Virtual coupling marshalling self-discipline control method and system for heavy-load combined train based on target distance

Abstract

According to the virtual coupling marshalling self-discipline control method and system for the long and large heavy-load combined train based on the target distance, various heterogeneous sensors are deployed on a locomotive and used for all-weather active distance measurement; fusing multi-sensor data, and outputting a real-time vehicle distance; calculating and predicting a target value of the safe distance between the marshalling locomotive and the tail of the front locomotive; a driver controller position instruction value equivalent to traction braking force is calculated and connected into a locomotive control system, and a unit train control closed loop with the distance as the target is formed; closed-loop control taking distance as a target is fused with a train control instruction transmitted by a main control locomotive through LOCOTROL, so that a long and large heavy-load combined train is equivalently decomposed into independent automatic driving unit trains for operation, and the longitudinal impact force of the long and large heavy-load train is eliminated; and the operation and control modes of the long and large heavy haul train and the virtual marshalling train of the heavy haul railway are overturned.

Description

Technical Field

[0001] The present invention relates to the fields of heavy-haul railway locomotive and train control, automatic detection, artificial intelligence, etc., and particularly relates to a cooperative control method and system for virtual coupled formation of heavy-haul combined trains. Background Art

[0002] Increasing the number of train formations and synchronously pulling multiple locomotives are common methods to improve the transportation capacity of heavy-haul railways. For a heavy-haul combined train composed of multiple trains connected by couplers, synchronous control of the locomotives of each unit train in the formation is a prerequisite for ensuring the safe and stable operation of the train. The master locomotive transmits control commands to the slave locomotives in real time, and multiple locomotives perform operations such as starting, accelerating, decelerating, and braking simultaneously to achieve synchronous control among the locomotives of the unit trains within the formation.

[0003] If the actions of the locomotives of each unit train are not synchronous, it will cause squeezing or pulling phenomena between the vehicles of the heavy-haul train, affecting railway transportation safety. Currently, there are two technologies to solve the problem of synchronous control of locomotives, namely the locomotive wireless synchronous control technology Locotrol and the electro-pneumatic braking technology ECP.

[0004] The longer the train length, the longer the transmission time of the braking control command, and the greater the longitudinal impulse between the vehicles. The technical requirements for the driver to operate the locomotive are high, and improper operation may cause serious accidents.

[0005] The introduction of virtual formation technology in heavy-haul railways breaks through the location-based train tracking concept of fixed block, quasi-moving block, and moving block, and adopts a more advanced train tracking concept to improve the transportation capacity of heavy-haul railways. In virtual formation, when heavy-haul trains are tracking and running, it no longer assumes that the leading train is stationary, but obtains the position, speed and other state information of the leading train through vehicle-to-vehicle communication. The trailing train tracks the speed and position of the leading train to achieve relative block, greatly shortening the train tracking interval and improving the transportation efficiency of heavy-haul railways.

[0006] Virtual formation technology usually realizes the dynamic cooperative operation of each unit train through wireless communication and intelligent control, breaks through the limitations of physical formation, and significantly improves railway transportation capacity and operation efficiency. This technology has become a research hotspot in the fields of heavy-haul railways and urban rail transit, and has significant potential especially in solving the control problems of long-formation trains and improving transportation density.

[0007] Current research on the operation control of heavy-haul railway virtual formation trains is based on vehicle-to-ground and vehicle-to-vehicle wireless communication, and adopts a master-slave synchronous control for the formation train composed of multiple trains. It is a complex system that uses dynamic cooperative control technology and train automatic driving technology within the formation. The vehicle-to-vehicle control tracking based on the speed-distance curve has a smaller tracking interval compared to moving block. Summary of the Invention

[0008] To solve the problems newly brought about by long train groups composed of multiple trains, such as large coupler impact force and serious wear of tracks with small turning radii, on the one hand, the present invention proposes a self-disciplined control method for virtual coupled formation of heavy-haul combined trains based on target distance. The method includes:

[0009] Step S1: Deploy a variety of heterogeneous sensors on the locomotive for all-weather active ranging;

[0010] Step S2: Perform multi-sensor data fusion to output the real-time train distance;

[0011] Step S3: Calculate the braking curve of this unit train and predict the target value of the safe distance between the formation locomotive and the rear end of the preceding train;

[0012] Step S4: Calculate the position command value of the controller equivalent to the traction and braking force, and access the locomotive control system to form a train closed-loop control system with distance as the target;

[0013] Step S5: Integrate / redundantly combine the closed-loop control with distance as the target and the train control command transmitted by the master locomotive through vehicle-to-vehicle communication to ensure the safe and stable operation of the virtual coupled formation long heavy-haul train.

