Dynamic positioning method and system for train

Abstract

The invention discloses a dynamic positioning method and system of a train, and relates to the technical field of dynamic positioning, a plurality of key areas are determined according to the identification of the geometrical morphology of a track, a data interaction system is determined based on each key area, a trackside UWB base station and a vehicle-mounted UWB label, and the dynamic positioning of the train is realized. The data combination of the train is determined according to detection of the data interaction system, and the dynamic train length of the train is determined according to identification of the data combination; determining a UWB ranging length based on the positioning data of the vehicle-mounted UWB tag of the vehicle head and the positioning data of the vehicle-mounted UWB tag of the vehicle tail; the UWB ranging precision is determined according to the difference value between the UWB ranging length and the dynamic train length of the train, meanwhile, previous error data of the train are collected, and a corresponding error management and control model is constructed according to the previous error data and the current driving scene of the train. The dynamic positioning data of the train is determined according to the error management and control model, the UWB distance measurement precision and the real-time position data of the train, and the accuracy of the dynamic positioning data of the train is improved.

Description

Technical Field

[0001] This invention relates to the technical field of dynamic positioning, and more particularly to a dynamic positioning method and system for trains. Background Technology

[0002] As the backbone system of modern urban public transportation, urban rail transit relies heavily on the accuracy of train positioning technology for its safety and operational efficiency. Currently, the main train positioning technologies rely on track circuits, beacon transponders, etc., which can affect the accuracy of train dynamic positioning due to environmental interference, rain, snow, slippery conditions, and other factors.

[0003] In existing technologies, real-time monitoring of train operation and obtaining multiple data points about the train, along with direct determination of corresponding positioning data based on the identification of these multiple data points, affects the accuracy of the train's dynamic positioning data. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a dynamic positioning method and system for trains.

[0005] This invention provides a dynamic positioning method for trains, comprising: A corresponding vehicle model database is constructed based on the vehicle basic information table, the UWB tag information table, and the vehicle dynamic parameter table; Collect track map database, and determine the track geometry of the section where the train is located based on the vehicle type database, track map database, and the initial location of the train; Based on the identification of the track geometry, multiple key areas are determined, and a data interaction system is determined based on each key area, trackside UWB base station, and vehicle-mounted UWB tag; the data interaction system includes an entity list, spatial topology, and interaction protocols and strategies. In the data interaction system, the train's data combination is determined based on the detection of the data interaction system, and the dynamic train length is determined based on the identification of the data combination; the UWB ranging length is determined based on the positioning data of the onboard UWB tag at the front of the train and the positioning data of the onboard UWB tag at the rear of the train; the UWB ranging length represents the distance between tags obtained by the UWB system through wireless signal measurement and geometric calculation. The UWB ranging accuracy is determined by the difference between the UWB ranging length and the dynamic train length. At the same time, historical error data of the train is collected, and a corresponding error control model is constructed based on this historical error data and the current train operating scenario. The dynamic positioning data of the train is determined based on this error control model, the UWB ranging accuracy, and the real-time position data of the train.

[0006] This invention provides a dynamic positioning system for trains, which is applied to the aforementioned dynamic positioning method for trains.

[0007] Compared with the prior art, the beneficial effects of the present invention are: A corresponding vehicle model database is constructed based on the vehicle basic information table, UWB tag information table, and vehicle dynamic parameter table. A track map database is collected, and the track geometry of the section where the train is located is determined based on the vehicle model database, track map database, and the initial position of the train. Based on the identification of this track geometry, multiple key areas are identified. A data interaction system is determined based on each key area, trackside UWB base station, and onboard UWB tag. The introduction of each key area and the overall consideration of each key area, trackside UWB base station, and onboard UWB tag improve the accuracy of the data interaction system.

