A train carriage relative positioning system and method based on distributed sensing and edge computing

Abstract

This invention discloses a train carriage relative positioning method and system based on distributed sensing and edge computing, belonging to the field of train navigation and positioning technology. The system includes: a sensing terminal module deployed in each carriage, comprising a multi-mode satellite navigation receiver, an IMU, wheel speed sensors, and a UWB communication module; a base station establishment unit configured at the front of the train; a relative positioning solution unit, which establishes a double-difference observation equation based on RTK technology to solve for relative coordinates; a multi-sensor fusion unit, which uses extended Kalman filtering to fuse satellite, IMU, and wheel speed data, and introduces track topology constraints to output relative position and attitude; and a communication and synchronization unit, which utilizes UWB to achieve data transmission between carriages and uses satellite second pulses to achieve high-precision time synchronization. This invention eliminates reliance on trackside equipment, possesses fully distributed redundant sensing capabilities, and can maintain continuous high-precision positioning even in environments with satellite signal obstruction, achieving real-time and accurate monitoring of the relative position and attitude between carriages.

Description

Technical Field

[0001] This invention relates to the field of train navigation and positioning technology, specifically to a train carriage relative positioning system and method based on distributed sensing and edge computing. Background Technology

[0002] With the rapid development of high-speed rail and heavy-haul freight technology, the requirements for safety and automation in train operation are increasing. Accurate and continuous acquisition of train position, especially the relative positions and running attitudes between multiple carriages in a train formation, is a key technological prerequisite for ensuring safe train operation and achieving intelligent dispatching. However, existing train positioning and monitoring technologies still have the following significant technical shortcomings when facing complex operating environments and higher safety requirements: Reliant on rail infrastructure, costly, and with limited coverage: Traditional communication-based train control systems (CBTC) and various signaling systems heavily depend on trackside infrastructure such as transponders, axle counters, loop lines, or wireless access points (APs) for train positioning and tracking. These trackside devices not only involve large upfront installations and high construction costs, but also require continuous maintenance. More importantly, in complex geographical environments such as mountainous areas, sea-crossing bridges, and deserts, the construction and maintenance of trackside equipment are extremely difficult, making it hard to achieve full-range, blind-spot-free coverage of the railway line. While solutions exist that use virtual transponders to replace physical transponders to reduce costs, their positioning accuracy and anti-interference capabilities are limited by software algorithms and external conditions, making it difficult to completely replace physical equipment.

[0003] Centralized sensing architectures pose a risk of single-node failure: Most current mainstream train positioning solutions employ a centralized architecture, where a high-precision positioning unit is installed only at the locomotive at the front of the train to sense the position and status of the entire train. Under this architecture, if the sensors, communication units, or computing nodes at the locomotive fail, the entire train will immediately lose its accurate positional awareness, severely impacting operational safety. In CBTC systems, when train-to-ground wireless communication or ground-based ATP equipment fails, the train is typically forced to degrade to backup mode, significantly reducing its tracking capabilities and system efficiency. Therefore, the reliability of a single node constitutes a major hidden danger to train operation safety.

[0004] Positioning Failure in Satellite Signal Obstruction Scenarios: The Global Navigation Satellite System (GNSS) provides all-weather, global absolute position information, a crucial foundation for train autonomous positioning. However, train routes often traverse complex environments such as tunnels, mountains, and forests. In these areas, GNSS satellite signals are easily obstructed and interfered with, leading to receiver signal loss. In severely obstructed environments, train positioning systems relying solely on GNSS cannot receive satellite signals, and the positioning error of the Inertial Navigation System (INS) accumulates rapidly over time, ultimately causing severe drift or even failure of the positioning results, failing to meet the stringent requirements for positioning continuity for train safety. Even in areas where signals can be received, multipath interference and signal obstruction can generate numerous unreliable positioning solutions, seriously threatening train operation safety.

[0005] Lack of dynamic attitude monitoring methods between carriages: Existing train positioning technologies mostly focus on obtaining the absolute position of the train as a whole, but lack accurate and real-time measurement methods for the dynamic relative attitudes within the train formation, especially between adjacent carriages, such as relative bending angles, roll, and pitch angles. These dynamic attitude parameters are crucial for the safe operation of heavy-haul freight trains, providing key data support for preventing major accidents such as coupler breakage and train derailment. Currently, monitoring the status of couplers, buffer systems, or carriage rollover mainly relies on installing dedicated sensors on specific mechanical structures such as couplers or bogies. These sensors have limited monitoring range, are complex to install and maintain, and are difficult to implement to form a comprehensive, multi-dimensional dynamic attitude perception capability covering the entire train.

