Railroad vehicle approach detection system

Abstract

The present invention relates to a railway vehicle approach detection system utilizing artificial intelligence technology for safe railway maintenance, comprising: an image device installed at a point to detect an approaching railway vehicle during track maintenance work for the safe operation of railway vehicles and capturing images; a calculation unit that detects a moving object from captured image information to obtain an optical flow image and calculates the speed at which the object moves; and a control unit that comprehensively considers the speed and position information of the moving object detected by a measurement unit monitoring the same point to determine whether a railway vehicle has entered, and notifies the control room and the workers on the track of a warning message and the time of arrival of the railway vehicle at the track maintenance work site. The invention has the advantage of reducing the risk of accidents caused by frequent errors in the method of using signals from existing detection sensors and reducing the cost required for sensor maintenance.

Description

Technology Field

[0001] The present invention relates to a railway vehicle approach detection system designed to eliminate frequent accidents involving railway vehicles and workers by installing a sensor capable of detecting the shape and driving speed of a vehicle along the track to ensure the safety of maintenance workers during maintenance work on the railway track, removing noise from the signal to confirm the accurate entry of the vehicle, and calculating the time it takes for the vehicle to reach the maintenance work site to provide sufficient evacuation time to the workers and track managers. Background Technology

[0002] To prevent vehicle accidents that may occur within railway track work sections, the most commonly used method involves stationing a checker on one side of the track and communicating with workers via hand signals when a railway vehicle approaches.

[0003] In addition, a train monitor that moves along the track to a warning point, warns the train driver that track work is in progress, detects the train, and outputs train approach information, and a remote controller that controls the train monitor to move to the warning point, receives train approach information from the train monitor, and warns the worker that the train is approaching, have been used, but this also has the problem that the remote controller may malfunction under bad weather conditions or at night.

[0004] Next, regarding a train approach warning system that detects naturally radiated noise signals from the power room of an electric train to warn of an approaching train, the technology includes a sensor module equipped with an antenna installed facing a propulsion control device installed in the power room of the electric train and receiving noise signals radiated from said propulsion control device; and a warning light that is illuminated by a driving signal output from said sensor module. However, this technology also has the disadvantage that the premise that the surrounding environment must be quiet enough to detect the noise must be satisfied, and if this is not the case, there is a high possibility of false alarms caused by other noise sources.

[0005] Next, there is a technology that improves worker safety by overcoming the limitations of information transmission by signalmen due to noise or worker concentration at the trackside work site, by using thermal imaging cameras and accelerometers (vibration sensors) to analyze movement information of trains approaching from a distance and issuing warnings in advance via flashing lights (alarms) and mobile communication terminals (mobile phones, PDAs, etc.) for the safety of workers (maintenance workers, etc.) working in the trackside work area. However, this method has the disadvantage that it is impossible to accurately confirm whether a railway vehicle is entering using the above method when the temperature of the railway vehicle and the ambient temperature are the same during the hot summer, or when multiple railway vehicles are entering the vicinity in succession.

[0006] Finally, while there is a traditional method of confirming a train's entry into the track based on railway signals, these signals operate by zone, and if the maintenance zone falls within the same area as the train's entry point, entry cannot be accurately confirmed. Furthermore, considering the frequent failures of track insulators, the current situation is that this method cannot completely protect track maintenance workers performing maintenance tasks on the track. Prior art literature

[0007] Korean Patent Publication No. 10-2008924 (Registered Aug. 02, 2019) Korean Patent Publication No. 10-2265726 (Registered June 10, 2021) Korean Patent Publication No. 10-0106186 (Registered July 14, 2021) Korean Patent Publication No. 10-1830388 (Registered Feb. 12, 2018) The problem to be solved

[0008] Conventional railway vehicle entry detection systems rely on signals from detection sensors installed on the tracks. When sensors located on tracks with high traffic are exposed for extended periods to extreme outdoor conditions with significant temperature and humidity fluctuations, they deteriorate rapidly, and problems frequently occur, particularly in the tracks transmitting electrical signals. This leads to frequent errors and incurs significant time and cost for sensor maintenance. Portable video sensors are limited to installation locations within 1 to 2 km of the work site, making them unsuitable for high-speed train operations, and they suffer from a rapid decline in nighttime recognition rates. Therefore, we aim to provide a railway vehicle approach detection system that can be installed without restrictions at a location sufficiently far from track maintenance work to prevent accident risks, is stable and error-free, and is not constrained by external environmental conditions. This system reliably detects whether a railway vehicle is approaching and promptly notifies the results.