[0008] Therefore, in the data interaction system, the train's data combination is determined based on the detection of this data interaction system, and the dynamic length of the train is determined based on the identification of this data combination; the UWB ranging length is determined based on the positioning data of the onboard UWB tags at the front and rear of the train; the UWB ranging accuracy is determined based on the difference between the UWB ranging length and the dynamic length of the train; simultaneously, historical error data of the train is collected, and a corresponding error control model is constructed based on this historical error data and the current driving scenario of the train; the dynamic positioning data of the train is determined based on this error control model, the UWB ranging accuracy, and the real-time position data of the train. The introduction of UWB ranging accuracy further controls the error control model, realizing the overall consideration of the error control model, UWB ranging accuracy, and the real-time position data of the train, thus improving the accuracy of the train's dynamic positioning data. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating the dynamic positioning method for trains in an embodiment of the present invention. Figure 2 This is a flowchart illustrating step S11 in the dynamic positioning method for trains according to an embodiment of the present invention. Figure 3 This is a flowchart illustrating step S12 in the dynamic positioning method for trains according to an embodiment of the present invention. Figure 4 This is a flowchart illustrating step S13 in the dynamic positioning method for trains according to an embodiment of the present invention. Figure 5 This is a flowchart illustrating step S14 in the dynamic positioning method for trains according to an embodiment of the present invention. Figure 6 This is a flowchart illustrating step S15 in the dynamic positioning method for trains according to an embodiment of the present invention. Figure 7 This is a schematic diagram of the structural composition of the dynamic positioning system for a train in an embodiment of the present invention; Figure 8 This is a framework diagram of the key area in S13 of the dynamic positioning method for trains in this embodiment of the invention. Detailed Implementation

[0010] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0011] Please see Figures 1 to 8 A dynamic positioning method for trains, applied to dynamic positioning scenarios; the dynamic positioning method for trains includes: Step S11: Construct the corresponding vehicle model database based on the vehicle basic information table, UWB tag information table, and vehicle dynamic parameter table; Step S12: Collect the track map database, and determine the track geometry of the section where the train is located based on the vehicle type database, the track map database, and the initial position of the train; Step S13: Based on the identification of the track geometry, determine multiple key areas, and determine the data interaction system based on each key area, trackside UWB base station and vehicle-mounted UWB tag; Step S14: In the data interaction system, the train's data combination is determined based on the detection of the data interaction system, and the dynamic length of the train is determined based on the identification of the data combination; the UWB ranging length is determined based on the positioning data of the onboard UWB tag at the front of the train and the positioning data of the onboard UWB tag at the rear of the train. Step S15: Determine the UWB ranging accuracy based on the difference between the UWB ranging length and the dynamic train length. Simultaneously, collect the train's past error data, construct a corresponding error control model based on the past error data and the train's current driving scenario, and determine the train's dynamic positioning data based on the error control model, the UWB ranging accuracy, and the train's real-time position data.

[0012] refer to Figure 2 In step S11, the specific steps are as follows: S111: Collect the train number and train database, determine the train's information space based on the matching of the train database and the train number, and determine the vehicle basic information table, UWB tag information table and vehicle dynamic parameter table based on the traversal of the train's information space. S112: Determine the first level of vehicle model content based on the vehicle basic information table and the UWB tag information table; determine the second level of vehicle model content based on the vehicle basic information table and the vehicle dynamic parameter table; and determine the corresponding vehicle model database based on the first level of vehicle model content and the second level of vehicle model content.

[0013] In the embodiments of this application, the train number and train database are collected, and the train information space is determined based on the matching of the train database and the train number. The vehicle basic information table, UWB tag information table and vehicle dynamic parameter table are determined based on the traversal of the train information space. This takes into account the overall consideration of the traversal of the train information space and ensures the accuracy of the vehicle basic information table, UWB tag information table and vehicle dynamic parameter table.

[0014] At this point, the system obtains the unique identifier of the currently operating train, i.e., the train number (e.g., Train_A_001), through the onboard interface or signaling system. The system uses this number as the primary key to query the central train database. This database is a relational database cluster or distributed storage system that stores data on all trains in the entire metro network. After a successful match, the system locks a logical data container, i.e., an information space, bound to the train number. This information space is not a single table, but a logical data view or pointer to a data set. It aggregates all static configuration data, dynamic parameters, and historical data related to the train. This process ensures that all subsequent data operations are strictly limited to the train and avoids data crosstalk.