[0006] Therefore, how to provide a train positioning solution that does not rely on large-scale trackside infrastructure, has fully distributed redundant sensing capabilities, can maintain high accuracy and continuous positioning even in harsh environments such as satellite signal blockage, and can monitor the dynamic relative position and attitude between carriages in real time is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0007] This invention aims to provide a train carriage relative positioning method and system based on distributed sensing and edge computing to solve the problems of existing technologies such as reliance on trackside infrastructure, risk of single-node failure, loss of accuracy under obstructed environments, and lack of dynamic attitude monitoring between carriages, so as to achieve high-precision and high-reliability real-time perception of the relative position and attitude of the entire train.

[0008] The present invention provides a train carriage relative positioning system based on distributed sensing and edge computing, comprising: The sensing terminal module is deployed in each carriage of the train and includes a multi-mode satellite navigation receiver, an inertial measurement unit (IMU), wheel speed sensors, and a UWB communication module. The base station establishment unit is configured in the locomotive carriage and is used to set the locomotive as a differential base station and obtain the base station coordinates; The relative positioning calculation unit is deployed at the edge computing node of the train head. It is used to establish the double difference observation equation between each carriage and the base station based on carrier phase differential RTK technology, and to calculate the relative coordinates of each carriage relative to the base station. The multi-sensor fusion unit is used to fuse the relative coordinates, IMU measurements and wheel speed and mileage data using the extended Kalman filter algorithm, and introduce track topology constraints to output the relative position and attitude information of each carriage. The communication and synchronization unit is used to transmit observation data and status information between carriages through the UWB communication module, and to achieve time synchronization of the entire train using the satellite pulse-per-second (PPS) signal.

[0009] Furthermore, in the relative positioning solution unit, the double-difference observation equation is constructed by building a single-difference observation model between each carriage and the base station, and selecting a reference satellite for inter-satellite difference construction; the relative positioning solution unit is also used to fix the integer ambiguity of the double-difference observation values ​​using the LAMBDA algorithm to obtain a high-precision relative positioning solution.

[0010] Furthermore, in the multi-sensor fusion unit, the state vector of the extended Kalman filter includes three-dimensional position error, three-dimensional velocity error, three-dimensional attitude error, accelerometer device error, and gyroscope device error; the track topology constraint condition is to restrict the motion of the carriage to the longitudinal direction along the track.

[0011] Furthermore, it also includes a status monitoring and early warning unit, which is used to calculate the longitudinal distance and lateral offset between adjacent carriages in real time, and trigger an early warning when the relative distance or attitude change exceeds a preset safety threshold.

[0012] Furthermore, the time synchronization accuracy achieved by the communication and synchronization unit through the satellite pulse-per-second (PPS) signal is better than 10 nanoseconds.

[0013] The present invention also provides a method for relative positioning of train carriages based on the above-described system, comprising the following steps: S1. Deploy sensing terminal modules in each carriage, using the front carriage as the reference station, and use satellite pulse 1PPS signal to achieve high-precision time synchronization of the entire train; S2. Each carriage collects its own satellite observation data, inertial measurement data, and wheel speed data, and transmits the data to the edge computing node at the front of the train via the UWB communication module; S3. At the edge of the train head, calculate the carrier phase double difference observation equation between each carriage and the base station based on RTK technology, fix the integer ambiguity, and solve the relative coordinates of each carriage relative to the base station. S4. The extended Kalman filter algorithm is adopted to fuse the relative coordinates, IMU measurements and wheel speed and mileage data, and introduce track topology constraints to output high-precision relative position and attitude information of each carriage.

[0014] Furthermore, in step S3, the LAMBDA algorithm is used to quickly fix the integer ambiguity of the double-difference observations in order to obtain centimeter-level relative positioning accuracy.

[0015] Furthermore, in step S4, when the satellite signal is lost, the system automatically switches to a trajectory estimation mode dominated by the IMU and wheel speed sensors, and continuously outputs positioning information in conjunction with the orbital topology constraints.