[0010] The technical problems that the present invention aims to solve are not limited to those mentioned above, and other unmentioned technical problems will be clearly understood by those skilled in the art to which the present invention belongs from the description below. means of solving the problem

[0011] A railway vehicle approach detection system according to the present invention comprises: an image unit installed spaced apart on one side of a track to identify the entry of a railway vehicle into a predetermined position on the track; a plurality of measurement units installed on one side of the image device to transmit a signal to the entry point of the railway vehicle and receive a reflected signal of said signal; a database loaded with machine-learned data to determine whether a railway vehicle has entered from the measurement information of the image device; and a calculation unit that performs a first step of confirming the entry of a railway vehicle into the track by comparing first information received from the image unit with the data loaded in the database, a second step of confirming the entry of a railway vehicle into the track by calculating the entry speed of the railway vehicle using the first information received from the image unit, and a third step of confirming the entry of a railway vehicle into the track by generating a continuous trajectory of the movement of the railway vehicle using second information received from the plurality of measurement units. It is composed of a control unit that, when the calculation result of the above-mentioned calculation unit confirms that the railway vehicle has entered the track, calculates the arrival time of the railway vehicle from a predetermined position on the track to a maintenance work site located on one side of the track and transmits a warning signal to the control room and the maintenance work site.

[0012] In an embodiment of the present invention, the first step comprises: arranging the first information of the image unit according to the elapsed time of a predetermined time and capturing an object moving above a reference value; comparing the moving object with an image of a railway vehicle loaded in the database to confirm the entry of the railway vehicle; and overlapping and arranging the outlines of the object according to the elapsed time of a predetermined time to create an optical flow.

[0013] In an embodiment of the present invention, the second step comprises: determining the time at which the outer edge end of the optical flow first enters the image unit; determining the time at which the outer edge end of the optical flow leaves the image unit; calculating the driving speed of the measurement target from the first captured time and the leaving time; and determining the measurement target as a railway vehicle if the driving speed does not exceed a reference value of the track's operating speed.

[0014] In an embodiment of the present invention, the third step comprises: a step in which the plurality of measuring units measure the shortest distance between each measuring unit and the track; a step in which the shortest distance between the trajectory and each measuring unit is measured; a step in which the entry speed of the trajectory is measured from each measuring unit; and a step in which the shortest distance between the measuring unit and the track, the shortest distance between the measuring unit and the trajectory, and the entry speed of the trajectory match the entry speed measured by the imaging unit, thereby determining that the trajectory is a railway vehicle.

[0015] In an embodiment of the present invention, the plurality of measuring units is characterized as being any one of a LIDAR that detects the speed, direction, and distance of an object by receiving a reflected signal after laser irradiation, a RADAR that detects the speed, direction, and distance of an object by receiving a reflected signal after irradiating electromagnetic waves, and an ultrasonic sensor that detects the speed, direction, and distance of an object by receiving a reflected signal after irradiating ultrasonic waves. Effects of the invention

[0016] This invention is effective in preventing or reducing train accidents involving track workers that may occur due to sensor malfunctions during track operations. In existing detection systems, when a sensor malfunction or breakdown occurs, time, effort, and cost are incurred to investigate the site or re-inspect the entire system to identify the cause; this invention effectively reduces these costs. Frequent sensor malfunctions disrupt railway operation schedules, causing damage to railway operators and inconvenience to passengers, but this invention significantly reduces these issues. Furthermore, this invention is not limited to railway work environments but can be universally applied to other fields for detecting approaching objects.

[0018] The effects of the present invention are not limited to the effects described above, and should be understood to include all effects that can be inferred from the configuration of the invention described in the detailed description of the invention or the claims. Brief explanation of the drawing

