the mean of the historical data, and σ is the standard deviation of the historical data;

[0070] S402, if yes, locate the intruding object, and comprehensively judge the characteristics of the intruding object for comprehensive analysis; if no, continue to collect train operation scene data in real time and perform iterative calculations.

[0071] It can be understood that step S402 includes S4021, S4022, S4023 and S4024, wherein:

[0072] S4021. Acquire the intrusion object characteristics in the real-time data, wherein the intrusion object characteristics include size, shape and speed;

[0073] S4022. Using the positioning system, calculate based on the acquired intrusion features to determine the location of the intrusion. The calculation formula is as follows:

[0074] P intruder =f(Z new , Θ, R) where f is the positioning function, Θ is the sensor angle, Znew is the mean of the newly collected data, R is the distance from the sensor to the center of the train, P intruder is location information;

[0075] S4023, using image processing technology to calculate the final size, shape and speed of the intruding object, extracting key features from the calculated size, shape and speed of the intruding object, and converting the extracted key features into a numerical vector to form a feature vector;

[0076] S4024. Based on the pre-trained machine learning model, the feature vector is input into the machine learning model for classification and property judgment to obtain output results, and the output results are integrated to finally obtain a comprehensive analysis and description result of the intrusion.

[0077] By combining artificial intelligence software with powerful computing performance, the data collected by sensors such as on-board cameras and lidar are analyzed and compared with the digital map of the railway environment. Static and dynamic obstacles on and beside the track are detected in real time. The collision risk is dynamically calculated based on the size, speed and position of the obstacle, combined with the safety margin and operation strategy of the train. Based on the risk assessment results, it is decided whether action is needed, such as slowing down or stopping. Based on actual events and feedback, the risk assessment model is continuously optimized to improve the safety and efficiency of train operation. Among them, the fault judgment is that there must be no encroaching objects on the track. After the encroaching objects appear, it is determined whether they are snow accumulation or fine sand cover. If it is judged to be minor or harmless, if it is determined to be gravel, the size of the gravel needs to be determined. If it is less than a certain diameter, it is judged to be minor or harmless. If it is greater than a certain index, such as greater than 10mm and less than 100mm, it is judged to have a certain degree of harm, and the snow remover can be started to clean out the secondary gravel. If it is greater than 100mm and the snow remover is difficult to handle, it is judged to be a serious hazard and emergency braking is performed to avoid a collision accident; if it is a biological invasion of the track line, it is determined whether it is a human invasion or an animal invasion. If it is a human invasion, it is determined how far the person is from the track, and the train whistle is started and the train is decelerated and braked to drive away. If the person is judged to be on the track, if it is suicide, emergency braking is performed; if it is an animal crossing the line to invade the track line, the distance between the track and the train is also judged. If it is a national protected animal, it is determined to sound the whistle and decelerate and brake to drive away the animal. If it is a general and small animal, it is determined to sound the whistle and drive away, and the snow remover is turned on and slows down to reduce damage to the animal.

[0078] It should be noted that step S4023 includes S40231, S40232 and S40233, wherein:

[0079] S40231, extracting features of the intrusion from the three-dimensional point cloud of the three-dimensional imaging sensor of the laser radar (12) using image processing technology, and distinguishing the background of the intrusion from the intrusion itself according to a region growing or clustering algorithm; estimating the size of the intrusion by calculating the range of the entire space occupied by the intrusion in the point cloud, and re-measuring the size using edge detection and geometric analysis methods to obtain final size information;

[0080] S40232. Based on the extracted intrusion image or point cloud data, use Fourier descriptors to characterize its shape features to obtain final shape information;

[0081] S40233. Obtain the time series information contained in the sensor data, calculate the speed of the intruding object by comparing its position changes at different time points, and then obtain the final speed information.