[0014] Further, in the step S1, the sensors include a binocular camera, a millimeter-wave radar, a lidar, and a Beidou positioning module; the binocular camera is used for visual detection, the millimeter-wave radar and the lidar are used for active ranging, and the Beidou positioning module is used to supplement the ranging of the millimeter-wave radar and the lidar.

[0015] Further, the step S2 specifically includes fusing multi-sensor data through the Kalman filter or particle filter algorithm to eliminate the error of a single sensor; realizing spatio-temporal consistency through the interpolation algorithm to obtain real-time train distance data;

[0016] Inputting multi-modal sensor data into a trained end-to-end neural network model to output real-time train distance data, and the real-time train distance data includes the acceleration, deceleration, and emergency braking states of the preceding train.

[0017] Further, the neural network model is of the CNN-LSTM or Transformer architecture.

[0018] Further, the step S3 specifically includes calculating the braking curve of this unit train based on the neural network model according to the line conditions, climate conditions, communication delay, and train operation status, dynamically predicting the minimum safe distance, and generating a distance control target value;

[0019] The neural network model is a multi-layer perceptron. The input parameters of the neural network model include line gradient, curvature, rain and snow weather, and train load, and the output is the dynamically adjusted minimum safety distance.

[0020] Further, step S4 specifically includes using the real-time vehicle distance data as the feedback quantity and the distance control target value as the given value to form a closed-loop control system with the distance between the train of this unit and the leading vehicle as the target, and correcting the following distance according to the deviation between the given value and the feedback quantity value.

[0021] According to the real-time vehicle distance deviation, calculate the position command value of the controller equivalent to the traction and braking force, and access the locomotive control system to form a control closed-loop with distance as the target.

[0022] Further, step S4 specifically includes redundantly fusing the distance closed-loop control with the control command transmitted by the master locomotive through vehicle-to-vehicle communication, enabling the unit trains within the formation to synchronously perform acceleration, deceleration, and emergency braking operations, adjusting the traction and braking force of the leading locomotive according to the given target using a self-disciplined flexible control strategy, quickly adjusting the distance between the formation locomotives and the end of the leading train, and pulling the train of this unit to run safely in formation.

[0023] Further, when vehicle-to-vehicle communication such as LOCOTROL system communication is interrupted, the closed-loop control command independently maintains the safe distance of the unit trains within the formation and resynchronizes with the master command after the communication is restored.

[0024] On the other hand, the present invention also proposes a self-disciplined control system for virtual coupled formation of heavy-haul combined trains based on target distance, including:

[0025] A multi-mode heterogeneous sensor module, deployed on the locomotive, for collecting the distance to the end of the leading train, track status, and environmental data;

[0026] An electronic control unit, including a high-performance processor, a memory, and input / output interfaces, for executing the steps of the above-mentioned self-disciplined control method for virtual coupled formation of heavy-haul trains based on target distance;

[0027] A communication module, supporting V2X communication and LOCOTROL system protocol, for transmitting commands between the master locomotive and the slave locomotives;

[0028] A locomotive control system, receiving the traction and braking force commands generated by the electronic control unit and controlling the locomotive to perform acceleration, deceleration, and braking operations.

[0029] Further, the electronic control unit is connected to the locomotive control system through a vehicle bus or a formation network and integrates a sensor signal preprocessing module for filtering, spatio-temporal alignment, and data compression.

[0030] The beneficial effects of the present invention:

[0031] (1) The locomotives of unit trains in virtual coupled train formations use autonomous flexible control of the following distance as the main control method, which realizes the equivalent decomposition of long and heavy-load combined trains into independent automatic driving unit trains, eliminates the longitudinal impulse of long and heavy-load trains, and subverts the operation and control mode of long and heavy-load trains and virtual train formations on heavy-load railways.

[0032] (2) The locomotives of unit trains in the virtual coupling train directly and actively measure the distance to the rear of the train in front, shortening the distance between trains and trains, further improving the efficiency of transportation organization. In terms of train operation control and dispatching command, it is no different from the traditional 10,000-ton combined train with a coupler.