[0019] Figure 1 shows a flowchart of the entire railway vehicle approach detection system. Figure 2 shows a schematic diagram representing the entire configuration of a railway vehicle approach detection system. Figure 3 is a diagram illustrating the process of the computing unit deriving optical flow from a moving object and extracting an image area during the first stage of railway vehicle approach detection. FIG. 4-1 is a diagram illustrating the process in which, in the second stage of railway vehicle approach detection, the computing unit captures the time when the railway vehicle first appears within the field of view of the camera image and the time when it reaches the end of the camera field of view in order to calculate the speed. Figure 4-2 is a diagram showing the dimensions that must be substituted into a formula to calculate the distance traveled by the railway vehicle during the time difference T1 to T2 in the process where the calculation unit calculates the speed during the second stage of the railway vehicle approach detection process. FIG. 5-1 is a diagram showing that in the three-stage process of detecting railway vehicle approach, when a railway vehicle enters a curved track, a plurality of measuring units receive their respective reflected signals and combine them to generate a movement trajectory. FIG. 5-2 is a diagram showing that in the three-stage process of detecting railway vehicle approach, when a railway vehicle enters a straight track, a plurality of measuring units receive reflected signals to generate a movement trajectory in a straight direction. Figure 6 shows a flowchart illustrating the overall flow of the railway vehicle approach detection system. Figure 7 shows a flowchart illustrating the first stage of the railway vehicle approach detection process. Figure 8 shows a flowchart illustrating the two-stage process of railway vehicle approach detection. Figure 9 shows a flowchart illustrating the three-stage process of detecting railway vehicle approach. Specific details for implementing the invention

[0020] Hereinafter, embodiments of the present invention will be described with reference to the attached drawings.

[0021] The parts necessary to understand the operation and function according to the present invention will be explained in detail, focusing on the parts necessary for understanding.

[0022] In describing the embodiments of the present invention, explanations of technical details that are well known in the technical field to which the present invention belongs and are not directly related to the present invention are omitted.

[0023] This is intended to convey the essence of the invention more clearly without obscuring it by omitting unnecessary explanations.

[0024] In addition, when describing the components of the present invention, different reference numerals may be assigned to components with the same name depending on the drawing, and the same reference numeral may be assigned even to different drawings.

[0025] However, even in such cases, it does not mean that the component has different functions depending on the embodiment or has the same function in different embodiments, and the function of each component should be determined based on the description of each component in the respective embodiment.

[0026] Furthermore, technical terms used in this specification shall be interpreted in the sense generally understood by those skilled in the art to which the present invention pertains, unless specifically defined otherwise in this specification, and shall not be interpreted in an overly broad or overly narrow sense.

[0027] Additionally, singular expressions used in this specification include plural expressions unless the context otherwise indicates.

[0028] In this application, terms such as "composed" or "comprising" should not be interpreted as necessarily including all of the various components or steps described in the specification, and should be interpreted as potentially excluding some of the components or steps, or including additional components or steps.

[0030] Figure 1 shows the overall flow of a railway vehicle approach detection system. In the present invention, a railway vehicle approaching the track is detected and verified in three stages.

[0031] The first step of the verification process is to determine whether a railway vehicle has entered by comparing the image of a moving object and the optical flow obtained through the information received from the video unit (100) with the data stored in the database (300).

[0032] The second verification process is to determine whether a railway vehicle enters by comparing the speed of a moving object calculated by the calculation unit (400) through the image information received from the image unit (100) with the track's scheduled speed. Here, the scheduled speed is a value obtained by dividing the travel distance of a means of transportation by the time required including the stopping time, and means a value obtained by dividing the transport distance by the total driving time including the stopping time.

[0033] The third verification process involves comparing the speed at which a moving object enters a trajectory received by the measuring unit (200) with the speed of the moving object calculated by the calculation unit (400) above, and comparing the distance between the trajectory of the moving object received by the measuring unit (200) and the measuring unit with the distance between the measuring unit and the track measured by the measuring unit to determine whether a railway vehicle has entered.

[0034] In existing methods for detecting the entry of railway vehicles, there was a probability of an accident occurring because the entering railway vehicle could not be detected due to sensor errors, etc., but the present invention has the advantage of being able to detect the approach of a railway vehicle more reliably and surely by presenting a three-step verification method.

[0036] FIG. 2 shows a schematic diagram illustrating the entire configuration of a railway vehicle approach detection system according to the present invention. An image unit (100) and a plurality of measurement units (200) are configured to be connected in a state where they can exchange data with the computation unit (400) and the control unit (500), which are connected to a machine-learned database (300) for railway vehicle approach detection. When a railway vehicle (10) is detected approaching from an installed location, the system notifies the control room (610) and the work site performing track maintenance work to prevent accidents from occurring.