[0082] It should be noted that the data collected in the preliminary test phase and the digital map of the railway environment are preloaded during the formal operation of the train, and the location information of the fixed equipment is recorded. Through GPS (or Beidou positioning system) and sensor data (acquired in real time), and the previously established digital map of the railway environment (a train scans once a month or a year to establish a digital map of the railway environment, which contains the three-dimensional geometric features of railway facilities such as buildings around the railway, poles and other railway facilities and accessories and the corresponding geographic coordinate system (excluding trains)), by identifying the static objects that do not change near the train (such as power towers), the identified object features are matched with the features of known objects in the digital map to determine the coordinate position of the object on the map. The calibrated location information is merged with the GPS positioning data to correct the GPS positioning data and achieve higher precision positioning. This method based on the combination of landmark positioning and GPS can achieve centimeter-level high-precision positioning. Each railway ancillary facility has its own unique three-dimensional geometric features, such as the arrangement of sleepers, the combination features beside sleepers, the geometric dimensions of power towers and transmission lines, etc., which are all three-dimensional data that can be extracted and corresponded to the corresponding geodetic coordinates one by one. First, a rough location interval is obtained, and the data of power towers and houses with high-significant features are identified again using artificial intelligence big data. Finally, the geometric features of sleepers and sand and gravel are identified to determine the precise geodetic location coordinates. At the same time, the presence of foreign objects and the distance of foreign objects from the train can be identified, and the degree of danger to the train can be judged based on the distance, and verified with the detection data of the on-board sensors.

[0083] Specifically, region growing or clustering algorithms are used to distinguish the background of the intrusion from the intrusion itself; region growing algorithms start from seed points and gradually merge adjacent points into regions until specific conditions are met; clustering algorithms, such as K-means or DBSCAN, help identify and distinguish different objects by grouping data points; and size estimation is achieved by calculating the range of the space occupied by the intrusion itself in the point cloud, which involves geometric analysis of the point cloud data, which may include the calculation of geometric shapes such as the minimum covering rectangle, the minimum covering circle, or the convex hull; edge detection algorithms, such as the Canny algorithm, can be used to identify the boundaries of the intrusion, and geometric analysis methods are used to extract size information from these boundaries. The technical effect of this step is to provide an automated method to identify and quantify the physical properties of intrusions from complex three-dimensional data.

[0084] In the next step, Fourier descriptor is a method that converts image or point cloud data into the frequency domain and extracts shape features from it. By performing Fourier transform on the boundary or contour of the shape, its frequency domain representation can be obtained, and then the key features describing the shape can be extracted, which helps the machine learning model understand the unique geometric properties of different intrusions, which is crucial for classifying and identifying the type of intrusions.

[0085] In step S40233, time series information is obtained from the sensor data and used to calculate the speed of the intruder, which involves comparing the position changes of the intruder at different time points. By measuring the position difference between two time points and dividing it by the time difference, the average speed of the intruder can be obtained. In actual research, the application of these steps may be combined with more advanced image and signal processing techniques, such as using deep learning for point cloud analysis, or using time series analysis methods to process sensor data. In addition, combined with multimodal data fusion technology, the accuracy of extracting and analyzing the characteristics of the intruder can be further improved.

[0086] S500: Output corresponding train operation instructions according to the comprehensive determination result to realize the prediction and response to the intrusion into the railway system.

[0087] It is understandable that in this step, the system connects the detection results and key information to the railway network and uploads them to the cloud, shares information such as common obstacles, improves recognition efficiency, and enables collaboration between different lines to better monitor the running status of trains, achieve intelligent scheduling and optimize operational efficiency. By connecting to the cloud, the data factory can use the powerful computing resources of the cloud to process large-scale data sets and support complex data analysis and machine learning model training. The elasticity and scalability of cloud services ensure that the system can operate stably when the amount of data surges or computing needs increase, avoiding service interruptions caused by peak loads.

[0088] Specifically, in this step, logical rules or machine learning models are used to evaluate the specific risk level of intruders to train operations. For example, large animals or objects moving at high speeds may require a higher level of response. According to the risk assessment results, corresponding train operation instructions are formulated, where possible instructions include slowing down, emergency braking, changing tracks, starting obstacle removal devices, or notifying maintenance personnel. In addition, taking into account the current status, speed, position of the train and the operation of the railway network, the output instructions are optimized to ensure safe and efficient train operation. The finalized operation instructions are communicated to the train driver or the automatic driving system through the train control system to ensure that the train can respond in time and take appropriate measures.

