Train leading device, train leading method and train leading system

Abstract

The invention relates to a train leading device, a train leading method and a train leading system, and relates to the field of rail transit. The vehicle leading device comprises a sensor unit, a communication unit and an edge calculation unit. The sensor unit is configured to be used for collecting railway freight station yard data; the communication unit is configured to be used for receiving a train control instruction and train running information sent by the train control host, and the train control instruction comprises train leading operation information; the edge calculation unit is configured to input the railway freight station yard data into the corresponding state monitoring algorithm model according to the train leading operation information to obtain corresponding to-be-detected target information; and inputting the to-be-detected target information and the train driving information into the behavior prediction algorithm model to obtain a train driving decision suggestion. Through the embodiment of the invention, the problems of safety risk, inaccurate road condition observation, low operation efficiency and the like caused by incapability of intelligently monitoring the road condition in front of the running of the train and only manual observation of the road condition in the running direction of the train in the train leading operation process are solved.

Description

Technical Field

[0001] The present invention relates to the field of rail transportation, and in particular to a train leading device, a train leading method and a train leading system. Background Art

[0002] Railway freight stations are the basic places for railway freight transportation and the key places for connecting with other modes of transportation. Therefore, they play an important role in railway freight transportation and the entire logistics system. Railway freight stations are places for freight companies to dispatch materials. When trains perform car-leading operations (for example, coupling operations and pushing operations) at railway freight stations, there are extremely high safety risks. Because when the train is performing pushing operations and coupling operations, the driver cannot directly obtain the road conditions ahead of the train. They must rely on the leader to ride in the last carriage of the train to monitor the road conditions ahead of the train in real time and communicate with the driver through intercom. This brings great safety risks to the train's pushing operations and coupling operations. In addition, because the pushing operations and coupling operations rely heavily on the leader to observe the road conditions ahead of the train, the judgment of the road conditions ahead of the train is inaccurate, prone to errors, and inefficient. Summary of the invention

[0003] In view of this, an embodiment of the present invention provides a train leading device, a train leading method and a train leading system, which solve the problems of safety risks, inaccurate road condition observation and low operation efficiency caused by the inability to intelligently monitor the road condition ahead of the train during the train leading operation and the only way to manually observe the road condition in the direction of the train.

[0004] On the one hand, an embodiment of the present invention provides a vehicle leading device, the vehicle leading device comprising: A sensor unit configured to collect railway freight station data; The communication unit is configured to receive the train control command and train travel information sent by the train control host, wherein the train control command includes train leading operation information; An edge computing unit is configured to input the railway freight station data into a corresponding state monitoring algorithm model according to the train lead operation information to obtain corresponding target information to be detected; The edge computing unit is also configured to input the target information to be detected and the train travel information into the behavior prediction algorithm model to obtain train travel decision recommendations.

[0005] In some embodiments, the railway freight yard data includes an environmental image of the railway freight yard and all object status data in the railway freight yard; The sensor unit includes a camera and a radar; wherein, The camera is configured to collect environmental images of railway freight stations; The radar is configured to collect all target status data in the railway freight yard.

[0006] In some embodiments, the communication unit is further configured to send the train driving decision suggestion to the train control host; or The communication unit is also configured to send the collected railway freight station data to the server.

[0007] On the other hand, an embodiment of the present invention further provides a train leading method, which is applied to the leading device in any of the above embodiments, and the train leading method includes: Collect railway freight station data; Receive train control instructions and train travel information sent by the train control host. The train control instructions include train leading operation information; According to the train lead operation information, the railway freight station data is input into the condition monitoring algorithm model to obtain the target information to be detected; The target information to be detected and the train driving information are input into the behavior prediction algorithm model to obtain train driving decision recommendations.

[0008] In some embodiments, the train leading operation information includes any one of push operation information and coupling operation information; The state monitoring algorithm model includes a first sub-state monitoring algorithm model and a second sub-state monitoring algorithm model; According to the train lead operation information, the railway freight station data is input into the corresponding state monitoring algorithm model to obtain the corresponding target information to be detected, including: In response to the train leading operation information being a push operation, the railway freight station data is input into the first sub-state monitoring algorithm model to obtain obstacle information that affects the train from executing the push operation within the first boundary range; or, In response to the train leading operation information being a coupling operation, the railway freight station data is input into the second sub-state monitoring algorithm model to obtain obstacle information and to-be-coupled vehicle information within the second boundary range that affect the train's execution of the coupling operation.

[0009] In some implementations, in response to the train leading operation information being a push operation, the railway freight station data is input into the first sub-state monitoring algorithm model to obtain obstacle information within the first boundary range that affects the train from performing the push operation, including: In response to the train leading operation information being a push operation, the railway freight station data is input into the first sub-state monitoring algorithm model, so as to perform the following steps based on the first sub-state monitoring algorithm model: Extract track information and all target information from railway freight station data respectively; determining a first boundary based on the track information; According to the first boundary and all target information, obstacle information within the first boundary that affects the train's propulsion operation is determined.

