Railway freight train adaptive asynchronous control system based on obstacle identification

Abstract

The present application belongs to the field of freight train control, and relates to obstacle identification technology, which is used to solve the problem that the single obstacle monitoring and identification means in the prior art cannot effectively deal with the obstacle identification avoidance of the whole process of freight train transportation, and specifically is a railway freight train adaptive asynchronous control system based on obstacle identification, which comprises obstacle monitoring modules, obstacle identification modules, response analysis modules and verification optimization modules connected in sequence, and the obstacle identification modules, the response analysis modules and the verification optimization modules are all connected with a database; the present application solves the problem of insufficient reliability of the single monitoring means in the prior art, improves the accuracy and anti-interference ability of obstacle detection through high-precision scanning of laser radar and quantitative analysis of obstacle value, provides a more reliable data basis for subsequent braking decision, improves the accuracy of obstacle identification through cross-validation mechanism, and automatically starts system self-checking and parameter optimization when the detection results conflict.

Description

Technical Field

[0001] This invention belongs to the field of freight train control and relates to obstacle recognition technology, specifically an adaptive asynchronous control system for railway freight trains based on obstacle recognition. Background Technology

[0002] The adaptive asynchronous control system for railway freight trains is an advanced train control system based on multi-sensor fusion and intelligent decision-making. It aims to improve the safety, operating efficiency, and energy utilization of freight trains. The system achieves coordinated and optimized control of multiple locomotives and multiple carriages by sensing environmental obstacles, train status, and track conditions in real time and adopting a distributed asynchronous control strategy.

[0003] The invention patent with publication number CN110027592B discloses a CBTC unmanned vehicle control system with intelligent obstacle detection and early warning functions. This control system can comprehensively realize unmanned operation control of the train based on route information, operation plan, and obstacles within the track and train operation limits during train operation. However, the system only uses a single method for obstacle detection, and the monitoring results cannot be verified. The sensors deployed on the rails and the cameras deployed on the train are affected by high temperature and vibration, which may cause them to malfunction. Therefore, a single obstacle detection and identification method cannot effectively cope with obstacle identification and avoidance throughout the entire freight train transportation process, resulting in low safety. In addition, the existing technology cannot combine obstacle identification verification results for optimization analysis, resulting in the inability to effectively improve the safety of the freight network.

[0004] To address the aforementioned technical problems, this application proposes a solution. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive asynchronous control system for railway freight trains based on obstacle recognition, which solves the problem that existing technologies using a single obstacle monitoring and recognition method cannot effectively address obstacle recognition and avoidance throughout the entire freight train transportation process.

[0006] The technical problem to be solved by this invention is: how to provide an adaptive asynchronous control system for railway freight trains based on obstacle recognition that can verify and analyze the results of obstacle detection and recognition.

[0007] The objective of this invention can be achieved through the following technical solutions:

[0008] An adaptive asynchronous control system for railway freight trains based on obstacle recognition includes an obstacle monitoring module, an obstacle recognition module, a response analysis module, and a verification and optimization module that are connected in sequence via communication. The obstacle recognition module, response analysis module, and verification and optimization module are all connected in communication with a database.

[0009] The obstacle monitoring module is used to monitor and analyze obstacles on the railway track: a lidar is deployed every K1 meters on the railway track, and the lidar monitors whether there are obstacles in the measurement area and marks the measurement area as a safe area or a risk area;

[0010] The obstacle recognition module is used to identify and analyze obstacles on the railway track: a panoramic camera is installed on the front of the railway freight train to identify obstacles.

[0011] The response analysis module is used to perform braking response analysis on railway freight trains: it compares the monitoring results and identification results of obstacles, generates braking control signals based on the comparison results, and sends the braking control signals to the client in the train driver's cab;

[0012] The verification and optimization module is used to perform verification and optimization analysis on the obstacle monitoring and identification process of railway freight trains.