[0016] Furthermore, it also includes step S5: calculating in real time the longitudinal distance and lateral offset between adjacent carriages, as well as the relative bending angle, pitch angle and roll angle between carriages, and triggering a safety warning when any parameter exceeds a preset safety threshold.

[0017] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the train carriage relative positioning method described above.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) This invention achieves fully autonomous relative positioning capability based entirely on the train itself by deploying independent sensing terminals in each carriage and using multi-mode satellite navigation receivers and the head reference station for RTK differential positioning. The entire positioning process does not rely on infrastructure such as transponders, loop lines or wireless access points along the track, reducing the initial construction investment and subsequent maintenance workload of railway lines. It is especially suitable for complex geographical environments such as mountainous areas and cross-sea bridges where it is difficult to deploy trackside equipment. (2) The present invention distributes sensing, communication and computing capabilities to each carriage to form a redundant sensing network. Even if the sensors or communication nodes of individual carriages fail, the remaining carriages can still maintain independent positioning capabilities and compensate through information exchange between adjacent carriages. This avoids the safety hazard of the entire train losing position perception due to the failure of a single node at the locomotive in the traditional solution, and improves the overall robustness and safety of train operation. (3) This invention uses a multi-sensor tightly coupled fusion strategy to deeply fuse GNSS, IMU, wheel speed sensor data with track topology constraints. When the train enters areas where satellite signals are blocked, such as tunnels or forest areas, causing GNSS to lose lock, the system automatically switches to a trajectory estimation mode dominated by IMU and wheel speed sensors. It also effectively suppresses the integral drift of the inertial navigation system by utilizing the physical constraint that "the carriage only moves longitudinally along the track", ensuring that high-precision continuous positioning output can still be maintained under long-term signal interruption, thus avoiding the driving risks caused by the failure of a single satellite navigation system. (4) This invention not only focuses on the absolute position of the train, but also calculates the relative position increment and attitude change between adjacent carriages in real time through a distributed sensing and edge computing architecture. The system can accurately output key parameters such as longitudinal distance, lateral offset, relative angle, pitch angle and roll angle between carriages, providing accurate and real-time data support for advanced safety functions such as anti-coupler breakage and anti-derailment of heavy-haul trains, filling the gap in the existing technology for dynamic attitude monitoring within train formations; (5) By combining carrier phase differential RTK technology, LAMBDA ambiguity fast fixing algorithm, and extended Kalman filter multi-sensor fusion, this invention achieves centimeter-level accuracy (better than 0.3 meters) for relative position measurement between carriages and better than 0.1° for relative attitude measurement in normal open environments. Simultaneously, a high-precision time synchronization network (synchronization accuracy better than 10 nanoseconds) built based on UWB ultra-wideband communication and satellite pulse-per-second (1PPS) ensures the real-time performance and consistency of differential observation data, providing a reliable data foundation for advanced applications such as precise train stopping, train formation integrity monitoring, and intelligent operation and maintenance. Attached Figure Description

[0019] Figure 1 This is a block diagram showing the connection relationship of the train carriage relative positioning system of the present invention.

[0020] Figure 2 This is a block diagram of the overall architecture of the train carriage relative positioning system of the present invention.

[0021] Figure 3 This is a logic block diagram of the fusion algorithm used in the multi-sensor fusion unit of this invention.

[0022] Figure 4 This is a flowchart of the relative positioning solution unit of the present invention performing satellite differential relative positioning. Detailed Implementation

[0023] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention. Example

[0024] See Figure 1 and Figure 2 This embodiment provides a train carriage relative positioning system based on distributed sensing and edge computing, which can be applied to high-speed trains or heavy-haul freight trains.

[0025] Each carriage of the train (including the locomotive carriage and all subsequent carriages) is equipped with a sensing terminal module. The hardware configuration of each sensing terminal module is as follows: Multi-mode satellite navigation receiver: Supports receiving signals from multiple satellite navigation systems such as BeiDou and GPS, and has the ability to output raw observation values ​​such as carrier phase and pseudorange. The receiver is connected to a multi-mode satellite navigation antenna mounted in an open area on the vehicle roof via a low-loss coaxial cable to ensure the quality of satellite signal reception.