[0010] In some implementations, in response to the train leading operation information being a coupling operation, the railway freight station data is input into the second sub-state monitoring algorithm model to obtain obstacle information and to-be-coupled vehicle information within the second boundary range that affect the train from executing the coupling operation, including: In response to the train leading operation information being a coupling operation, the railway freight station data is input into the second sub-state monitoring algorithm model, so as to perform the following steps based on the second sub-state monitoring algorithm model: Extract track information and all target information from railway freight station data respectively; determining a second boundary based on the track information; According to the second boundary and all target information, obstacle information and information of vehicles to be coupled that affect the coupling operation of the train within the second boundary are determined.

[0011] In some embodiments, before collecting railway freight station data, the method further includes: Collecting initial railway freight station data, and inputting the initial railway freight station data into a perspective correction model to obtain a corrected data collection perspective; The collection of railway freight station data includes: The railway freight station data is collected from the corrected data collection perspective.

[0012] On the other hand, an embodiment of the present invention further provides a train leading system, including a train leading device as in any of the above embodiments.

[0013] In some embodiments, the vehicle leading system further comprises: The train control host is configured to send train control instructions and train travel information to the train leader; The server is configured to train the artificial intelligence vision model according to the train leading operation information and the railway freight station data to obtain the state monitoring algorithm model; and train the artificial intelligence prediction model according to the target information to be detected and the train travel information to obtain the behavior prediction algorithm model.

[0014] On the other hand, an embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the method described in any of the above embodiments are performed.

[0015] The train leader of the embodiment of the present invention can be installed on the terminal carriage of a train, and collect the railway freight station data ahead of the train in real time through the sensor unit; after receiving the train control instruction and train travel information containing the train leading operation information sent by the train control host through the communication unit, the railway freight station data is input into the corresponding state monitoring algorithm model through the edge computing unit according to the train leading operation information to obtain the corresponding target information to be detected, and the obtained target information to be detected and the train travel information are input into the behavior prediction algorithm model to obtain the train travel decision suggestion, thereby realizing the real-time collection of the railway freight station data ahead of the train through the train leader, and automatically generating the train travel decision suggestion according to the real-time collected railway freight station data, so that the train can perform the train control operation according to the train travel decision suggestion, thereby reducing the labor cost, avoiding the problem that the manual observation of the road conditions ahead of the train is prone to errors and is greatly affected by the environment, and improving the safety and work efficiency of the train leading operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other embodiments can be obtained based on these drawings without paying creative work.

[0017] Figure 1a The figure is a schematic diagram of a train leading operation; Figure 1b A schematic diagram of the structure of a vehicle guide device provided by an embodiment of the present invention; Figure 2 A flowchart of a data transmission method provided by an embodiment of the present invention; Figure 3 A flow chart of a train leading method provided by an embodiment of the present invention; Figure 4 A flowchart of a vision correction method provided by an embodiment of the present invention; Figure 5 A flow chart of a train leading method provided by an embodiment of the present invention; Figure 6 A schematic structural diagram of a train leading system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the embodiments of the present invention are further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0019] It should be noted that all expressions using "first" and "second" in the embodiments of the present invention are for distinguishing two non-identical entities with the same name or non-identical parameters. It can be seen that "first" and "second" are only for the convenience of expression and should not be understood as limitations on the embodiments of the present invention. The subsequent embodiments will not explain this one by one.

[0020] Railway freight stations are the basic places for trains to transport goods. Trains can include locomotives and multiple carriages connected to the locomotives. Train leading operations can include coupling operations and pushing operations. Figure 1a In the example shown, the train includes a locomotive, and carriages 1 and 2 connected to the locomotive along the X direction and carriages 1' and 2' connected along the Y direction. When the train is performing the train leading operation, the train can push several carriages forward along the X direction or the Y direction. For the convenience of description, in this example, the train needs to push the carriage m along the X direction and the carriage m' in the Y direction as an example for description. When the train pushes the carriage m along the X direction, the driver's line of sight will be blocked by the carriage in front of the line of sight, and the road conditions in front of the train cannot be directly observed. When the train pushes the carriage m' along the Y direction, the train driver cannot directly observe the road conditions in front of the train. Therefore, no matter whether the train pushes the carriage along the X direction or the Y direction, the train driver cannot directly observe the road conditions in front of the train. A leader is required to ride in the end carriage to follow the train, observe the road conditions in front of the train instead of the driver, and communicate with the driver through the intercom. Therefore, there is a very high safety risk in the train leading operation.

[0021] The following uses the pushing operation and coupling operation as examples to explain in detail why the train leading operation has extremely high safety risks. The pushing operation mainly involves the train pushing the carriage from a certain position to the destination. In addition to the train pushing the carriage from a certain position to the destination, the coupling operation also requires constantly determining the distance between the train's end carriage and the vehicle to be coupled. Whether it is a pushing operation or a coupling operation, the train driver cannot directly observe the road conditions ahead of the train. The train leader must ride in the train's end carriage to monitor the train's situation in real time and communicate with the driver through intercom. This brings extremely high safety risks to the train leading operation and the train leader, especially in complex operating environments, such as at night or in bad weather, which greatly increases the safety risks. In addition, because the pushing operation and coupling operation rely heavily on the train leader to observe the road conditions ahead, the judgment of the road conditions ahead of the train is inaccurate, prone to errors, and inefficient.