[0013] Furthermore, the specific process of monitoring whether there are obstacles in the measurement area using lidar includes: generating and emitting a laser with a wavelength of 905nm, a pulse width of 3-10ns, and a repetition frequency of 50-200kHz using lidar. Each laser pulse corresponds to a measurement point, and the density of measurement points is 200-400 points / m² for each measurement area. The difference between the theoretical time difference and the actual time difference between the measurement point and the lidar is marked as the obstacle value. The measurement area is marked as a safe area or a risk area based on the obstacle value.

[0014] Furthermore, the specific process of marking the measurement area as a safe area or a risk area includes: comparing the obstacle value with a preset obstacle threshold; if the obstacle value is less than the obstacle threshold, the corresponding measurement point is marked as a free-flowing point; if the obstacle value is greater than or equal to the obstacle threshold, the corresponding measurement point is marked as an obstacle point; the ratio of the number of obstacle points to the number of measurement points in the measurement area is marked as the obstacle coefficient, and the obstacle coefficient is compared with a preset obstacle threshold; if the obstacle coefficient is less than the obstacle threshold, it is determined that there are no obstacles in the measurement area, and the corresponding measurement area is marked as a safe area; if the obstacle coefficient is greater than or equal to the obstacle threshold, it is determined that there are obstacles in the measurement area, and the corresponding measurement area is marked as a risk area.

[0015] Furthermore, the specific process of obstacle identification using a panoramic camera includes: capturing images of the train's direction of travel using the panoramic camera to obtain a traveling image; enlarging the traveling image into a pixel grid image and performing grayscale transformation; segmenting the traveling image according to the measurement area to obtain several recognition areas; performing contour segmentation on the recognition areas using a binary method to obtain several reference areas; acquiring the recognition features of the reference areas, including the shape, maximum height, area value, and average grayscale value of the reference areas; and marking the corresponding recognition areas as risk areas or safe areas based on the recognition features.

[0016] Furthermore, the specific process of marking the identification area as a risk area or a safe area includes: comparing the identification features of the control area with the comparison features of all obstacle types in the database; if the identification features of the control area correspond to the comparison features of any obstacle type, then the identification area to which the corresponding control area belongs is marked as a risk area; otherwise, the identification area to which the corresponding control area belongs is marked as a safe area.

[0017] Furthermore, the specific process of comparing the monitoring results and identification results of obstacles includes: if both the measurement area and its corresponding identification area are marked as safe areas, the passage risk of the measurement area is determined to meet the requirements; if both the measurement area and its corresponding identification area are marked as risk areas, the passage risk of the measurement area is determined to not meet the requirements, a braking control signal is generated and sent to the client in the train driver's cab; otherwise, the monitoring results and identification results are determined to be inconsistent, a braking control signal is generated and sent to the client in the train driver's cab, and a verification optimization signal is generated and sent to the verification optimization module.

[0018] Furthermore, the specific process of the verification and optimization module for verifying and optimizing the obstacle monitoring and identification process of railway freight trains includes: generating an optimization period, obtaining the number of times the monitoring results and identification results of the same railway track are inconsistent within the optimization period and marking them as monitoring optimization values, comparing the monitoring optimization values ​​with preset monitoring optimization thresholds: if the monitoring optimization value is less than the monitoring optimization threshold, it is determined that the railway track does not have monitoring optimization characteristics; if the monitoring optimization value is greater than or equal to the monitoring optimization threshold, it is determined that the railway track has monitoring optimization characteristics, and the lidar of the railway track is maintained or updated.

[0019] Furthermore, the specific process of the verification and optimization module for verifying and optimizing the obstacle monitoring and identification process of railway freight trains also includes: obtaining the number of times the monitoring results and identification results of the same railway freight train are inconsistent within the optimization period and marking them as identification optimization values; comparing the identification optimization values ​​with the preset identification optimization threshold; if the identification optimization value is less than the identification optimization threshold, it is determined that the railway freight train does not have identification optimization features; if the identification optimization value is greater than or equal to the identification optimization threshold, it is determined that the railway freight train has identification optimization features, and the lidar of the railway freight train is maintained or updated.