[0015] After determining the train's information space, the system needs to perform structured parsing and extraction of the data within that space. Following a predefined data model, the system accesses the core data tables contained within the information space one by one. These core data tables include at least: a vehicle basic information table: this table provides the train's inherent physical attributes and static configuration information. This data is determined when the train leaves the factory and changes very infrequently; a UWB tag information table: this table records the detailed configuration of the UWB tags specifically installed for positioning, which is the physical basis for achieving UWB positioning; and a vehicle dynamic parameter table: this table stores data collected in real time by sensors that changes continuously with the train's operating status. This data reflects the train's current operating condition.

[0016] Specifically, the train (model M6, serial number M6-01-03) is preparing to enter the main line; after the onboard positioning unit is activated, it reads its unique serial number M6-01-03 from the Train Control System (TCMS), and this serial number is sent to the onboard computer; the computer then sends a query request to the central database: please provide all data associated with train serial number M6-01-03; the database system responds to this request, locates the record with Train_ID M6-01-03 in the data table, and returns a reference or ID pointing to the train's dedicated dataset; at this point, the train's information space is successfully determined; all subsequent data calls to the train will be based on this determined information space.

[0017] After determining the information space of the train (M6-01-03), the onboard computer begins the traversal operation: Based on the vehicle basic information table, the system queries the table and obtains the following key static data about the train: rain_ID:M6-01-03; Train_Type:M6; Design_Length:120.5 meters (this is the design length of the train under standard conditions); Elastic_Coeff:0.0008 / °C (this is the coefficient of thermal expansion of the car body material, used to calculate the length expansion and contraction of the car due to temperature changes).

[0018] The system then queries the UWB tag information table to obtain the configuration of the UWB tags installed on the train: Tag_ID: Tag_Head_A01, Train_ID: M6-01-03, Install_Position: CabA (front of the train), Height: 3.8 meters, Offset: 0.1 meters; Tag_ID: Tag_Tail_A01, Train_ID: M6-01-03, Install_Position: CabB (rear of the train), Height: 3.8 meters, Offset: 0.1 meters. This data accurately describes the three-dimensional installation position of the two UWB tags on the train, which is crucial for subsequently calculating the train position from the tag coordinates. For the vehicle dynamic parameter table, the system accesses this table in real time (or subscribes directly via the vehicle bus) to obtain the dynamic data at the current moment: Train_ID:M6-01-03; Timestamp:2023-10-27T10:30:05Z; Temperature:35.2°C; Axle_Weight:15.2 tons (current axle load, which affects the compression of the suspension system, and thus the vehicle length). The system constructs a complete data snapshot for the train, including static attributes, label configuration, and real-time operating conditions. This structured dataset provides the most direct and comprehensive raw material for constructing the vehicle model database in step S112.

[0019] In addition, Ultra-Wideband (UWB) technology has shown broad application prospects in the field of rail transit train positioning due to its advantages such as accurate positioning and resistance to multipath interference. UWB technology achieves centimeter-level ranging accuracy through nanosecond-level pulse signals. Its high resistance to multipath interference can effectively suppress track reflection interference, and it supports real-time position calculation using dynamic TDOA (Time Difference of Arrival) algorithms without relying on track infrastructure modifications.

[0020] Furthermore, the first level of vehicle model content is determined based on the vehicle basic information table and the UWB tag information table, and the second level of vehicle model content is determined based on the vehicle basic information table and the vehicle dynamic parameter table. The corresponding vehicle model database is determined based on the first level of vehicle model content and the second level of vehicle model content, which takes into account the overall consideration of the first level of vehicle model content and the second level of vehicle model content, and ensures the accuracy of the corresponding vehicle model database.