[0022] In order to solve at least one of the above technical problems, a first aspect of an embodiment of the present invention provides a vehicle guiding device.

[0023] Figure 1b The structure diagram of a vehicle guide device provided by an embodiment of the present invention is shown in FIG. Figure 1b As shown, the train leader 10 provided by the embodiment of the present invention includes: a sensor unit 11, a communication unit 12 and an edge computing unit 13. The sensor unit 11 is configured to collect railway freight station data; the communication unit 12 is configured to receive the train control command and train travel information sent by the train control host, and the train control command includes the train leading operation information; the edge computing unit 13 is configured to input the railway freight station data into the corresponding state monitoring algorithm model according to the train leading operation information to obtain the corresponding target information to be detected; the target information to be detected and the train travel information are input into the behavior prediction algorithm model to obtain the train travel decision suggestion.

[0024] The train may include a locomotive and at least one carriage connected to the locomotive. The locomotive is equipped with a train control host for controlling the operation of the train.

[0025] The leader 10 may be arranged on the end car connected to the locomotive. In some examples, the leader 10 may be arranged on the side of the end car of the train away from the locomotive.

[0026] The sensor unit 11 may include devices such as cameras and radars that can collect railway freight station data. The railway freight station data may include environmental images of the railway freight station and all target status data in the railway freight station. Targets refer to all objects that appear at the railway freight site, such as people, vehicles, tracks, flowers, trees, etc. The target status data refers to whether the object is stationary or moving, and if it is a moving object, how fast the object moves.

[0027] The collected railway freight station data can be stored in the cache of the vehicle collector.

[0028] The communication unit 12 is used to realize the communication between the vehicle control device and an external unit, for example, a vehicle control host or a cloud server, etc. The communication unit 12 can be implemented based on 3G, 4G or 5G communication technology.

[0029] The train leading operation information may include the train leading operation information to be executed. A train control instruction may include one type of train leading operation information.

[0030] The edge computing unit 13 is deployed with a state monitoring algorithm model and a behavior prediction algorithm model. The state monitoring algorithm model is used to obtain the target information to be detected based on the railway freight station data. The behavior prediction algorithm model is used to obtain train travel decision suggestions based on the target information to be detected and the train travel information.

[0031] The target to be detected can be understood as an object that affects the train leading operation when the train is performing the train leading operation, such as a person, a vehicle, a stone, a box, etc. The target to be detected is different according to the different train operation modes. Here, we only take the train leading operation including the propulsion operation and the coupling operation as an example to illustrate the target to be detected. For example, when the train is performing the propulsion operation, the target to be detected is an obstacle that affects the propulsion of the train. However, when the train is performing the coupling operation, the target to be detected is the obstacle that affects the train's travel and the vehicle to be coupled. At this time, the vehicle to be coupled does not belong to an obstacle.

[0032] The train driving decision suggestion may be driving at a constant speed according to the current speed, accelerating to a certain speed, decelerating to a certain speed, or stopping, etc. The train driving decision suggestion may be sent to the control host through the communication unit, so that the control host can control the train according to the train driving decision suggestion.

[0033] The target information to be detected and the train driving decision suggestions can be saved in the cache of the train leader.

[0034] The embodiment of the present invention collects the railway freight station data when the train is leading the train in real time through the sensing unit, realizes the automatic detection of the road conditions ahead when the train is performing the train leading operation, does not need to monitor the road conditions manually, and improves the safety of the train leading operation. The embodiment of the present invention also processes the collected railway freight station data through the state monitoring algorithm model to obtain the corresponding target information to be detected, and intelligently analyzes the target information to be detected and the train travel information through the behavior prediction algorithm model to obtain the train travel decision suggestion, realizes the automatic processing of the real-time collected data, improves the accuracy of the train travel decision suggestion, avoids the problem of easy error and great environmental impact when manually observing the road conditions ahead of the train, and improves the safety and work efficiency of the train leading operation. The embodiment of the present invention also improves the processing speed of the real-time collected data through the pre-deployed state monitoring algorithm model, and improves the speed of generating the train travel decision suggestion through the pre-deployed behavior prediction algorithm model, which improves the real-time performance and reliability of the leading device as a whole.

[0035] The embodiment of the present invention replaces the train leader with a train leader device, realizes real-time detection of the road conditions ahead of the train and automatic generation of train travel suggestions, and realizes an intelligent train leader method.

[0036] In some embodiments, Figure 1b As shown, the sensor unit 11 includes a camera 111 and a radar 112. The camera is configured to collect environmental images of the railway freight station; and the radar is configured to collect all target status data in the railway freight station.

[0037] The camera can be a visible light camera or an infrared camera. Visible light cameras provide high-resolution visual information of the environment; infrared cameras perform well at night and in low-light conditions.