[0020] The present invention has the following beneficial effects:

[0021] 1. This application solves the problem of insufficient reliability of single monitoring methods in the prior art. By using high-precision scanning and obstacle value quantitative analysis of lidar, it improves the accuracy and anti-interference ability of obstacle detection, and provides a more reliable data basis for subsequent braking decisions.

[0022] 2. This application improves the accuracy of obstacle recognition through a cross-validation mechanism, and automatically initiates system self-checks and parameter optimization when detection results conflict, ensuring the safety of railway freight trains and the reliability of the system in complex operating environments;

[0023] 3. This application can solve the problem of insufficient reliability of monitoring equipment in the prior art. Through periodic verification and automatic maintenance mechanisms, it can reduce the probability of missed or false detection of obstacles caused by the performance degradation of lidar, and improve the overall safety level of railway freight system. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a system block diagram of Embodiment 1 of the present invention;

[0026] Figure 2 This is a flowchart of the method in Embodiment 2 of the present invention. Detailed Implementation

[0027] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] In existing technologies, obstacle detection for railway freight trains mainly relies on a single sensor or camera, such as a LiDAR system deployed along the track or a camera mounted on the train. However, the track environment is complex and variable. LiDAR may experience measurement errors due to high temperatures, and camera image quality may be affected by train vibrations. Single detection methods are prone to misjudgments or missed detections. Although the patent with publication number CN110027592B achieves obstacle detection and train control, it lacks a multi-source data verification mechanism. When sensors or cameras malfunction, it cannot promptly correct erroneous judgments, resulting in insufficient reliability of braking signal generation. Furthermore, existing systems lack continuous optimization analysis of the monitoring equipment's operating status, making it difficult to adapt to performance degradation issues during long-term operation.

[0029] To address the aforementioned issues, the inventors discovered that obstacle detection errors primarily stem from environmental interference and equipment performance degradation. Relying solely on a single detection method results in low system fault tolerance. Analysis revealed that lidar and cameras are complementary in obstacle detection: lidar excels in precise distance measurement, while cameras can identify object shape features. Cross-validating the detection results of both significantly reduces the probability of misjudgment. Furthermore, equipment performance degradation often manifests as periodic anomalies; statistically analyzing the frequency of these anomalies can trigger maintenance warnings. Based on this, the inventors proposed constructing a multi-module collaborative asynchronous control system, achieving dual protection through data comparison and periodic verification.

[0030] Example 1: As Figure 1 As shown, the adaptive asynchronous control system for railway freight trains based on obstacle recognition includes an obstacle monitoring module, an obstacle recognition module, a response analysis module, and a verification and optimization module that are connected in sequence. The obstacle recognition module, response analysis module, and verification and optimization module are all connected to a database.

[0031] The obstacle monitoring module is used to monitor and analyze obstacles on railway tracks. A lidar is deployed every K1 meters along the track. The lidar generates and emits laser light with a wavelength of 905nm, a pulse width of 3-10ns, and a repetition frequency of 50-200kHz. K1 is a numerical constant, set by administrators, typically 500. Each laser pulse corresponds to a measurement point, with a density of 200-400 points / m² per measurement area. The difference between the theoretical and actual time difference between the measurement point and the lidar signal is marked as the obstacle value. The measurement point is compared with a preset obstacle threshold: if the obstacle value is less than the obstacle threshold, the corresponding measurement point is marked as a clear point; if the obstacle value is greater than or equal to the obstacle threshold, the corresponding measurement point is marked as an obstacle point. The ratio of the number of obstacle points to the number of measurement points in the measurement area is marked as the obstacle coefficient. The obstacle coefficient is compared with a preset obstacle threshold: if the obstacle coefficient is less than the obstacle threshold, it is determined that there is no obstacle in the measurement area, and the corresponding measurement area is marked as a safe area; if the obstacle coefficient is greater than or equal to the obstacle threshold, it is determined that there is an obstacle in the measurement area, and the corresponding measurement area is marked as a risk area.