[0037] The reason for attempting to detect the approach of a railway vehicle (10) through the above-mentioned video unit (100) and the above-mentioned measurement unit (200) is that the conventional method of detecting the approach of a railway vehicle by attaching sensors to the tracks is prone to sensor failures and errors. This is because it consumes a significant amount of time and cost to visit each site individually to investigate or re-examine the entire system in order to identify the cause of the malfunction.

[0038] The above-mentioned video unit (100) and the above-mentioned measurement unit (200) are not fixed but movable and can be installed immediately where needed. To protect workers at a site requiring track maintenance work, an optimal location is determined where the approach of a railway vehicle (10) must be detected, and the movable video unit (100) and the plurality of measurement units (200) are installed there. The above-mentioned video unit (100) and the plurality of measurement units (200) are connected to the above-mentioned computation unit (400) and the above-mentioned control unit (500), and are connected to the control room (610) and the track maintenance work site so as to transmit the results of the railway vehicle approach detection.

[0039] The above-mentioned video unit (100) for capturing video is positioned at a point on one side of the track at a location where the approach of a railway vehicle is to be detected, at a certain distance from the centerline of the track. The above-mentioned video unit (100) is equipped with a camera having a predetermined viewing angle, and is installed according to the predetermined specifications for the distance (LD) from the centerline of the track in a direction perpendicular to the track and the angle (α) formed by the camera direction and the track direction.

[0040] The above database (300) contains an artificial intelligence program

[0041] a) Receive video information provided along with time information and extract corresponding photos at set time intervals

[0042] b) After arranging the extracted photos according to the passage of time, classify image regions where the distance traveled in any direction during the passage of time is less than or equal to a threshold as static images.

[0043] c) After arranging the extracted photos according to the passage of time, classify image regions where the distance moved in any direction during the passage of time exceeds a threshold as dynamic images.

[0044] d) Distinguish railway vehicles by comparing various types of images classified as dynamic images with railway vehicle images by type.

[0045] Data learned through the machine learning process is loaded.

[0047] FIG. 3 illustrates the process in which a processing unit processes first information received from the image unit (100) during the first stage verification process of railway vehicle approach detection, and shows the process of deriving optical flow from a moving object captured by the image unit (100) and extracting an image area. The processing unit (400) performs a first stage of confirming whether a moving object that has entered the track is a railway vehicle by comparing the first information received from the image unit (100) with data loaded into the database (300) using machine learning.

[0048] The above calculation unit (400)

[0049] e) Receiving first information from the above image unit (100)

[0050] f) dynamic images of the railway vehicle (10) are extracted at regular time intervals, and an optical flow is generated by overlapping these images.

[0051] g) Extract the outline of the image area from the optical flow of the railway vehicle (10).

[0052] h) Compare with machine-learned data in the above database

[0053] A process is performed to first verify whether the object entering is a railway vehicle.

[0055] FIG. 4-1 shows the second stage verification process of railway vehicle approach detection, and the process of the calculation unit (400) capturing the time when the railway vehicle (10) first appears within the field of view of the camera image and the time when it reaches the end of the field of view of the camera in order to calculate the speed of the entering object.

[0056] The above calculation unit (400)

[0057] i) Detect the corresponding image frame in the received original image to determine when the point in time (T1) when the outline extracted from the optical flow of the railway vehicle (10) first appears on the image screen as the object moves, and determine the exact time.

[0058] j) Detecting the corresponding image frame in the received original image to determine when the point in time (T2) when the outline extracted from the optical flow of the railway vehicle (10) reaches the edge of the field of view of the image screen as the object moves, and determining the exact time.

[0059] Through this process, the time difference (ΔT = T2 - T1) from when the edge of the image outline of a moving object is first captured in the video until it leaves the video screen is calculated.

[0061] FIG. 4-2 illustrates the second stage verification process of railway vehicle approach detection, and shows the process in which the calculation unit (400) calculates the speed of a moving object. It shows the dimensions that must be substituted into the formula to calculate the distance traveled by the railway vehicle during the time difference T1 to T2.