[0089] In actual research and application, the output of train operation instructions can be further integrated with the train's automatic driving system to achieve fully automated train operation control. In addition, these instructions can also be combined with the railway traffic management system to achieve optimized scheduling of the entire railway network. With the development of the Internet of Things and vehicle networking technologies, the output and reception of train operation instructions will be more real-time and accurate, and the communication between trains and trackside sensors, traffic control systems, and other trains will make train operations more intelligent and collaborative.

[0090] It should be noted that, regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment of the method, and will not be elaborated here.

[0091] Embodiment 3:

[0092] Corresponding to the above method embodiment, this embodiment also provides a rail train anti-collision identification device in an unmanned driving state. The rail train anti-collision identification device in an unmanned driving state described below and the rail train anti-collision identification method in an unmanned driving state described above can be referenced to each other.

[0093] Figure 3 FIG. 8 is a block diagram of a rail train anti-collision identification device 800 in an unmanned driving state according to an exemplary embodiment. Figure 3 As shown, the rail train anti-collision identification device 800 in the unmanned driving state includes: a processor 801 and a memory 802. The rail train anti-collision identification device 800 in the unmanned driving state also includes one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0094] The processor 801 is used to control the overall operation of the rail train anti-collision identification device 800 in the unmanned driving state to complete all or part of the steps in the above-mentioned rail train anti-collision identification method in the unmanned driving state. The memory 802 is used to store various types of data to support the operation of the rail train anti-collision identification device 800 in the unmanned driving state. These data may include, for example, instructions for any application or method used to operate on the rail train anti-collision identification device 800 in the unmanned driving state, and application-related data, such as contact data, sent and received messages, pictures, audio, video, etc. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, disk or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 802 or sent via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, and the above-mentioned other interface modules can be a keyboard, a mouse or buttons, etc. These buttons can be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the rail train anti-collision identification device 800 in the unmanned driving state and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 805 can include: a Wi-Fi module, a Bluetooth module or an NFC module.

[0095] In an exemplary embodiment, the rail train anti-collision identification device 800 in an unmanned driving state can be implemented by one or more application specific integrated circuits (Application Specific Integrated Circuit, referred to as ASIC), digital signal processors (Digital Signal Processor, referred to as DSP), digital signal processing devices (Digital Signal Processing Device, referred to as DSPD), programmable logic devices (Programmable Logic Device, referred to as PLD), field programmable gate arrays (Field Programmable Gate Array, referred to as FPGA), controllers, microcontrollers, microprocessors or other electronic components, and is used to execute the above-mentioned rail train anti-collision identification method in an unmanned driving state.

[0096] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the steps of the above-mentioned rail train anti-collision identification method in an unmanned driving state are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 802 including program instructions, and the above-mentioned program instructions can be executed by the processor 801 of the rail train anti-collision identification device 800 in an unmanned driving state to complete the above-mentioned rail train anti-collision identification method in an unmanned driving state.

[0097] Embodiment 4:

[0098] Corresponding to the above method embodiment, a readable storage medium is also provided in this embodiment. The readable storage medium described below and the rail train anti-collision identification method in an unmanned driving state described above can be referred to each other.

[0099] A computer program is stored on the readable storage medium, and when the computer program is executed by the processor, the steps of the rail train anti-collision identification method in an unmanned driving state of the above method embodiment are implemented.

[0100] The readable storage medium may specifically be a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or other readable storage medium that can store program codes.