[0038] The radar can be a laser radar or a 4D millimeter-wave radar. The 4D millimeter-wave radar can work stably in various weather and lighting conditions and accurately measure the distance and speed of the target.

[0039] Through cameras and radars, the environmental data of railway freight station sites can be fully and accurately perceived. In particular, the sensor unit composed of visible light cameras, infrared cameras and 4D millimeter wave radars further improves the adaptability of the train leader, and can fully and accurately perceive the environmental data of railway freight station sites in different environments, improving the adaptability of the train leader to complex environments and improving the safety of the train leader when working in complex environments.

[0040] In some embodiments, Figure 1b As shown, the train leader 10 may further include a power supply 14 and a fixing unit 15. The power supply 14 may supply power to the train leader separately. The fixing unit 15 may fix the train leader on the train.

[0041] In some embodiments, Figure 1b As shown, the communication unit 12 is also configured to send the train driving decision suggestion to the train control host; or, The communication unit 12 is also configured to send the collected railway freight station data to the server.

[0042] The server may be a cloud server, and the communication unit 12 may communicate with the cloud server via a wireless network, which may be a wireless local area network, a 4G / 5G communication network, or the like.

[0043] In order to improve the data transmission efficiency of the train leader. The embodiment of the present invention can divide the operation mode of the train leader into an operating state and a non-operating state. In the operating state, the train leader only transmits key data related to train control, such as train driving decision recommendations. The collected railway freight station data, the data obtained based on the state monitoring algorithm model and the intermediate data generated in the state monitoring algorithm model, the data obtained based on the behavior prediction algorithm model and the intermediate data generated in the behavior monitoring algorithm model, can all be temporarily stored in the cache of the train leader. In the non-operating state, the train leader obtains all the cached data from the cache and transmits it to the train control host or server. Thereby, the data transmission efficiency of the train leader in the operating state is improved, the real-time performance of the data transmitted by the train leader is improved, and the problems of data delay, packet loss, etc. are avoided.

[0044] In some examples, when the vehicle leader is not in operation, it can obtain all cached data from the cache and transmit it to the cloud server for storage for subsequent use.

[0045] In some examples, the cache data upload process also has an interruption recovery function to ensure the integrity and reliability of data transmission.

[0046] Figure 2 A flowchart of a data transmission method provided by an embodiment of the present invention.

[0047] In some embodiments, the present invention Figure 1b The vehicle guide device 10 shown can implement the present invention as follows Figure 2 The data transmission method shown is as follows Figure 2 As shown, the data transmission method includes steps S201 to S207.

[0048] S201, receiving an operation status instruction sent by a vehicle control host, where the operation status instruction includes an operation status of a vehicle leader.

[0049] S202: Determine whether the operating state of the vehicle leader is in the operating state.

[0050] S203: When the operating state of the train leader is the operating state, the train driving decision suggestion is transmitted to the train control host.

[0051] S204: When the operating state of the vehicle leader is a non-operating state, obtain cache data from the cache and transmit the cache data to the cloud server.

[0052] S205, monitoring the cache progress to determine whether the cache data transmission is interrupted. When the running state of the vehicle leader switches from the non-running state to the running state, the cache process will be interrupted. If the cache interruption is detected, steps S206 and S207 are executed.

[0053] S206: When the cache data transmission is interrupted, save the cache data transmission progress.

[0054] S207. When the running state of the vehicle leader is switched from the running state to the non-running state, the cache data transmission progress is read, and the cache data transmission is resumed according to the cache data transmission progress until all the cache data are uploaded to the cloud server.

[0055] The above steps S201 to S207 may be executed by the communication unit 12 of the vehicle leader 10 .

[0056] The vehicle leader of the embodiment of the present invention provides a decoupled transmission strategy for operating and non-operating data. In the operating state, the vehicle leader only transmits the detection and decision result data related to the vehicle control to the vehicle control host, and other data, such as the data collected by the sensor unit and the intermediate data and detection results obtained based on the state monitoring algorithm model and the behavior monitoring algorithm model, can be temporarily stored in the cache, thereby reducing the occupation of the vehicle leader resources by data transmission, and improving the real-time data transmission and the operating efficiency of the vehicle leader. When the vehicle leader is in the non-operating state, all cached data is uploaded to the cloud server for storage for subsequent model optimization and data backtracking. If the vehicle leader switches from the non-operating state to the operating state during the data upload to the cloud server, the data transmission interruption operation is triggered, and the current cache data transmission progress is saved. When the vehicle leader switches from the operating state to the non-operating state, the data upload will be automatically resumed, thereby ensuring the integrity and reliability of the data transmission.

[0057] On the other hand, an embodiment of the present invention further provides a train leading method, which is applied to the train leading device in any of the above embodiments.

[0058] Figure 3 The present invention provides a flow chart of a train leading method. Figure 3 The train leading method shown in the figure can be implemented by the present invention as follows: Figure 1b The driver shown in the figure is executed. Figure 3 As shown, the train leading method includes steps S301-S304.

[0059] S301. Collect railway freight station data.