[0062] The above calculation unit (400)

[0063] k) The distance the camera is located from the centerline of the track in a direction perpendicular to the track (L D ) and, the angle (α) formed by the camera direction and the track direction, the camera's field of view (β), and the distance (L) from the track centerline to the side of the railway vehicle W), camera view distance (L) representing the actual distance traveled by the railway vehicle through movement within the camera's field of view. V ), camera out-of-view distance (L), which indicates the distance the railway vehicle travels from the point it moves out of the camera's field of view to the location where the camera is installed. R The process of receiving ) from the user

[0064] l) A process of referring to the result of calculating the time difference from when a railway vehicle appears on the video screen until it reaches the edge of the screen through the above optical flow and image region generation.

[0065] m) Derived from the principles of trigonometric functions by referring to Fig. 4, which illustrates the process of calculating the entry speed of a railway vehicle

[0066] [Formula 1]

[0067] at,

[0068] [Equation 2]

[0069] To obtain,

[0070] Derived from the principles of trigonometric functions

[0071] [Equation 3]

[0072] to

[0073] L R By substituting, the actual distance traveled from when the railway vehicle appears on the video screen until it reaches the edge of the video screen is

[0074] [Equation 4]

[0075]

[0076] Obtaining this, and dividing it into time intervals from when the railway vehicle appears on the video screen until it reaches the end, regarding the speed of the approaching railway vehicle

[0077] [Formula 5]

[0078]

[0079] The process of obtaining

[0080] A second verification step is performed to determine that the object (10) entering through the calculation of the entry speed of the object (10) is a railway vehicle if the calculated entry speed does not exceed the reference value of the track's speed.

[0081] The above calculation unit (400) receives distance information from the user from the location where the above video unit (100) is installed to the track maintenance work site, and divides this by the above-calculated entry speed (V) to calculate the time it takes for the railway vehicle (10) to reach the track maintenance work site. When sending a warning message to the control room (610) and the track maintenance work site to notify the entry of the railway vehicle, the time it takes for the railway vehicle (10) to reach the track maintenance work site, calculated through this process, is also transmitted.

[0083] FIG. 5-1 illustrates the third stage of the railway vehicle approach detection process, showing that when a railway vehicle (10) enters a curved track, a plurality of measuring units (200) receive their respective reflected signals and combine them to generate a movement trajectory. A specific measuring unit (200) also measures the distance between other measuring units and the track.

[0084] The plurality of measurement units (200) above provide a means to determine whether a railway vehicle has entered the track by another method in addition to the information received from the image unit (100).

[0085] The third step to check whether a railway vehicle has entered the track is

[0086] n) A step in which the plurality of measuring units (200) measure the shortest distance between each of the measuring units (200) and the track;

[0087] o) A step of measuring the shortest distance between each of the measurement units and the trajectory of a moving object calculated by the calculation unit (400) through signals received from the plurality of measurement units (200);

[0088] p) A step of measuring the entry speed of the trajectory from each of the above measuring units (200);

[0089] q) A step of comparing the shortest distance between the measuring unit (200) and the track with the shortest distance between the measuring unit (200) and the trajectory.

[0090] r) A step of determining that the trajectory is a railway vehicle if the entry speed of the trajectory matches the entry speed measured by the image unit;

[0091] It goes through.

[0093] FIG. 5-2 shows the third step of the railway vehicle approach detection process, illustrating that when a railway vehicle enters a straight track, a plurality of measuring units (200) receive reflected signals to generate a movement trajectory in a straight direction. The approach of a railway vehicle can be determined through the same process as described in FIG. 5-1. In a straight track, the number of measuring units installed can be reduced, and it is possible to determine whether an object entering is a railway vehicle simply by comparing the direction information of the straight line and the track created by connecting the movement trajectories of the moving object.

[0095] FIG. 6 shows a flowchart representing the overall flow of the railway vehicle approach detection system. The first information received from the image unit (100) and the second information received from the measurement unit (200) are received by the calculation unit. The calculation unit (400) verifies whether the object entering the third stage is a railway vehicle.

[0096] If the first stage of verification is passed, it proceeds to the second stage of verification, and if the second stage of verification is passed, it proceeds to the third stage of verification. If it fails to pass each stage, it is determined that the object entering is not a railway vehicle, and it returns to the beginning to repeat the activity of receiving the first information and the second information from the image unit (100) and the measurement unit (200).

[0097] When it is confirmed that the entering object is a railway vehicle after passing all three stages of verification, a warning message is transmitted to the control room and the track maintenance work site through the control unit.

[0099] FIG. 7 shows a flowchart illustrating the first stage of the railway vehicle approach detection process. It involves the process of capturing a moving object (S100), verifying it by comparing it with a database (S200), and generating an optical flow (S300).