[0032] The laser wavelength of 905nm refers to the wavelength of the light emitted by the laser emitter, which can be achieved using a semiconductor laser. This wavelength has a low attenuation rate when penetrating atmospheric particles, making it suitable for long-distance monitoring in railway track environments. The pulse width of 3-10ns refers to the duration of a single laser pulse, which can be adjusted by regulating the laser drive circuit. A shorter pulse width helps improve distance resolution. The repetition frequency of 50-200kHz refers to the emission rate of the laser pulse, which can be controlled using a frequency divider circuit. A higher repetition frequency increases the amount of data collected per unit time. The measurement point density of 200-400 points / m² refers to the number of sampling points scanned by the lidar per unit area, which can be adjusted by regulating the lidar's scanning angle and rotation speed. A high density of measurement points enhances obstacle edge detection capabilities. The obstacle value refers to the deviation between the laser reflection time difference and the theoretical value, which can be measured using a time-to-digital converter. This parameter quantifies the degree of obstruction along the laser path.

[0033] Specifically, lidar is deployed at fixed intervals along the railway track, emitting laser beams with specific parameters to scan the measurement area. Each laser pulse reflects back to the radar upon encountering an obstacle. The obstacle value is obtained by calculating the difference between the theoretical propagation time and the actual reception time. The high-density distribution of measurement points allows for more precise capture of obstacle contours, reducing the possibility of missed detections. When the obstacle value exceeds a preset threshold, the corresponding area is identified as a risk area, triggering subsequent control procedures.

[0034] The obstacle recognition module is used to identify and analyze obstacles on railway tracks. A panoramic camera is mounted on the front of the freight train to capture images of the train's direction of travel. These images are then magnified into pixel-level images and subjected to grayscale transformation. The images are segmented into several recognition regions based on the measurement area. A binary method is used to segment the recognition regions into contours, resulting in several reference regions. The recognition features of the reference regions are obtained, including their shape, maximum height, area, and average grayscale value. These features are compared with the comparison features of all obstacle types in the database. If the recognition features of a reference region correspond to the comparison features of any obstacle type, the corresponding recognition region is marked as a risk area; otherwise, it is marked as a safe area.

[0035] In this context, a pixelated image refers to a gridded image composed of independent pixel units, formed by magnifying the original image. This can be achieved using image interpolation algorithms, increasing image resolution to facilitate subsequent detailed analysis. Grayscale transformation converts a color image to a grayscale image, which can be achieved using weighted averaging or component methods, enhancing contour contrast by eliminating color interference. The recognition region refers to a local image patch formed by segmenting the image based on the boundaries of the measurement area, achieved using grid partitioning or edge detection algorithms, narrowing the analysis scope through region segmentation. Binarization converts a grayscale image to a black-and-white binary image, achieved using fixed or adaptive thresholding algorithms, highlighting object contours by simplifying image information. The contrast region refers to an independent closed region formed after binary segmentation, achieved using edge tracking or region growing algorithms, separating potential obstacles through contour extraction. Recognition features are quantitative indicators used to describe the shape and attributes of an object, achieved using geometric feature extraction or texture analysis algorithms, improving recognition accuracy through multi-dimensional feature combination.

[0036] Specifically, after the panoramic camera captures an image of the train's path, the image is first magnified to enhance detail and then converted to grayscale to reduce data processing complexity. The image is then segmented into multiple recognition regions corresponding to the measurement area. A binary method is used to perform contour segmentation on each recognition region, forming multiple independent control regions. Features such as shape, height, area, and mean grayscale value of each control region are extracted and matched against obstacle features in the database. If a feature match is successful, the recognition region is marked as a risk region; otherwise, it is marked as a safe region.

[0037] The identification features refer to obstacle attribute parameters extracted through image processing technology. Specifically, contour segmentation algorithms can be used to obtain shape features, area values ​​and average gray values ​​can be calculated using pixel grid grayscale, and maximum height can be measured using 3D modeling technology. These features are then used for pattern matching with standard obstacle data in a database.