[0060] Step S301 can be performed by Figure 1b The sensor unit 11 of the vehicle leader 10 is shown to execute the operation. The railway freight station data is multi-dimensional data, and may include an environmental image of the railway freight station and all target status data in the railway freight station.

[0061] S302, receiving a train control instruction and train travel information sent by a train control host, wherein the train control instruction includes train leading operation information.

[0062] Step S302 can be performed by Figure 1b The communication unit 12 of the train leader 10 is shown to execute the train leader operation information. The train leader operation information may include any one of the coupling operation information and the propulsion operation information.

[0063] S303: Inputting railway freight station data into a status monitoring algorithm model according to train lead operation information to obtain corresponding target information to be detected.

[0064] Step S303 can be performed by Figure 1bThe edge computing unit 13 of the leader 10 shown in the figure executes. Different state monitoring algorithm models can be obtained by comparing the artificial intelligence visual model with the application scenario. That is, in the coupling operation scenario, a state monitoring algorithm model is obtained by comparing the artificial intelligence visual model with the historically collected railway freight station data. In the push operation scenario, another state monitoring algorithm model can be obtained by comparing the artificial intelligence visual model with the historically collected railway freight station data.

[0065] The training process of the condition monitoring algorithm model can be carried out on the cloud server, thereby saving the local resources of the vehicle leader.

[0066] S304: Input the target information to be detected and the train travel information into the behavior prediction algorithm model to obtain train travel decision suggestions.

[0067] Step S304 can be performed by the edge computing unit 13 of the train leader 10 shown in Figure 1. The behavior prediction algorithm model can be obtained by training the artificial intelligence prediction model based on the target information to be detected and the train travel information collected historically according to the state monitoring algorithm model. The training process can be performed on the cloud server, thereby saving local resources of the train leader.

[0068] By inputting the target information to be detected and the train driving information into the behavior prediction algorithm model, train driving decision suggestions can be obtained, such as constant speed, acceleration, deceleration and parking operations.

[0069] The embodiment of the present invention realizes automatic detection of the road conditions ahead of the train when the train performs the train leading operation by real-time collection of railway freight station data when the train performs the train leading operation, without manual monitoring of the road conditions, thereby improving the safety of the train operation. The embodiment of the present invention also processes the collected railway freight station data through the state monitoring algorithm model to obtain the target information to be detected, and intelligently analyzes the target information to be detected and the train travel information through the behavior prediction algorithm model to obtain the train travel decision suggestion, thereby realizing the automatic processing of the real-time collected data, improving the accuracy of the train travel decision suggestion, avoiding manual observation of the road conditions ahead of the train travel, and solving the problems of inability to intelligently monitor the road conditions ahead of the train travel during the train leading operation, and only manually observing the road conditions in the direction of train travel, which brings safety risks, inaccurate road condition observation, great environmental impact and low operation efficiency, thereby improving the safety and work efficiency of the train leading operation. The embodiment of the present invention also improves the processing speed of the real-time collected data through the pre-deployed state monitoring algorithm model, and improves the speed of generating the train travel decision suggestion through the pre-deployed behavior prediction algorithm model, thereby improving the real-time and reliability of the leading method as a whole.

[0070] The embodiment of the present invention replaces the train leader with a train leader device, realizes real-time detection of the road conditions ahead of the train and automatic generation of train travel suggestions, and realizes an intelligent train leader method.

[0071] In some embodiments, the train leading operation information mentioned in the embodiments of the present invention may include push operation information or coupling operation information; the state monitoring algorithm model may include a first sub-state monitoring algorithm model and a second sub-state monitoring algorithm model.

[0072] In some embodiments, inputting railway freight station data into a corresponding state monitoring algorithm model according to train lead operation information to obtain corresponding target information to be detected may include: In response to the train leading operation information being push operation information, the railway freight station data is input into the first sub-state monitoring algorithm model to obtain obstacle information that affects the train from executing the push operation within the first boundary range; or, In response to the train leading operation information being coupling operation information, the railway freight station data is input into the second sub-state monitoring algorithm model to obtain obstacle information and to-be-coupled vehicle information within the second boundary range that affect the train's execution of the coupling operation.

[0073] The condition monitoring algorithm model may include a first sub-condition monitoring algorithm model corresponding to the propulsion operation and a second sub-condition monitoring algorithm model corresponding to the coupling operation.

[0074] Since the train leading operation information can include the advancing operation information and the coupling operation information. The advancing operation requires the leading device to monitor the target information to be detected (also called obstacle information, such as people, cars, etc.) in front of the train that affects the travel of the train in real time, and to feed back the category information and distance information of the obstacles to the vehicle control host in real time. In addition to the above information, the coupling operation also requires the identification of the vehicle to be coupled, and sends the remaining distance between the end car of the train and the vehicle to be coupled to the vehicle control host in real time. Therefore, the advancing operation and the coupling operation have different targets to be processed according to the different operation purposes. Therefore, in some examples, the artificial intelligence visual model and the train leading operation information can be trained according to the advancing operation information and the railway freight station data to obtain the first sub-state monitoring algorithm model. In some examples, the artificial intelligence visual model and the train leading operation information can also be trained according to the coupling operation information and the railway freight station data to obtain the second sub-state monitoring algorithm model.