[0101] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0102] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A rail train anti-collision identification device in an unmanned driving state, characterized in that: include: A data acquisition system, an intelligent detection system and a positioning system, wherein the data acquisition system comprises a vehicle body (8), a driver's cab structure (29) is installed on one side of the vehicle body (8), a positioning system is arranged on the side beam of the vehicle body (8), bogies (27) are installed on the left and right ends of the lower side of the vehicle body (8), wherein the positioning system comprises a GPS positioning system (11) and a three-dimensional digital map imaging technology, and the intelligent detection system comprises an artificial intelligence detection algorithm of machine learning; a zoom camera (6) is installed inside the windshield of the driver's cab structure (29), a thermal imager (5) is installed outside the windshield, and a pitot tube (4) is installed on the lower side of the thermal imager (5), wherein the thermal imager (5) is used to detect invading organisms with thermal effects, thereby realizing real-time imaging; wherein the pitot tube (4) is used to measure the vehicle The thermal imaging camera (5) is used to detect the external temperature, pressure and speed and to serve as an early warning for extreme weather and as an input for safe operation of the train; wherein the zoom camera (6) is used to detect images less than 500 meters, thereby realizing the detection and data transmission of intrusions (2); a telephoto camera (7) is installed above the thermal imaging camera (5), a pressure sensor (3) is installed on the skin plate of the coupler opening and closing mechanism, a deflector (17) is installed on the lower side of the opening and closing mechanism, and a laser radar (12) is installed on the outside of the opening and closing mechanism and the skirt (28), wherein the laser radar (12) is used to detect and scan the surrounding scenes of the vehicle on the track (1) and to construct a three-dimensional map and realize the detection and early warning of different objects approaching the train; wherein the telephoto camera (7) is used to realize imaging greater than 500 meters, thereby realizing the detection and data transmission of intrusions (2).

2. The anti-collision identification device for a railway train in an unmanned driving state according to claim 1 is characterized in that: The bogie (27) comprises a wheelset (18), a frame (21), an axle box cover (19), a first tie spring (20) and a second tie spring (22); a first acceleration sensor (14), a second acceleration sensor (15) and an angular momentum encoder (13) are arranged on the axle box cover (19); and a frame (21) is arranged between the two wheelsets (18); The first acceleration sensor (14), the second acceleration sensor (15) and the angular momentum encoder (13) are used to monitor and identify the foreign matter crushing the track (1), collect the vibration of the wheelset (18) in real time, and transmit the impact signal to slow down the train or perform emergency braking.

3. The anti-collision identification device for a railway train in an unmanned driving state according to claim 1 is characterized in that: A front camera (9) and a rear camera (10) are installed above the vehicle body (8); The camera (9) on the front of the vehicle and the camera (10) on the rear of the vehicle are used for intrusion detection and imaging of falling objects.

4. The anti-collision identification device for a railway train in an unmanned driving state according to claim 1, characterized in that: A storage exchange system is arranged below the driver's cab structure (29), wherein the storage exchange system comprises a data exchanger (23), a data storage device (24), a port (25) and a computer (26); The data exchanger (23) is used to forward the data acquisition system to the correct port by identifying the destination address in the data acquisition system, thereby realizing rapid data exchange; the data storage device (24) is used to store the equipment for collecting and saving data of the unmanned rail train; the port (25) is used to distinguish different services or processes, wherein each network connection needs to specify a source port and a destination port; the computer (26) is used to perform computing tasks and process data.

5. The anti-collision identification device for a railway train in an unmanned driving state according to claim 4 is characterized in that: An opening and closing device (16) is provided below the data exchanger (23) and the data storage device (24).

6. A method for anti-collision identification of a rail train in an unmanned driving state, characterized in that: The anti-collision identification device for a rail train in an unmanned driving state according to any one of claims 1 to 5 comprises the following steps: Acquiring three-dimensional imaging data of the periphery of the train body using a laser radar (12), and detecting and determining potential intrusions using other onboard sensors, thereby obtaining initial data of the train operation scene; The initial data of the train operation scenario is accumulated as initial data, and combined with the collected historical detection data, it is stored to form a complete railway network big data detection cluster; The railway network big data detection cluster is trained through artificial intelligence machine learning algorithms to obtain a railway train intrusion judgment prediction model, where the model is used to classify intrusions, determine their locations, and predict the degree of danger to train operation; Based on the railway train intrusion judgment and prediction model, real-time intrusion detection is carried out and the detection results are comprehensively judged; Based on the comprehensive judgment results, the corresponding train operation instructions are output to achieve the prediction and response to intrusions into the railway system.