[0021] At this point, for the first layer of vehicle model content, the aim is to obtain the inherent and unchanging physical attributes and configuration information of the train. The system accurately matches the train number in the vehicle basic information table to obtain the core design parameters of the train, such as the model, design length, and material properties. The system uses the same train number as an index to query the UWB tag information table to obtain the detailed configuration of all positioning tags bound to the train. These two parts of data together constitute the static content of the train, defining its basic physical form and the layout of positioning sensors.

[0022] For the second type of vehicle content, the aim is to capture the real-time status information of the train at the current operating moment, which is a dynamic data acquisition process. The system queries the latest record in the vehicle dynamic parameter table based on the train number and the current timestamp. The data stored in this table reflects the real-time changes of the train caused by factors such as environment and load, such as car body temperature, axle load, and suspension system status. These dynamic parameters are crucial for correcting the static model and achieving accurate real-time positioning.

[0023] The system organically integrates the first (static) and second (dynamic) vehicle model information obtained in the first two steps to form a structured and integrated data set, which is the exclusive vehicle model database for the train. It not only includes the train's factory settings but also overlays real-time operating conditions, thereby constructing a digital snapshot that can comprehensively and accurately describe the current physical state of the train. This database is the direct data source for subsequent positioning algorithms to perform coordinate calculation, error compensation, and state estimation.

[0024] Specifically, for the train (Train_ID=A12345), the system performs the following query: For the vehicle basic information table, the system locates the unique record in the table based on Train_ID "A12345" and extracts the following key static data: Train_ID:A12345; Design_Length: 200.0 meters; Elastic_Coeff: 0.01; Regarding the UWB tag information table, the system also uses Train_ID "A12345" as the key to retrieve all UWB tag records installed on the train: Front tag record: Tag_ID: Tag_001, Install_Position: CabA, Height: 1.5 meters, Offset: 0.0 meters; Rear tag record: Tag_ID: Tag_002, Install_Position: CabB, Height: 1.5 meters, Offset: 0.0 meters; At this point, the first layer of train model content is determined, which includes all static specifications of the train model and the precise installation parameters of its front and rear UWB tags; At this point, the first layer of train model content is determined, which includes all static specifications of the train as the corresponding model and the precise installation parameters of its front and rear UWB tags.

[0025] After determining the static information, the system immediately performs a dynamic data query. Based on the vehicle dynamic parameter table, the system obtains the latest real-time operating condition data using Train_ID "A12345" and the current timestamp: Train_ID:A12345; Timestamp:2023-11-0110:00:00; Temperature:25.0 degrees Celsius; Axle_Weight:15.0 tons. This data constitutes the second layer of vehicle model information, accurately reflecting the train's instantaneous status at the moment of positioning calculation.

[0026] The onboard computer integrates the static and dynamic information retrieved above to construct a complete vehicle model database for train (A12345), the contents of which are as follows: Static information: Train number: A12345; Design length: 200.0 meters; Elastic expansion coefficient: 0.01; Front UWB tag number: Tag_001, Installation position: CabA, Height: 1.5 meters, Offset: 0.0 meters; Rear UWB tag number: Tag_002, Installation position: CabB, Height: 1.5 meters, Offset: 0.0 meters; Dynamic information: Real-time temperature: 25.0 degrees Celsius; Axle load: 15.0 tons; Through step S112, the system successfully constructs a vehicle model database containing all relevant static and dynamic information for the train, providing complete, accurate, and reliable data support for subsequent dynamic positioning.

[0027] refer to Figure 3 In step S12, the specific steps are as follows: S121: Collect the train's travel trajectory, determine the corresponding track area based on the identification of the travel trajectory, mark the track information corresponding to the track area, and determine the track map database based on the tracing of the track information; S122: Collect the initial position of the train, determine the train's operating section based on the initial position and vehicle type database, determine the overall shape of the track based on the track map database based on the initial position and vehicle type database, and determine the track geometry of the section where the train is located based on the overall shape of the track and the train's operating section.

[0028] In the embodiments of this application, the train's travel trajectory is collected, the corresponding track area is determined based on the identification of the travel trajectory, and the track information corresponding to the track area is marked. The track map database is determined based on the tracing of the track information, which is compatible with the overall consideration of track information tracing and ensures the accuracy of the track map database.