[0026] Inertial Measurement Unit (IMU): Internally integrates a three-axis gyroscope and a three-axis accelerometer for real-time measurement of the carriage's attitude angle changes and acceleration. The IMU is rigidly mounted to the carriage's underframe to accurately sense changes in the carriage's attitude.

[0027] Wheel speed sensor: Installed at the end of the wheel axle of the bogie, it is used to measure the rotational speed of the wheel in real time and convert it into the travel speed and cumulative mileage of the car, and is used as an odometer.

[0028] UWB communication module: Employs ultra-wideband wireless communication technology and connects to a UWB communication antenna to achieve low-latency transmission of raw observation data and status information between carriages.

[0029] Edge computing nodes: One edge computing node is deployed in the electrical cabinet of each carriage to preprocess sensor data, timestamp it, and execute local computation algorithms. All sensor data is aggregated to the edge computing node via the carriage's internal data bus.

[0030] In addition to the standard sensing terminal modules installed in the front carriage, its edge computing node is also configured as the master control node for the entire trainset. The following functional units are further deployed on this master control node (see...). Figure 2 ): Base station establishment unit: The high-precision position information of the locomotive and carriage is set as the differential base station coordinates within the train group, and the base station coordinates and necessary raw GNSS observation data are broadcast to other carriages through the UWB communication module.

[0031] Relative positioning calculation unit: Receives and processes GNSS carrier phase and pseudorange observations reported by the sensing terminal modules of each carriage, and performs RTK relative positioning calculation.

[0032] Multi-sensor fusion unit: It collects RTK calculation results, IMU measurement values ​​of each compartment, wheel speed and mileage data, executes multi-source information fusion algorithm, and outputs the final relative position and attitude information.

[0033] All GNSS receivers within the sensing terminal modules of each carriage are equipped with satellite pulse-per-second (1PPS) output functionality. The system utilizes this 1PPS signal to achieve high-precision time synchronization among all sensing terminal modules across the train. All data exchange between carriages, including base station observation data, raw data from sensors in each carriage, and intermediate processing results, is transmitted through a low-latency wireless communication network comprised of the aforementioned UWB communication modules. Example

[0034] Based on the system hardware of Embodiment 1 above, the specific implementation process of the train carriage relative positioning method provided by the present invention is as follows, and can be found in [reference needed]. Figure 4 The flowchart shown is for satellite differential relative positioning calculation.

[0035] Step S1: System power-on and initialization After the train is powered on, the sensing terminal modules in each carriage perform power-on self-tests. Each GNSS receiver begins satellite acquisition, and the IMU completes initial alignment (determining the initial attitude). The head control node initiates a full-network time synchronization process using the satellite second pulse signal to ensure that the clock synchronization accuracy of all sensing terminal modules on the entire train meets the requirements of carrier phase differential positioning.

[0036] Step S2: Base station setup and observation data acquisition The reference station establishment unit in the locomotive carriage marks itself as a differential reference station. Its multi-mode satellite navigation receiver continuously outputs raw satellite observation data, including pseudorange, carrier phase observations, and ephemeris data.

[0037] Meanwhile, each subsequent carriage serves as a mobile station, with its sensing terminal module independently receiving satellite signals and outputting raw satellite observation data of the same type as the base station. In addition, each carriage's IMU (Inertial Measurement Unit) outputs angular velocity and acceleration data, and the wheel speed sensors output wheel speed pulse counts.

[0038] The edge computing nodes in each carriage add local timestamps to the collected IMU data and wheel speed data, and encapsulate them together with GNSS observation data into data frames via the UWB communication module before sending them to the relative positioning calculation unit of the main control node at the front of the train. The observation data from the base station is also collected by the main control node itself.

[0039] Step S3: RTK Relative Positioning Solution After receiving data from all carriages, the relative positioning calculation unit of the head unit performs relative positioning calculations based on real-time kinematics (RTK) technology. The specific process is as follows: Data preprocessing: Cycle slip detection and repair are performed on the carrier phase observations reported by each carriage and base station, and gross errors are removed.