7. The method for anti-collision identification of a rail train in an unmanned driving state according to claim 6, characterized in that: The railway train intrusion judgment prediction model is used to detect the real-time intrusion and make a comprehensive judgment on the detection results, including: Based on the railway train intrusion prediction model, the real-time collected data is matched with the historical training data, and the judgment formula is used to identify whether an intrusion has appeared. The judgment formula is as follows: Where d is the standardized value of the data difference, Znew is the mean of the newly collected data, Zhistory is the mean of the historical data, and σ is the standard deviation of the historical data; If there is, the intrusion will be located, and the characteristics of the intrusion will be comprehensively judged and analyzed; if not, the real-time collection of train operation scene data will continue and iterative calculations will be performed.

8. The method for anti-collision identification of a rail train in an unmanned driving state according to claim 7, characterized in that: The method of locating the intrusions and comprehensively analyzing the characteristics of the intrusions includes: Obtaining intrusion characteristics in real-time data, wherein the intrusion characteristics include size, shape and speed; Using the positioning system, the position of the intruder is determined based on the acquired features of the intruder. The calculation formula is as follows: P intruder =f(Z new ,Θ,R) Where f is the positioning function, Θ is the sensor angle, and Z new is the mean of the newly collected data, R is the distance from the sensor to the center of the train, P intruder is location information; The final size, shape and speed of the intrusion object are calculated by using image processing technology, and key features are extracted from the calculated size, shape and speed of the intrusion object, and the extracted key features are converted into numerical vectors to form feature vectors; Based on the pre-trained machine learning model, the feature vector is input into the machine learning model for classification and property judgment to obtain the output results, which are then integrated to finally obtain a comprehensive analysis and description of the intrusion.

9. The method for anti-collision identification of a rail train in an unmanned driving state according to claim 8, characterized in that: The image processing technology is used to calculate the final size, shape and speed of the intrusion, including: The features of the intrusion object are extracted from the three-dimensional point cloud of the three-dimensional imaging sensor of the laser radar (12) by using image processing technology, and the background of the intrusion object and the intrusion object body are distinguished according to a region growing or clustering algorithm; the size of the intrusion object is estimated by calculating the range of the entire space occupied by the intrusion object body in the point cloud, and the size is measured again by edge detection and geometric analysis methods to obtain final size information; According to the extracted intrusion image or point cloud data, the Fourier descriptor is used to characterize its shape features to obtain the final shape information; The time series information contained in the sensor data is obtained, and the speed of the intruding object is calculated by comparing its position changes at different time points, thereby obtaining the final speed information.

10. The method for anti-collision identification of a rail train in an unmanned driving state according to claim 6, characterized in that: The railway network big data detection cluster is trained by artificial intelligence machine learning algorithm to obtain a railway train intrusion judgment prediction model, wherein the model is used to realize the classification and location judgment of intrusions and the prediction of the degree of danger to train operation, including: Clean and format the data in the railway network big data detection cluster to remove noise and outliers and obtain the final railway network big data detection cluster; The final railway network big data detection cluster is labeled to distinguish different types of intrusions and the different degrees of harm of each intrusion to train operation, and the labeled railway network big data detection cluster is obtained; The labeled railway network big data detection cluster is trained based on the random forest algorithm. During the training process, the labeled railway network big data detection cluster is divided into a test set and a validation set. The test set is used for training, and the remaining validation set is used for validation. This cycle is repeated multiple times to obtain a railway train intrusion judgment prediction model, and the railway train intrusion judgment prediction model is adjusted and optimized to obtain a final railway train intrusion judgment prediction model, wherein the final railway train intrusion judgment prediction model includes adjusting model parameters and selecting different feature subsets.

Citation Information

Patent Citations

  • Positioning method for roadside autonomous vehicle

    CN110220529A

  • System for intelligently monitoring behaviors of rail train driver and intelligently detecting obstacle invasion

    CN113562005A

  • Intrusion detection system and method for track construction interval and early warning system

    CN114802354A

  • Long-distance personnel intrusion detection method and device under illumination-free condition

    CN114973564A

  • Vehicle obstacle avoidance method and system, vehicle and storage medium

    CN117087664A

Cited By

  • Passive obstacle detection method and system for train

    CN119636855A

  • Track turnout foreign matter detection and type analysis method and system

    CN120411656A

  • A method and system for detecting and analyzing foreign objects in track switches

    CN120411656B