[0075] In some embodiments, in response to the train leading operation information being push operation information, inputting the railway freight station data into the first sub-state monitoring algorithm model to obtain obstacle information within the first boundary range that affects the train from performing the push operation may include: In response to the train leading operation information being the advancing operation information, the railway freight station data is input into the first sub-state monitoring algorithm model, so as to perform the following steps based on the first sub-state monitoring algorithm model: Extract track information and all target information from railway freight station data; determining a first boundary based on the track information; According to the first boundary and all target information, obstacle information within the first boundary that affects the train's propulsion operation is determined.

[0076] When the train leading operation information is the advancing operation information, the railway freight station data is input into the first sub-state monitoring algorithm model. In the first sub-state monitoring algorithm model, the track detection results and all target information ahead of the train are first obtained according to the railway freight station data. The target information includes the type and distance of the target. Then, the first boundary is determined according to the track detection results. The edge of the track can be used as the first boundary, or the boundary at a certain distance from the edge of the track can be used as the first boundary. All targets within the first boundary range are regarded as obstacles that affect the travel of the train.

[0077] In some embodiments, in response to the train leading operation information being coupling operation information, the railway freight station data is input into the second sub-state monitoring algorithm model to obtain obstacle information and to-be-coupled vehicle information within the second boundary range that affect the train from executing the coupling operation, including: In response to the train leading operation information being the coupling operation information, the railway freight station data is input into the second sub-state monitoring algorithm model, so as to perform the following steps based on the second sub-state monitoring algorithm model: Extract track information and all target information from railway freight station data respectively; determining a second boundary based on the track information; According to the second boundary and all target information, obstacle information and information of vehicles to be coupled that affect the coupling operation of the train within the second boundary are determined.

[0078] In the case where the train leading operation information is the coupling operation information, the railway freight station data is input into the second sub-state monitoring algorithm model. In the first sub-state monitoring algorithm model, the track detection results and all target information ahead of the train are first obtained based on the railway freight station data. The target information includes the type and distance of the target. Then, the second boundary is determined according to the track detection results. The edge of the track can be used as the second boundary, or the boundary at a certain distance from the edge of the track can be used as the second boundary. All targets within the second boundary range are regarded as obstacles that affect driving. In addition, based on the track detection results, the current track needs to be extended to extract the vehicle information to be detected that appears in the current track. The vehicle information to be detected can include the vehicle type and the distance between the vehicle and the leading device.

[0079] In some examples, it is also possible to check whether the vehicle to be detected that appears in the current track is in a preconfigured vehicle category library. If the vehicle to be detected is in the vehicle category library, the verification is normal. If the vehicle to be detected is in the vehicle category library, the verification is abnormal, and the abnormal information is fed back to the vehicle control host.

[0080] In some embodiments, in the present invention, Figure 3 In the illustrated embodiment, step S300 may be performed before step S301.

[0081] S300: Collect initial railway freight station data, and input the initial railway freight station data into a perspective correction model to obtain a corrected data collection perspective.

[0082] The perspective correction module is obtained by training the artificial intelligence perspective model based on the track positive sample data and the track negative sample data. The track positive sample data and the track negative sample data are corresponding. The track positive sample data can be data that meets the preset conditions, and the track negative sample data can be data that does not meet the preset conditions. The preset conditions are set manually.

[0083] The initial railway freight station data can be understood as the first N frames of data collected by the sensor unit after the sensor unit of the leader is started, where N≥1 and N is manually set.

[0084] The initial railway freight station data has an initial railway freight station data acquisition perspective, and the initial railway freight station data acquisition perspective can be understood as the initial coordinate system of the railway freight station environment image in the initial railway freight station data. In some examples, the perspective correction model can extract the track image where the train is located from the railway freight station environment image under the initial coordinate system, and judge whether the initial coordinate system needs to be corrected by whether the track axis of the track image where the train is located coincides with the axis of the railway freight station environment image along the track extension direction. However, it is not limited to this. In some examples, it is also possible to judge whether the initial coordinate system needs to be corrected by whether the distance between the track axis of the track image where the train is located and the axis of the railway freight station environment image along the track extension direction is within a threshold range.

[0085] Through the calibrated data collection perspective, the train leader can collect railway freight station data with the calibrated data collection perspective, so that the train leader can collect data with a precise perspective no matter where it is, thereby ensuring the stability and reliability of subsequent operations and improving the accuracy of train driving decision recommendations.

[0086] Figure 4 Flow chart of a visual correction method provided by an embodiment of the present invention. Figure 4 As shown, step S300 can be performed based on the perspective correction model, and specifically includes S401 to S403.

[0087] S401. Extract track information from initial railway freight station data. The track information may include the position and direction of the track.

[0088] S402: Calculate a correction matrix according to the orbit information.

[0089] S403: multiplying the correction matrix by the coordinate system of the initial railway freight station data to obtain a corrected coordinate system to adjust the railway freight station data collection viewing angle.