[0033] (3) Whether the locomotives of unit trains in a virtual coupled train are directly measuring distances or controlling the distances between unit trains, they do not rely on complex means such as "train-to-train communication" and "ground-space-time" to "precisely locate" each unit train. They do not rely on ground signal systems and are not constrained or affected by external conditions such as ground facilities (such as communication base stations), which greatly simplifies the control structure of the virtual train system.

[0034] (4) The autonomous flexible control method and system of the present invention allows unit trains in a train set to be pulled by locomotives of different models and types, such as a virtual coupled train set consisting of any mix of DC locomotives, AC locomotives, etc.

[0035] (5) The autonomous flexible control method and system of the present invention does not require the train formation to adopt a specific or unified braking system or braking method, and allows the existing braking system and braking method (such as air line braking or ECP braking) of each train before formation to be retained.

[0036] (6) The autonomous flexible control method and system of the present invention has redundant functions with the LOCOTROL control of the entire train set. Even if an abnormality such as a short-term interruption of the remote communication between locomotives occurs, it will not affect the normal operation of the entire train set, and the operation safety and reliability are higher. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The present invention has the following accompanying drawings:

[0038] Figure 1 Flow chart of the method of the present invention;

[0039] Figure 2 The system structure diagram of the present invention;

[0040] Figure 3 The closed-loop control system based on target distance formed by the method of the present invention and the original locomotive control system and its working principle;

[0041] Figure 4The method and system of the present invention constitute the implementation process of autonomous control for virtual coupled formation of heavy-haul trains. Detailed implementation manners

[0042] To make the objectives, technical solutions and advantages of the present application clearer and more understandable, the following further elaborates on the present application in detail with reference to specific embodiments and the accompanying drawings.

[0043] Embodiment 1:

[0044] As Figure 1 shown, a method for autonomous control of virtual coupled formation of heavy-haul trains based on target distance specifically includes:

[0045] Step S1, deploying a variety of heterogeneous sensors on the locomotive to ensure all-weather ranging reliability;

[0046] Step S2, fusing multi-sensor data to reliably measure the real-time distance between the formation locomotive and the rear end of the leading train;

[0047] Step S3, calculating the braking curve of this unit train and predicting the target value of the safe distance between the locomotive in the formation and the rear end of the leading train;

[0048] Step S4, calculating the position command value of the controller equivalent to the traction and braking force, accessing the locomotive control system, and forming a closed-loop control of the unit train with distance as the target;

[0049] Step S5, fusing the closed-loop control with distance as the target and the train control command transmitted by the master locomotive via LOCOTROL to ensure the safe and stable operation of the long heavy-haul train with virtual coupled formation.

[0050] In step S1, a variety of sensors including binocular cameras, millimeter-wave / laser radars, etc. are deployed on the locomotive. The camera is used for visual detection, the radar is used for active ranging, and Beidou, etc. are used as supplements to ensure all-weather ranging reliability.

[0051] In step S2, the Kalman filter or particle filter algorithm is used to eliminate the error of a single sensor, and the interpolation algorithm is used to ensure spatio-temporal consistency. An end-to-end neural network model is trained, and multi-modal sensor data is input to output the real-time vehicle distance (including the acceleration / deceleration and emergency braking states of the leading vehicle).

[0052] In step S3, a neural network model is used to calculate the braking curve of this unit train according to line and climate conditions, communication and calculation delays, train status, etc., predict the minimum safe distance, and give the control target value of the distance from the leading vehicle.

[0053] In step S4, the formation of trailing locomotives takes the distance from the rear of the preceding train formation as the control target, uses the ranging value obtained by multi-sensor data fusion (Claim 3) as the feedback quantity, and uses the calculated distance control target value (Claim 4) as the given value to form a closed-loop control system for this unit train with the distance from the preceding train as the target. The following distance is corrected according to the deviation between the given value and the feedback quantity value.

[0054] In step S5, the distance closed-loop control is integrated / redundant with the train control instructions transmitted by the master locomotive via LOCOTROL to improve the response speed and reliability of each slave locomotive to synchronously execute starting acceleration, deceleration, and emergency braking. The traction and braking force of the leading locomotive are adjusted using a self-disciplined flexible control strategy according to the given target, quickly adjusting or maintaining the distance between the formation locomotives and the rear of the preceding train formation, and towing this unit train to run safely in formation.