[0029] At this point, the onboard system generates a continuous travel trajectory of the train over a period of time by fusing multiple sensor sources (such as odometers, transponders, inertial measurement units, etc.). This trajectory is represented as a path with a specific direction and curvature on a two-dimensional plane. The system compares this real-time generated trajectory with a pre-stored overall network map containing all line topology relationships. By analyzing the trajectory's starting point, direction, curvature changes, and other features, the system can identify the physical track line where the train is currently located with high confidence (e.g., Line 1, Line 2, or a specific connecting line). The identification result is the logical identifier of a track area.

[0030] After the track area is determined, the system needs to perform a logical-to-physical data mapping. Each track area identifier (such as the uplink of Line 1) is bound to a specific, high-precision track map database file or database index in the system configuration. This binding relationship is established when the system is deployed. Based on the identified track area identifier, the system retrieves the database information it is bound to, and the result of tracing is to obtain a pointer or path to a specific track map database.

[0031] Based on the database pointer or path obtained in the previous step, the system accesses the storage system (local storage or network storage) and loads the corresponding track map database into the memory of the onboard computer. This database contains precise geometric information of all track sections of the line, such as mileage, curvature, gradient, speed limits, etc. After loading, the system has prepared a complete environmental data foundation for subsequent precise positioning and calculation.

[0032] Specifically, the train departs from the depot, passes through the switch area, and enters the main line; the onboard system records its path characteristics from the depot to the main line; the system matches these path characteristics with the overall line network map and identifies that the endpoint of the path connects to the northbound direction of Line 1; therefore, the system determines that the current track area of ​​the train is the northbound direction of Line 1.

[0033] After the system identified that the train was located in the uptrack area of ​​Line 1, it immediately queried the data source corresponding to that area in the internal configuration table. The configuration table showed that the data source corresponding to the uptrack area of ​​Line 1 was / map_data / line1_uptrack.db. The system completed the information marking and traced back to the specific track map database file.

[0034] The onboard system reads the database file from the solid-state drive according to the path / map_data / line1_uptrack.db and loads it into memory. At this point, the train's dynamic positioning system has successfully retrieved the complete track map database for the up-line of Line 1, which contains information on all sections from the starting point to the end point, thus preparing for the next step in S122 to accurately determine the geometry of the section where the train is located.

[0035] Furthermore, the preliminary location of the train is collected, and the train's operating section is determined based on the preliminary location and the train model database. The overall shape of the track is determined based on the track map database, and the track geometry of the train's operating section is determined based on the overall shape of the track and the train's operating section. This approach takes into account both the overall shape of the track and the overall consideration of the train's operating section, ensuring the accuracy of the track geometry of the train's operating section.

[0036] At this point, the system obtains the train's preliminary position, which is typically the three-dimensional coordinates of the train's head or center in the global coordinate system, calculated in real time by the UWB system. Simultaneously, the system retrieves the train's physical length from the train model database built in S11. Using these preliminary position coordinates, the system performs a spatial query in the track map database loaded in S121. By calculating the shortest distance between this coordinate point and the track centerline, its projection point on the track is determined. Based on the position of this projection point in the track data link, its corresponding track mileage is calculated. By comparing this mileage value with the start and end mileages (Start_Mileage, End_Mileage) of each section, the system accurately pinpoints the train's current operating section.

[0037] Acquiring information about the track environment surrounding the train provides forward-looking data for dynamic compensation calculations. Knowing only the current section is insufficient, as the train is a rigid body, with its front and rear traversing different sections, and the characteristics of the upcoming section will also affect the current calculations. Centered on the initial position, the system retrieves a data window from the track map database containing several consecutive sections before and after it. The data set within this window constitutes the overall shape of the track, reflecting the current and upcoming track environment that the train will face, such as continuous curve combinations and gradient changes.