[0090] By extracting the position and direction of the track based on the initial railway freight station data, a correction matrix can be calculated and the correction matrix can be multiplied by the coordinate system of the initial railway freight station data to obtain the corrected coordinate system, thereby adjusting the viewing angle of the initial railway freight station data collection, thereby ensuring the stability and reliability of subsequent operations and improving the accuracy of train driving decision recommendations.

[0091] Figure 5 A flow chart of a train leading method provided by an embodiment of the present invention.

[0092] In some embodiments, Figure 5 The train leading method shown in the figure is executed by the train leading device shown in Figure 1 of the present invention. Figure 5As shown, the train leading method may include steps S500-S510.

[0093] S500: After the vehicle leader is started, data collection viewing angle correction is first performed. After the vehicle leader is started, it is in a non-operating state.

[0094] S501. Collect railway freight station data in real time with a corrected viewing angle and cache the collected railway freight station data.

[0095] S502: Upload cached data to the cloud server.

[0096] S503, receiving the operation status instruction sent by the control host, and configuring the operation status of the vehicle leader according to the operation status instruction, wherein the operation status instruction includes the operation information of the vehicle leader, and the operation information includes the operation state or the non-operation state.

[0097] S504, monitoring the operating state of the vehicle leader; in response to the operating state of the vehicle leader being in an operating state, executing step S505; in response to the operating state of the vehicle leader being in a non-operating state, executing step S502.

[0098] S505, receiving a train control instruction, and configuring the operation mode of the train leader according to the train control instruction. The train control instruction may include train leader operation information, and the train leader operation information may include push operation information or coupling operation information.

[0099] S506, determining whether the operation mode of the leader is a push operation or a coupling operation, and in response to the leader being in a push operation, executing step S507; in response to the leader being in a coupling operation, executing step S508.

[0100] S507: When the operation mode of the leader is a push operation, output obstacle information according to the state monitoring algorithm model and cache the obstacle information.

[0101] S508: When the operation mode of the vehicle leader is coupling operation, output obstacle information and information of the vehicle to be coupled according to the state monitoring algorithm model, and cache the obstacle information and information of the vehicle to be coupled.

[0102] S509. The obstacle information obtained in step S507 or the obstacle information obtained in step S508 is input into the behavior prediction algorithm model. The behavior prediction algorithm model performs vehicle behavior prediction analysis based on the received information, outputs the best driving decision suggestion (i.e., the train driving decision suggestion) and caches the train driving decision suggestion.

[0103] S510, sending the train driving decision suggestion to the train control host.

[0104] On the other hand, an embodiment of the present invention further provides a train leading system. Figure 6 The structure diagram of a train leading system provided by an embodiment of the present invention is shown in FIG. Figure 6 As shown, the train leading system 60 of the embodiment of the present invention includes the present invention as shown in FIG. Figure 1b The vehicle leader 10 is shown.

[0105] In some embodiments, Figure 6 As shown, the train leading system also includes: a train control host 61 and a server 62. The train control host 61 is configured to send train control instructions and train travel information to the train leading device; the server 62 is configured to train the artificial intelligence visual model according to the train leading operation information and the railway freight station data to obtain the state monitoring algorithm model; the artificial intelligence prediction model is trained according to the target information to be detected and the train travel information to obtain the behavior prediction algorithm model.

[0106] The train leader 10 is installed at the end of the train and is equipped with a sensor unit 11, a communication unit 12 and an edge computing unit 13. It collects and processes data from the intelligent train leader operation in real time and deploys AI algorithms for intelligent status analysis.

[0107] The train control host 61, as a third-party interface module, can communicate directly with the train leader. The train control host can send instructions to the train leader, such as train control instructions, operating status instructions, etc. The train control host can receive information sent by the train leader and control the train, for example, receive information sent by the train leader, such as train driving decision suggestions, and control the train according to the train driving decision suggestions.

[0108] Cloud Server 62 is responsible for the training, optimization and data storage of artificial intelligence models, ensuring continuous system improvement and performance enhancement. The cloud server is a training platform for artificial intelligence (AI) visual models and is also responsible for data storage. It not only supports the training and optimization of AI visual models, but also processes and manages a large amount of historical data. The service has powerful computing power and large-capacity storage, covering the entire process of data storage, preprocessing, labeling, and model training and optimization, significantly improving the performance and reliability of the system.

[0109] The AI ​​visual model involved in the embodiment of the present invention ensures the smooth progress of large-scale data sample labeling tasks through the automatic labeling tool provided by the cloud server. The cloud server also provides supercomputing support to meet the large number of parameter calculation requirements in model training. The AI ​​visual model model can use a general basic large model as a pre-training model, and use a large amount of railway freight station data to train the secondary industry large model, so that the trained model (condition monitoring algorithm model and behavior prediction algorithm model) finally has both the generalization of the basic large model and the accuracy of the industry large model. After the trained model is deployed to the edge computing unit of the leader, it can fully identify and process all relevant elements in the freight station, including obstacles and objects that need to interact, such as personnel, vehicles, tracks, and vehicles to be coupled during the coupling process. At the same time, the cloud server will continuously optimize the condition monitoring algorithm model and the behavior prediction algorithm model based on the feedback data, and continuously improve the performance and accuracy of the condition monitoring algorithm model and the behavior prediction algorithm model.