[0055] Embodiment 2:

[0056] As Figure 2 shown, on each formation unit locomotive, a binocular camera, a radar sensor, a V2X communication module, and an ECU controller are deployed. The camera is used for medium-distance assistance in visual detection, calculates the three-dimensional coordinates of the preceding train through the parallax principle, and combines with a deep learning model to achieve target detection (can simultaneously identify the contour of the preceding train, the track, etc., and can be used for detecting the intrusion of foreign objects on the track. Infrared is used to supplement in low light at night) and distance estimation; millimeter-wave / 64-line lidar is used as the main long-distance sensor and is redundant for active ranging, with the advantage of penetrating bad weather such as rain, snow, haze, etc. Beidou, etc. are used as supplements to ensure all-weather ranging reliability.

[0057] The Kalman filter or particle filter algorithm is used to fuse the data of sensors such as vision, radar, and ultrasonic, establish a multi-sensor state space model, allocate weights, and eliminate the errors of single sensors (such as the camera being sensitive to strong light and the radar misdetecting metal obstacles). The radar timestamp (in μs) and the camera image frame (30Hz) are aligned through an interpolation algorithm to ensure spatio-temporal consistency.

[0058] The multi-modal sensor data is directly input into a trained end-to-end model (such as CNN-LSTM, Transformer), and after fusing the data, the real-time vehicle distance and the state of the preceding vehicle (including acceleration / deceleration, emergency braking) are output.

[0059] A neural network model (such as MLP) is used to consider line conditions, vehicle data, operating status, weather conditions, communication control calculation delay, etc., predict the minimum safe distance, and give the target value for controlling the distance from the preceding vehicle.

[0060] As Figures 3 - 4As shown, different from the traditional train control method, the slave locomotives in the formation take the distance from the rear of the preceding train as the target, use the ranging value fused from multi-sensor data as the feedback quantity, and use the calculated distance control target value as the given value to form a closed-loop train control system for the train unit.

[0061] In the operation control of the whole train, the distance closed-loop control is integrated / redundant with the train control instructions transmitted by the master locomotive via LOCOTROL, improving the response speed and reliability of each slave locomotive to synchronously execute starting acceleration, deceleration, and emergency braking. The traction and braking forces of the locomotives in the train unit are adjusted according to the given target by using the self-disciplined flexible control strategy, so as to dynamically adjust the distance between the slave locomotives in the formation and the rear of the preceding train, and tow the train unit to run safely in formation.

[0062] In the present invention, virtual coupling means that the train units in the formation are not connected by couplers and keep a following distance of dozens of meters or shorter; flexibility means that this following "distance" is not constant and has a certain elasticity like a spring; self-disciplined control means that the locomotive of the train unit has the right to adjust the traction and braking forces of the train unit within a certain range.

[0063] The ECU collects the image data of the camera, the active ranging data of sensors such as radar, etc., performs fusion to obtain the real-time distance between train units, calculates and predicts the target value of the safe distance, converts it into a traction and braking force instruction as the input of the locomotive control system, and forms a control closed-loop with the locomotive control system with the distance as the target to ensure that the distance from the rear of the preceding train changes elastically within a certain safe range.

[0064] The ECU can be connected to the vehicle bus (such as MVB) / network (such as ECN) of the locomotive. Structurally, the ECU is a multi-functional vehicle-mounted electronic control unit with multiple analog and digital input interfaces, bus / network interfaces, a high-performance CPU, a memory, etc. The above various sensors, V2X modules, Beidou satellite antenna signals, etc. are physically directly connected to the input interface of the ECU. First, the input signals of the sensors are preprocessed in the ECU, and data processing such as filtering and spatio-temporal alignment is performed, and then the above data is fused to obtain real-time distance data.

[0065] The embodiments of the present application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present application shall be included within the protection scope of the present application.

Claims

1. A method for autonomous control of virtual coupling marshaling of heavy-load combined trains based on target distance, characterized in that: The method comprises: Step S1: deploying a variety of heterogeneous sensors on the locomotive for all-weather active ranging; Step S2: multi-sensor data fusion, output real-time vehicle distance; Step S3: Calculate the braking curve of the unit train and predict the target value of the safe distance between the locomotive of the unit train and the tail of the preceding train in the formation; Step S4: Calculate the driver controller position command value equivalent to the traction braking force, connect it to the locomotive control system, and form a control closed loop with distance as the target; Step S5: Integrate the closed-loop control with distance as the target with the train control command transmitted from the master locomotive via train-to-train communication to ensure the safe and stable operation of the virtual coupled marshaling large-scale and heavy-load combination train.