[0101] Figure 8 shows a flowchart illustrating the second stage of the railway vehicle approach detection process. Through the process of determining the time when the end of a video image extracted from a moving object first appears in the video (S400) and the process of determining the time when the end leaves the video (S500), the process of calculating the entry speed of the entering object (S600) and the process of comparing this entry speed with the direction speed (S700) are performed.

[0103] FIG. 9 shows a flowchart illustrating the three-stage process of detecting the approach of a railway vehicle. Information necessary for the three-stage verification is obtained through the following steps: a process (S800) in which the shortest distance between a plurality of measuring units (200) and the track is measured at each measuring unit (200); a process (S900) in which the distance between the measuring unit and the moving object's movement trajectory received from the measuring unit is measured; and a process (S1000) in which the measuring unit measures the entry speed of the moving object. The results of the distance comparison S800 and S900 are compared, and the results of the speed comparison S600 and S1000 are compared to finally verify whether the approaching object is a railway vehicle.

[0105] 1. First embodiment

[0106] In this invention, only the first method is applied among the three verification steps proposed to determine whether an entering object is a railway vehicle (10). It is possible to determine whether a moving object entering the track is a railway vehicle simply by comparing the first information received from the image unit (100) with the data loaded into the database (300) via machine learning. In order to prepare for various situations that may occur in the machine learning

[0107] s) Learning about various driving sections, such as when a railway vehicle travels on a straight section, when a railway vehicle operates on a curved section, and when a railway vehicle exits a tunnel section.

[0108] t) Learning about various situations or environmental conditions, such as when a railway vehicle runs in the rain, snow, or fog.

[0109] u) Learning about various lighting conditions, such as when railway vehicles run at dawn or sunrise, when railway vehicles run in the evening or sunset, and when railway vehicles run in complete darkness without lighting.

[0110] Through this, the ability to distinguish railway vehicles (10) can be improved.

[0111] The above optical flow can be generated from an image of a railway vehicle (10) and used to make a judgment to distinguish the railway vehicle (10), and the following

[0112] v) Learning to generate the above optical flow for cases where railway vehicles of various types and forms are in operation

[0113] w) Learning to generate the above optical flow by receiving input images of cases that are not railway vehicles, such as flying birds or airplanes.

[0114] x) Classify the various types of optical flows generated above into cases corresponding to railway vehicles and cases that are not railway vehicles, and train to selectively recognize the optical flows of railway vehicles.

[0115] Through this, the ability to distinguish railway vehicles (10) can be improved.

[0117] Second embodiment

[0118] In the present invention, the first and second verification steps are applied out of the three verification steps required to determine whether the entering object proposed in the present invention is a railway vehicle (10). In the first embodiment above, after confirming the entry of the railway vehicle through an image in the first step, the entry speed of the railway vehicle is checked in the second step to confirm whether the railway vehicle has entered, so that the determination can be made more certainly.

[0119] First, the entry of a railway vehicle is determined by comparing the first information received from the video unit (100), which is the process in the first verification stage, with the data loaded into the database (300) through machine learning.

[0120] In the second verification step, the outline of the optical flow is extracted from the entering railway vehicle, and the time taken for the end to first appear on the image screen and for the end to leave the image unit is determined; the distance traveled by the railway vehicle during the time taken is calculated using the formulas; and the speed of the entering railway vehicle is calculated by dividing the distance traveled by the time taken. If the driving speed calculated through the information received from the image unit (100) does not exceed the reference value of the track's operating speed, the process of determining the measurement target as a railway vehicle is performed.

[0121] Instead of stopping at a first verification step where the entering object is confirmed to be a railway vehicle (10) based only on the image, if a second verification step is taken to determine whether the entering object is a railway vehicle (10) by calculating the entry speed, it becomes possible to prepare more reliably for the entry of a railway vehicle.

[0123] Third embodiment

[0124] In order to determine whether the object entering is a railway vehicle (10) as proposed in the present invention, all three verification steps are taken without omission. In order to confirm the entry of a railway vehicle (10) most certainly and reliably and to deliver a warning message, it is best to go through all three steps 1 to 3 presented in the present invention. Following the first step of determining by comparing images received from the video and the second step of determining by the entry speed obtained through calculation based on information received from the video, the third step adds a step of comparing distance and speed information measured by the plurality of measurement units (200) to more certainly determine whether a railway vehicle has entered.