[0110] On the other hand, an embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the method described in any of the above embodiments are performed.

[0111] It should be clear that the present invention is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present invention.

[0112] The functional blocks shown in the structural block diagram described above can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present invention are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier. "Machine-readable medium" may include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROMs, flash memories, erasable ROMs (EROMs), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0113] It should also be noted that the exemplary embodiments mentioned in the present invention describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiments, or in a different order from the embodiments, or several steps can be performed simultaneously.

[0114] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0115] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A vehicle guide device, characterized in that: include: A sensor unit configured to collect railway freight station data; A communication unit configured to receive a train control instruction and train travel information sent by a train control host, wherein the train control instruction includes train leading operation information; An edge computing unit is configured to input the railway freight station data into a corresponding state monitoring algorithm model according to the train lead operation information to obtain corresponding target information to be detected; The edge computing unit is also configured to input the target information to be detected and the train travel information into a behavior prediction algorithm model to obtain train travel decision recommendations.

2. The vehicle driver according to claim 1, characterized in that: The railway freight station data includes an environmental image of the railway freight station and all target status data in the railway freight station; The sensor unit includes a camera and a radar; wherein, The camera is configured to collect environmental images of a railway freight station; The radar configuration is used to collect all target status data in the railway freight station.

3. The vehicle driver according to claim 1, characterized in that: The communication unit is further configured to send a train driving decision suggestion to the train control host; or The communication unit is also configured to send the collected railway freight station data to the server.

4. A train leading method, characterized in that: Applied to the vehicle guide device according to any one of claims 1 to 3, the method comprises: Collect railway freight station data; Receiving train control instructions and train travel information sent by the train control host, wherein the train control instructions include train leading operation information; Inputting the railway freight station data into the corresponding state monitoring algorithm model according to the train lead operation information to obtain the corresponding target information to be detected; The target information to be detected and the train travel information are input into a behavior prediction algorithm model to obtain a train travel decision suggestion.

5. The method according to claim 4, characterized in that The train leading operation information includes any one of push operation information and coupling operation information; The state monitoring algorithm model includes a first sub-state monitoring algorithm model and a second sub-state monitoring algorithm model; The method of inputting the railway freight station data into a corresponding state monitoring algorithm model according to the train leading operation information to obtain corresponding target information to be detected includes: In response to the train leading operation information being push operation information, the railway freight station data is input into a first sub-state monitoring algorithm model to obtain obstacle information that affects the train from executing the push operation within a first boundary range; or, In response to the train leading operation information being the coupling operation information, the railway freight station data is input into the second sub-state monitoring algorithm model to obtain obstacle information and to-be-coupled vehicle information within a second boundary range that affect the train's coupling operation.

6. The method according to claim 5, characterized in that In response to the train leading operation information being a push operation, the railway freight station data is input into the first sub-state monitoring algorithm model to obtain obstacle information within the first boundary range that affects the train from executing the push operation, including: In response to the train leading operation information being a push operation, the railway freight station data is input into a first sub-state monitoring algorithm model, so as to perform the following steps based on the first sub-state monitoring algorithm model: Extracting track information and all target information from the railway freight station data respectively; determining a first boundary according to the track information; Obstacles within the first boundary that affect the propulsion operation of the train are determined according to the first boundary and all the target information.

7. The method according to claim 5, characterized in that In response to the train leading operation information being the coupling operation, the railway freight station data is input into the second sub-state monitoring algorithm model to obtain obstacle information and to-be-coupled vehicle information within a second boundary range that affect the train from executing the coupling operation, including: In response to the train leading operation information being a coupling operation, the railway freight station data is input into a second sub-state monitoring algorithm model, so as to perform the following steps based on the second sub-state monitoring algorithm model: Extracting track information and all target information from the railway freight station data respectively; determining a second boundary according to the track information; According to the second boundary and all the target information, obstacles and information of vehicles to be coupled that affect the coupling operation of the train within the second boundary are determined.

8. The method according to claim 4, characterized in that Before collecting the railway freight station data, the method further includes: Collecting initial railway freight station data, and inputting the initial railway freight station data into a perspective correction model to obtain a corrected data collection perspective; The collection of railway freight station data includes: The railway freight station data is collected from the corrected data collection perspective.

9. A train leading system, characterized in that: It comprises a vehicle leader as claimed in any one of claims 1 to 3.

10. The vehicle piloting system according to claim 9, characterized in that: Also includes: A train control host, configured to send train control instructions and train travel information to the train leader; The server is configured to train the artificial intelligence visual model according to the train leading operation information and the railway freight station data to obtain a state monitoring algorithm model; and train the artificial intelligence prediction model according to the target information to be detected and the train travel information to obtain a behavior prediction algorithm model.

Images