[0040] A double-difference observation equation is constructed: First, the carrier phase observations of the base station and rover station about the same satellite are subtracted to obtain an inter-station single-difference model, eliminating satellite clock bias. Then, a reference satellite is selected, and the single-difference observations of other satellites are subtracted again to obtain an inter-station-inter-satellite double-difference observation model. Through double-difference processing, common errors such as receiver clock bias, ionospheric delay, and tropospheric delay are effectively eliminated or significantly reduced.

[0041] Integer ambiguity fixation: The LAMBDA algorithm is used to search for and fix double-difference integer ambiguities. Once the ambiguity is successfully fixed, a fixed ambiguity solution is obtained.

[0042] Baseline vector calculation: Substitute the fixed integer ambiguity into the double-difference observation equation to calculate the three-dimensional baseline vector of the rover station relative to the base station.

[0043] Coordinate transformation: Using the coordinates of the locomotive's reference station as the origin, the calculated three-dimensional baseline vector is transformed to the Northeast-Eastern-Upper-Southern (ENU) local tangent plane coordinate system to obtain the precise position increment of each car relative to the locomotive. This is the core expression of the relative position between the cars.

[0044] Step S4: Multi-sensor tightly coupled fusion positioning The multi-sensor fusion unit receives RTK relative position results from the relative positioning solution unit, as well as IMU and wheel speed mileage data from each vehicle's sensing terminal module. It uses the extended Kalman filter (EKF) algorithm for tight coupling fusion to further improve the continuity, smoothness, and robustness of the positioning results.

[0045] Integration process such as Figure 3 As shown, it specifically includes: System initial alignment: Initial attitude determination is completed using IMU data.

[0046] Normal mode (when satellite signal is valid): When the satellite signal is valid and the RTK solution is a fixed solution, the RTK relative position result is used as the measurement input for EKF to construct the measurement equation. Attitude, velocity, and position are recursively calculated using IMU data to obtain the state prediction value. Within the EKF framework, the Kalman gain is calculated, and the predicted value is updated and corrected using the measurement value to obtain the optimal state estimate. Simultaneously, the estimated IMU bias error is fed back to the IMU raw data for compensation.

[0047] Obstruction Mode (Satellite Signal Invalid): When the train enters environments with obstructed satellite signals, such as tunnels or forest areas, causing a loss of satellite signal lock, the system automatically switches to obstruction mode. At this time, RTK measurement updates are interrupted. The fusion unit relies solely on IMU data for attitude and velocity extrapolation, and combines this with mileage increments provided by wheel speed sensors for position updates, thus entering a pure inertial / track extrapolation mode.

[0048] Introducing Track Topology Constraints: To suppress the divergence of position errors in pure inertial navigation mode, this step introduces track topology constraints. These constraints are based on the physical fact that "train carriages can only move along the longitudinal direction of the track," limiting the lateral and vertical movements of the carriages to the allowable range of the track. By correcting the state prediction values ​​through these constraints, IMU integral drift can be effectively suppressed, maintaining the continuity of positioning output during periods of satellite signal loss.

[0049] Output fusion results: Regardless of whether it is in normal mode or occlusion mode, the multi-sensor fusion unit outputs smooth and continuous relative position, speed and attitude information of each carriage (including relative angle, pitch angle and roll angle).

[0050] Step S5: Status Monitoring and Safety Early Warning The condition monitoring and early warning unit receives precise condition data for each carriage from the multi-sensor fusion unit and performs the following real-time analysis: Car spacing monitoring: Calculates the spatial distance between adjacent cars. If an abnormal sudden change in the spacing between adjacent cars is detected, exceeding a preset safety threshold, it is determined that there is a safety risk of coupler breakage or train separation.

[0051] Carriage relative attitude monitoring: Continuously monitors the relative angle of yaw, pitch angle difference, and roll angle between adjacent carriages. If any attitude angle exceeds the safe range allowed by the track design, it is determined that there is a risk of derailment or rollover.

[0052] Warning Trigger: Once any of the above monitoring items exceeds the preset safety threshold, the status monitoring and warning unit immediately generates an alarm message and sends it to the train control system, triggering the corresponding emergency braking or alarm prompt.