[0125] First, the entry of a railway vehicle is determined by comparing the first information received from the video unit (100), which is the process in the first verification stage, with the data loaded into the database (300) through machine learning.

[0126] In the process of the second verification step, the outline of the optical flow is extracted from the entering railway vehicle, the time taken for the end to exit the image section after it first appears on the image screen is determined, the distance traveled by the railway vehicle during the time taken is calculated using the formulas, the speed of the entering railway vehicle is calculated by dividing the distance traveled by the time taken, and this is compared with the scheduled speed to determine whether the railway vehicle has entered.

[0127] In the third verification step, the plurality of measuring units (200) each measure the shortest distance between the track and the measuring unit (200); the shortest distance between the trajectory obtained through the measuring unit (200) and each measuring unit (200); the entry speed of the trajectory from each measuring unit; and if the shortest distance between the measuring unit (200) and the track, the shortest distance between the measuring unit (200) and the trajectory, and the entry speed of the trajectory match the entry speed measured by the image unit, the step of determining that the object moving along the trajectory is a railway vehicle (10). Explanation of the symbols

[0128] 10: Railway rolling stock 100 : Video section 200 : Measurement section 300 : Database 400 : Operation unit 500 : Control unit 610 : Control Room

Claims

Claim 1 An image unit (100) installed spaced apart on one side of the track to identify the entry of a railway vehicle (10) into a predetermined position on the track; a plurality of measurement units (200) installed on one side of the image unit (100) to transmit a signal to the entry point of the railway vehicle and receive a reflected signal of the signal; a database (300) loaded with machine-learned data to determine whether the railway vehicle has entered from the measurement information of the image unit (100); a first step of confirming the entry of the railway vehicle (10) into the track by comparing first information received from the image unit (100) with the data loaded in the database (300); a second step of confirming the entry of the railway vehicle into the track by calculating the entry speed of the railway vehicle (10) using the first information received from the image unit (100); and a third step of performing calculations to confirm the entry of the railway vehicle (10) into the track by generating a continuous trajectory of the movement of the railway vehicle (10) using second information received from the plurality of measurement units (200). The method includes a calculation unit (400); and a control unit (500) that, when the calculation result of the calculation unit (400) confirms that the railway vehicle (10) has entered the track, calculates the arrival time of the railway vehicle from a predetermined position on the track to a maintenance work site located on one side of the track and transmits a warning signal to the control room (610) and the maintenance work site. The first step comprises: a step (S100) of capturing an object moving above a reference value after arranging the first information of the image unit (100) according to the elapsed time of a predetermined time; a step (S200) of confirming the entry of the railway vehicle by comparing the moving object with a railway vehicle image loaded in the database (300); and a step (S300) of creating an optical flow by overlapping and arranging the outlines of the object according to the elapsed time of a predetermined time. The second step comprises: a step (S400) of determining the time when the end of the outline of the optical flow first enters the image unit (100); and the end of the outline of the optical flow entering the image unit (100). A step of determining the time of departure (S500); a step of determining the entry speed of the measurement target from the initial entry time and the time of departure (S600);A railway vehicle approach detection device characterized by including a step (S700) of determining that the measurement target is a railway vehicle when the entry speed does not exceed the reference speed of the track. Claim 2 delete Claim 3 delete Claim 4 A railway vehicle approach detection device according to claim 1, wherein the third step comprises: a step (S800) in which the plurality of measuring units (200) measure the shortest distance between each measuring unit (200) and the track; a step (S900) in which the shortest distance between the trajectory and each measuring unit (200); a step (S1000) in which the entry speed of the trajectory is measured from each measuring unit (200); and a step (S1100) in which the shortest distance between the measuring unit (200) and the track, the shortest distance between the measuring unit (200) and the trajectory, and the entry speed of the trajectory match the entry speed measured by the calculation unit (400). Claim 5 A railway vehicle approach detection device according to claim 1, characterized in that the plurality of measuring units are any one of a LIDAR that detects the speed, direction, and distance of an object by receiving a reflected signal after laser irradiation, a RADAR that detects the speed, direction, and distance of an object by receiving a reflected signal after irradiating electromagnetic waves, and an ultrasonic sensor that detects the speed, direction, and distance of an object by receiving a reflected signal after irradiating ultrasonic waves.

Images