[0038] The comparison features refer to obstacle type feature data pre-stored in the database. Specifically, a feature library can be established by collecting the shape, height, area, and grayscale parameters of typical obstacles. The database can adopt a distributed storage architecture to achieve multi-node data synchronization, ensuring the real-time performance and accuracy of the comparison process.

[0039] Specifically, the panoramic camera captures a traveling image that is segmented into multiple recognition regions. Each region is then binarized to generate a control region. A four-dimensional feature vector is constructed by extracting the shape, maximum height, area, and average grayscale value of the control region. This vector is then compared with obstacle feature vectors stored in the database for similarity calculation. If the feature vector matches a feature vector of any obstacle type in the database with a similarity exceeding a preset threshold, the recognition region is determined to pose an obstacle risk. For example, if a control region is detected to have a cylindrical outline, a height exceeding 0.5 meters, and a grayscale value below a set range, it can be determined to match the "fallen log" obstacle feature in the database.

[0040] The response analysis module is used to perform braking response analysis on railway freight trains: it compares the monitoring results and identification results of obstacles; if both the measured area and its corresponding identification area are marked as safe areas, the traffic risk of the measured area is determined to meet the requirements; if both the measured area and its corresponding identification area are marked as risk areas, the traffic risk of the measured area is determined to not meet the requirements, a braking control signal is generated and sent to the client in the train driver's cab; otherwise, the monitoring results and identification results are determined to be inconsistent, a braking control signal is generated and sent to the client in the train driver's cab, and a verification optimization signal is generated and sent to the verification optimization module.

[0041] The comparison between monitoring and identification results refers to cross-verifying the measurement area status monitored by LiDAR with the identification area status identified by the panoramic camera. This can be achieved using data fusion algorithms, such as using a logical judgment module to perform consistency analysis on the marking results from the two sources. A safe zone refers to an area where no obstacles are detected or where the impact of obstacles does not exceed a preset threshold. This can be achieved through LiDAR obstacle coefficient calculation and image recognition feature matching. For example, a safe zone is defined as one where the obstacle coefficient is below the threshold and the identification feature does not match an obstacle type. A risk zone refers to an area where obstacles exist or where the impact of obstacles exceeds a preset threshold. This can be achieved through LiDAR obstacle point density and image recognition contour feature identification. For example, a risk zone is defined as one where the obstacle coefficient exceeds the threshold or the identification feature matches obstacle features in the database. A braking control signal is a command that triggers train braking operations. This can be achieved through the communication interface of the train control system, such as transmitting a signal to the braking execution unit to adjust the vehicle speed or stop the vehicle. A verification and optimization signal is a command that triggers system self-checks or parameter adjustments. This can be achieved through optimization algorithms to evaluate the working status of the monitoring and identification modules, such as initiating sensor calibration or camera focus adjustment when data is inconsistent.

[0042] The verification and optimization module is used to perform verification and optimization analysis on the obstacle monitoring and identification process of railway freight trains: It generates an optimization cycle, obtains the number of times the monitoring results and identification results of the same railway track are inconsistent within the optimization cycle and marks them as monitoring optimization values, and compares these values ​​with a preset monitoring optimization threshold: if the monitoring optimization value is less than the monitoring optimization threshold, the railway track is determined not to have monitoring optimization characteristics; if the monitoring optimization value is greater than or equal to the monitoring optimization threshold, the railway track is determined to have monitoring optimization characteristics, and the LiDAR of the railway track is maintained or updated. It also obtains the number of times the monitoring results and identification results of the same railway freight train are inconsistent within the optimization cycle and marks them as identification optimization values, comparing these values ​​with a preset identification optimization threshold: if the identification optimization value is less than the identification optimization threshold, the railway freight train is determined not to have identification optimization characteristics; if the identification optimization value is greater than or equal to the identification optimization threshold, the railway freight train is determined to have identification optimization characteristics, and the LiDAR of the railway freight train is maintained or updated.