[0053] In summary, this invention, through the aforementioned distributed hardware architecture and multi-level fusion positioning method, successfully constructs a train car relative positioning system that does not rely on trackside equipment, has full-car perception redundancy, and can maintain high-precision positioning continuity even under satellite signal blockage. This system organically combines high-precision RTK relative positioning, tight-coupled fusion of multiple sensors, track topology constraints, and inter-car UWB low-latency communication, achieving high-precision monitoring of relative position and attitude between cars. This provides a reliable data foundation for advanced safety functions of heavy-haul trains, such as preventing coupler breakage and derailment.

[0054] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the inventive concept of the present invention, and all such modifications or additions should fall within the protection scope of the present invention.

Claims

1. A train carriage relative positioning system based on distributed sensing and edge computing, characterized in that, include: The sensing terminal module is deployed in each carriage of the train and includes a multi-mode satellite navigation receiver, an inertial measurement unit (IMU), wheel speed sensors, and a UWB communication module. The base station establishment unit is configured in the locomotive carriage and is used to set the locomotive as a differential base station and obtain the base station coordinates; The relative positioning calculation unit is deployed at the edge computing node of the train head. It is used to establish the double difference observation equation between each carriage and the base station based on carrier phase differential RTK technology, and to calculate the relative coordinates of each carriage relative to the base station. The multi-sensor fusion unit is used to fuse the relative coordinates, IMU measurements and wheel speed and mileage data using the extended Kalman filter algorithm, and introduce track topology constraints to output the relative position and attitude information of each carriage. The communication and synchronization unit is used to transmit observation data and status information between carriages through the UWB communication module, and to achieve time synchronization of the entire train using the satellite pulse-per-second (PPS) signal.

2. The system according to claim 1, characterized in that, In the relative positioning solution unit, the double-difference observation equation is constructed by building a single-difference observation model between each carriage and the base station, and selecting a reference satellite for inter-satellite difference construction; the relative positioning solution unit is also used to fix the integer ambiguity of the double-difference observation values ​​using the LAMBDA algorithm to obtain a high-precision relative positioning solution.

3. The system according to claim 1, characterized in that, In the multi-sensor fusion unit, the state vector of the extended Kalman filter includes three-dimensional position error, three-dimensional velocity error, three-dimensional attitude error, accelerometer device error, and gyroscope device error; the track topology constraint condition is to restrict the motion of the carriage to the longitudinal direction along the track.

4. The system according to claim 1, characterized in that, It also includes a status monitoring and early warning unit, which is used to calculate the longitudinal distance and lateral offset between adjacent carriages in real time, and trigger an early warning when the relative distance or attitude change exceeds a preset safety threshold.

5. The system according to claim 1, characterized in that, The time synchronization accuracy achieved by the communication and synchronization unit through the satellite pulse-per-second (PPS) signal is better than 10 nanoseconds.

6. A method for relative positioning of train carriages based on the system described in any one of claims 1 to 5, characterized in that, Includes the following steps: S1. Deploy sensing terminal modules in each carriage, using the front carriage as the reference station, and use satellite pulse 1PPS signal to achieve high-precision time synchronization of the entire train; S2. Each carriage collects its own satellite observation data, inertial measurement data, and wheel speed data, and transmits the data to the edge computing node at the front of the train via the UWB communication module; S3. At the edge of the train head, calculate the carrier phase double difference observation equation between each carriage and the base station based on RTK technology, fix the integer ambiguity, and solve the relative coordinates of each carriage relative to the base station. S4. The extended Kalman filter algorithm is adopted to fuse the relative coordinates, IMU measurements and wheel speed and mileage data, and introduce track topology constraints to output high-precision relative position and attitude information of each carriage.

7. The method according to claim 6, characterized in that, In step S3, the LAMBDA algorithm is used to quickly fix the integer ambiguity of the double-difference observations in order to obtain centimeter-level relative positioning accuracy.

8. The method according to claim 6, characterized in that, In step S4, when the satellite signal is lost, the system automatically switches to a trajectory estimation mode dominated by the IMU and wheel speed sensor, and continuously outputs positioning information in combination with the orbital topology constraints.

9. The method according to claim 6, characterized in that, It also includes step S5: calculating in real time the longitudinal distance and lateral offset between adjacent carriages, as well as the relative bending angle, pitch angle and roll angle between carriages. When any parameter is detected to exceed the preset safety threshold, a safety warning is triggered.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the train carriage relative positioning method as described in any one of claims 6 to 9.

Images