[0038] From the analyzed overall track morphology, the key geometric parameters for subsequent calculations are precisely extracted. The system focuses on the current running section occupied by the train (determined by S122-1) and, in conjunction with the overall track morphology (look-ahead information provided by S122-2), extracts the specific geometric feature values ​​of the section and its influence range. These parameters are the direct inputs for calculating the changes in train length (∆Lcurve and ∆Lgradient) caused by curves and slopes.

[0039] Specifically, the UWB tag Tag_Head_A01 on the front of the train (M6 type, designed length 120.5 meters) is located in the global coordinate system (X=10532.4, Y=8231.7); the system performs the following operations: the system matches this coordinate with the loaded Line 1 up-line track map and calculates the projection point of this point on the track centerline; the system calculates the corresponding line mileage K10+350 based on the index of the projection point in the track data; the system traverses the section data in the track map and finds that mileage K10+350 falls within the section with starting mileage K10+200 and ending mileage K10+500; therefore, the system determines that the current running section of the train is Section_ID:S1-08.

[0040] After determining that the train is located in section S1-08, the system will perform overall track morphology analysis: taking section S1-08 as the center, the system will pre-fetch two sections forward and back one section backward; the system retrieves continuous data of sections S1-06, S1-07, S1-08, S1-09, and S1-10; by analyzing this data window, the system identifies that the train is at the beginning of a right turn (S1-08) with a radius of 300 meters, which will continue to section S1-09, and there is a 0.3% uphill slope in the current section.

[0041] The system ultimately focuses on section S1-08 where the train is located and extracts its specific geometric parameters from the overall track shape data: Section_ID: S1-08; Start_Mileage: K10+200; End_Mileage: K10+500; Curve: 300 (meters, representing the radius of curvature, a positive value indicates a right turn); Slope: 3 (‰, representing a 0.3% uphill slope). Through the above process, the system provides accurate track geometry data for the train's dynamic positioning. This data will be directly used in subsequent steps to calculate the changes in train length caused by curves and slopes, thereby ensuring the accuracy of the verification.

[0042] refer to Figure 4 In step S13, the specific steps are as follows: S131: Dynamically identify the geometry of the track and determine multiple track regions during the identification process. Determine the grade coefficient of each track region based on its location, corresponding shape, and past train travel events. Based on the comparison of the grade coefficients of each track region, determine multiple key regions. S132: In multiple key areas, the corresponding orbital space is determined based on the identification of each key area, and the marking position of the trackside UWB base station is determined according to the spatial shape of the orbital space and the corresponding surrounding environment. S133: Determine the marking location of onboard UWB tags based on train detection, and determine the data interaction system based on the marking locations of various key areas, trackside UWB base stations, and onboard UWB tags.

[0043] In the embodiments of this application, the track geometry is dynamically identified, and multiple track regions are determined during the identification process. The grade coefficient of each track region is determined based on its location, corresponding shape, and past train travel events. Multiple key regions are determined based on the comparison of the grade coefficients of each track region, which takes into account the overall consideration of comparing the grade coefficients of each track region and ensures the accuracy of multiple key regions.

[0044] At this point, the system traverses and scans the overall track shape (e.g., multiple consecutive sections) determined in S12. The system has a built-in recognition rule base for automatically detecting track areas with specific geometric features, including but not limited to: curvature change rate: when the radius of curvature is less than a preset threshold (e.g., less than 500 meters), the system identifies it as a small-radius curve area; slope change rate: when the absolute value of the slope is greater than a preset threshold (e.g., greater than 4‰), the system identifies it as a large-slope area; special structural points: the system identifies turnout areas, tunnel entrance / exit areas, platform areas, etc., through attribute markers in the track map database. Through this process, the system divides and marks a continuous track line into multiple track areas with different features.