2. A method for controlling a heavy-load combined train virtual coupling marshaling autonomously based on target distance as claimed in claim 1, characterized in that: In step S1, the sensor includes a binocular camera, a millimeter-wave radar, a laser radar and a Beidou positioning module; the binocular camera is used for visual detection, the millimeter-wave radar and the laser radar are used for active ranging, and the Beidou positioning module is used to supplement the ranging of the millimeter-wave radar and the laser radar.

3. The method for controlling a heavy-load combined train virtual coupling marshaling autonomously based on target distance according to claim 1, characterized in that: The step S2 specifically includes fusing multi-sensor data through Kalman filtering or particle filtering algorithm to eliminate single sensor errors; achieving spatiotemporal consistency through interpolation algorithm; and training an end-to-end neural network model, inputting multi-modal sensor data, and outputting real-time vehicle distance data, wherein the real-time vehicle distance data includes acceleration, deceleration, and emergency braking status information of the preceding vehicle.

4. A method for controlling a heavy-load combined train virtual coupling marshaling autonomously based on target distance as claimed in claim 3, characterized in that: The neural network model is a CNN-LSTM or Transformer architecture.

5. The method for controlling the virtual coupling marshaling of a heavy-load combined train based on target distance according to claim 1, characterized in that: The step S3 specifically includes, based on the neural network model, calculating the braking curve of the unit train according to the line conditions, climate conditions, communication delay, and train operation status, dynamically predicting the minimum safe distance and generating a distance control target value; the neural network model is a multi-layer perceptron, and the neural network model input parameters include line slope, curvature, rainy and snowy weather, and train load, and outputs a dynamically adjusted minimum safe distance.

6. The method for controlling a heavy-load combined train virtual coupling marshaling autonomously based on target distance according to claim 1, characterized in that: The step S4 specifically includes: taking the real-time vehicle distance data as feedback and the distance control target value as a given value, forming a train closed-loop control system with the distance between the unit train and the preceding vehicle as the target, and correcting the following vehicle distance according to the deviation between the given value and the feedback value; calculating the driver controller position command value equivalent to the traction braking force according to the real-time vehicle distance deviation, connecting it to the locomotive control system, and forming a control closed loop with distance as the target.

7. The method for autonomous control of heavy-load train virtual coupling marshaling based on target distance as claimed in claim 1, characterized in that: The step S4 specifically includes redundantly integrating the distance closed-loop control with the train control instructions transmitted from the master locomotive via train-to-train communication, so as to achieve synchronous acceleration, deceleration and emergency braking operations of the unit trains within the formation, and adjusting the traction braking force of the locomotive of the unit train according to the given target by adopting the autonomous flexible control strategy, so as to quickly adjust or maintain the distance between the formation locomotive and the rear of the preceding train, and to pull the unit train to run safely.

8. The method for controlling the virtual coupling marshaling of a heavy-load combined train based on target distance according to claim 7, characterized in that: When the vehicle-to-vehicle communication between the master locomotive and the slave locomotive is interrupted, the unit train locomotive independently maintains the safe distance of the unit trains in the formation and resynchronizes with the master control command after the communication is restored.

9. A heavy-load train virtual coupling marshaling autonomous control system based on target distance, characterized in that: include: Multi-mode heterogeneous sensor modules are deployed on locomotives to collect data on the distance between the front and rear trains, track status, and environment. An electronic control unit, comprising a powerful computing processor, a memory and an input / output interface, for executing the steps of a method for autonomous control of a heavy-load train virtual coupling marshaling based on a target distance as described in any one of claims 1 to 8; Communication module, supporting V2X communication and LOCOTROL system protocol, used for command transmission between the master locomotive and the slave locomotive workshop; The locomotive control system receives the traction braking force command generated by the electronic control unit and controls the locomotive to perform acceleration, deceleration and braking operations.

10. A virtual coupling marshaling autonomous control system for heavy-load combined trains based on target distance as claimed in claim 9, characterized in that: The electronic control unit is connected to the locomotive control system via a vehicle bus or a marshaling network, and is integrated with a sensor signal preprocessing module for filtering, time-space alignment and data compression.

Images