[0043] The optimization cycle refers to a pre-defined time interval, which can be implemented using a fixed cycle or a dynamically adjusted cycle, and is used to periodically evaluate the working status of the monitoring equipment. The monitoring optimization value refers to the cumulative number of times the monitoring results and identification results for the same railway track are inconsistent within the optimization cycle, which can be implemented through a data statistics module and is used to quantify the reliability deviation of the monitoring equipment. The monitoring optimization threshold is a preset numerical benchmark, which can be set based on historical data or empirical values, and is used to determine whether to trigger maintenance operations. Maintenance or update refers to calibrating, repairing, or replacing the lidar, which can be implemented through manual operation or automated equipment, and is used to restore the accuracy of the monitoring equipment.

[0044] Among them, the identification optimization value refers to the cumulative number of times that the LiDAR and the panoramic camera on the same train have inconsistent obstacle judgment results. Specifically, it can be realized by the data acquisition unit to count the number of difference events within the period. This parameter is used to quantify the reliability of the collaborative work of the on-board sensors. The identification optimization threshold refers to the critical judgment value that triggers equipment maintenance. Specifically, it can be determined by fitting historical data or experimental testing. For example, it can be set as the maximum number of differences allowed within a single period. Its function is to distinguish the boundary conditions between normal equipment fluctuations and performance degradation.

[0045] Example 2: Figure 2 As shown, the adaptive asynchronous control method for railway freight trains based on obstacle recognition includes the following steps:

[0046] Step 1: Monitor and analyze obstacles on the railway track: Deploy lidar every K1 meters on the railway track, monitor the measurement area for obstacles using lidar, and mark the measurement area as a safe area or a risk area;

[0047] Step 2: Identify and analyze obstacles on the railway track: Install a panoramic camera on the front of the railway freight train to identify obstacles;

[0048] Step 3: Perform braking response analysis on railway freight trains: compare the monitoring and identification results of obstacles, generate braking control signals based on the comparison results, and send the braking control signals to the client in the train driver's cab;

[0049] Step 4: Verify and optimize the obstacle monitoring and identification process for railway freight trains.

[0050] The obstacle recognition-based adaptive asynchronous control system for railway freight trains operates by deploying lidar every kilometer along the railway track. The lidar monitors the measurement area for obstacles and marks the area as either safe or risky. A panoramic camera is mounted on the front of the freight train to identify obstacles. The monitoring and identification results are compared, and a braking control signal is generated and sent to the client in the train driver's cab based on the comparison. The obstacle monitoring and identification process for the railway freight train is then validated, optimized, and analyzed.

[0051] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

[0052] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0053] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. Railway freight train adaptive asynchronous control system based on obstacle recognition, characterized in that, It includes an obstacle monitoring module, an obstacle identification module, a response analysis module, and a verification optimization module that are connected in sequence. The obstacle identification module, response analysis module, and verification optimization module are all connected in communication with the database. The obstacle monitoring module is used to monitor and analyze obstacles on the railway track: a lidar is deployed every K1 meters on the railway track, and the lidar monitors whether there are obstacles in the measurement area and marks the measurement area as a safe area or a risk area; The obstacle recognition module is used to identify and analyze obstacles on the railway track: a panoramic camera is installed on the front of the railway freight train to identify obstacles. The response analysis module is used to perform braking response analysis on railway freight trains: it compares the monitoring results and identification results of obstacles, generates braking control signals based on the comparison results, and sends the braking control signals to the client in the train driver's cab; The verification and optimization module is used to perform verification and optimization analysis on the obstacle monitoring and identification process of railway freight trains; The specific process of monitoring whether there are obstacles in the measurement area using lidar includes: generating and emitting a laser with a wavelength of 905nm, a pulse width of 3-10ns, and a repetition frequency of 50-200kHz using lidar. Each laser pulse corresponds to a measurement point, and the density of measurement points is 200-400 points / m² for each measurement area. The difference between the theoretical time difference and the actual time difference between the measurement point and the lidar is marked as the obstacle value. The measurement area is marked as a safe area or a risk area based on the obstacle value. The specific process of marking a measurement area as a safe area or a risk area includes: comparing the obstacle value with a preset obstacle threshold; if the obstacle value is less than the obstacle threshold, the corresponding measurement point is marked as a freeway point; if the obstacle value is greater than or equal to the obstacle threshold, the corresponding measurement point is marked as an obstacle point; the ratio of the number of obstacle points to the number of measurement points in the measurement area is marked as the obstacle coefficient, and the obstacle coefficient is compared with a preset obstacle threshold; if the obstacle coefficient is less than the obstacle threshold, it is determined that there are no obstacles in the measurement area, and the corresponding measurement area is marked as a safe area; if the obstacle coefficient is greater than or equal to the obstacle threshold, it is determined that there are obstacles in the measurement area, and the corresponding measurement area is marked as a risk area. The specific process of comparing the monitoring and identification results of obstacles includes: if both the measurement area and its corresponding identification area are marked as safe areas, the passage risk of the measurement area is determined to meet the requirements; if both the measurement area and its corresponding identification area are marked as risk areas, the passage risk of the measurement area is determined to not meet the requirements, a braking control signal is generated and sent to the client in the train driver's cab; otherwise, the monitoring and identification results are determined to be inconsistent, a braking control signal is generated and sent to the client in the train driver's cab, and a verification optimization signal is generated and sent to the verification optimization module.

2. The obstacle identification based adaptive asynchronous control system for railway freight trains of claim 1, wherein, The specific process of obstacle identification using a panoramic camera includes: capturing images of the train's direction of travel using the panoramic camera to obtain a traveling image; enlarging the traveling image into a pixel grid image and performing grayscale transformation; segmenting the traveling image according to the measurement area to obtain several recognition areas; performing contour segmentation on the recognition areas using a binary method to obtain several reference areas; acquiring the recognition features of the reference areas, including the shape, maximum height, area value, and average grayscale value of the reference areas; and marking the corresponding recognition areas as risk areas or safe areas based on the recognition features.

3. The obstacle identification based adaptive asynchronous control system for railway freight trains of claim 2, wherein, The specific process of marking an identified area as a risk area or a safe area includes: comparing the identification features of the control area with the comparison features of all obstacle types in the database; if the identification features of the control area correspond to the comparison features of any obstacle type, then the identification area to which the corresponding control area belongs is marked as a risk area; otherwise, the identification area to which the corresponding control area belongs is marked as a safe area.

4. The obstacle identification based adaptive asynchronous control system for railway freight trains of claim 3, wherein, The specific process of the verification and optimization module for verifying and optimizing the obstacle monitoring and identification process of railway freight trains includes: generating an optimization period, obtaining the number of times the monitoring results and identification results of the same railway track are inconsistent within the optimization period and marking them as monitoring optimization values, and comparing the monitoring optimization values ​​with preset monitoring optimization thresholds: if the monitoring optimization value is less than the monitoring optimization threshold, it is determined that the railway track does not have monitoring optimization characteristics; if the monitoring optimization value is greater than or equal to the monitoring optimization threshold, it is determined that the railway track has monitoring optimization characteristics, and the lidar of the railway track is maintained or updated.

5. The obstacle identification based adaptive asynchronous control system for railway freight trains of claim 4, wherein, The specific process of the verification and optimization module for verifying and optimizing the obstacle monitoring and identification process of railway freight trains also includes: obtaining the number of times the monitoring results and identification results of the same railway freight train are inconsistent within the optimization period and marking them as identification optimization values; comparing the identification optimization values ​​with the preset identification optimization threshold; if the identification optimization value is less than the identification optimization threshold, it is determined that the railway freight train does not have identification optimization features; if the identification optimization value is greater than or equal to the identification optimization threshold, it is determined that the railway freight train has identification optimization features, and the panoramic camera of the railway freight train is maintained or updated.

Images