[0045] For each track area, the system calculates a comprehensive rating coefficient, a dimensionless quantitative indicator used to comprehensively measure the potential negative impact of the area on UWB positioning accuracy. The calculation is based on three dimensions: Area Location: assessing the area's importance within the track network; for example, areas located at crossovers or hubs have more severe consequences for positioning errors and are therefore given a higher base weight; Area Morphology: scoring based on the severity of geometric features; smaller curve radii and steeper ramps result in higher scores; for example, a curve with a 300-meter radius receives a higher morphology score than a 600-meter radius curve; Historical Operating Events: the system queries the historical operating database to analyze whether the area has frequently experienced UWB positioning-related anomalies (such as signal loss, sudden drops in positioning accuracy, severe multipath effects, etc.); the higher the frequency and severity of historical anomalies, the higher the area's score. The system uses a weighted fusion algorithm to comprehensively calculate the scores from these three dimensions, ultimately deriving the rating coefficient for each track area.

[0046] The system sorts the grade coefficients of all track regions; the system sets one or more filtering thresholds, or directly selects the top N regions; all track regions with grade coefficients exceeding the thresholds, or ranked within N, will be officially identified and marked as key regions. These regions are the core targets for key optimization and resource allocation in subsequent steps.

[0047] Specifically, the system analyzed the overall shape of the Line 1 up-line track that the train was about to enter and performed dynamic identification: the system scanned that the radius of curvature of section S1-08 was 300 meters, which is less than the threshold of 500 meters, so it was identified and marked as a small-radius curve area; the system scanned that the slope of section S1-10 was 5‰, which is greater than the threshold of 4‰, so it was identified and marked as a high-slope area; the system found in the track map attributes that the starting point of section S1-12 was marked as a tunnel entrance, so it was identified and marked as a tunnel entrance / exit area; thus, the system identified three potentially risky track areas from the continuous track.

[0048] The system calculates grade coefficients for the three identified areas: For the small-radius curve area S1-08: Area location: ordinary mainline section, basic weight is 1.0; Area shape: radius 300 meters, shape score is 80 points (out of 100); Historical events: data shows that the area occasionally experiences signal fluctuations, historical score is 30 points; Comprehensive grade coefficient calculation: 1.0 x 80 + 30 = 110; For the steep slope area S1-10: Area location: ordinary mainline section, basic weight is 1.0; Area shape: slope 5‰, morphological score of 70; Historical events: data is stable, historical score of 10; Comprehensive grade coefficient calculation: 1.0X70+10=80; For tunnel entrance and exit area S1-12: Area location: key environmental transition point, basic weight of 1.2; Area morphology: tunnel entrance, morphological score of 85; Historical events: data shows that more than 30% of trains experience NLOS (non-line-of-sight) problems when passing through, historical score of 60; Comprehensive grade coefficient calculation: 1.2X85+60=162.

[0049] The system compares the calculated three level coefficients: S1-08 (110), S1-10 (80), and S1-12 (162). The system sets the threshold to 100 and selects the top two regions. S1-12 has a level coefficient of 162 > 100 and ranks first, so it is identified as critical region 1. S1-08 has a level coefficient of 110 > 100 and ranks second, so it is identified as critical region 2. S1-10 has a level coefficient of 80 < 100 and is not selected as a critical region, but it is still considered as a general area of ​​concern. Through the complete process of S131, the system intelligently identifies S1-12 (tunnel entrance) and S1-08 (small radius curve), the two most challenging critical regions for the current operation of the train, from the static track data, providing clear targets for the next step of optimization deployment.

[0050] Furthermore, in multiple key areas, the corresponding track space is determined based on the identification of each key area. The marking position of the trackside UWB base station is determined according to the spatial shape of the track space and the corresponding surrounding environment. This takes into account the overall consideration of the spatial shape of the track space and the corresponding surrounding environment, ensuring the accuracy of the marking position of the trackside UWB base station.

[0051] At this point, for each key area identified in S131, the system extracts its precise track space data from the track map database. This is not just a simple set of start and end coordinates, but a dataset containing complete three-dimensional geometric information of the area, such as: centerline three-dimensional coordinate sequence: continuous three-dimensional points that constitute curves or ramps; spatial envelope: describing the volume occupied by the track area in three-dimensional space; structural boundaries: precise geometric models of large obstacles such as tunnel walls, platform edges, and viaduct piers. This track space is the basis for wireless signal simulation.

[0052] The system utilizes the established orbital space and structural boundaries, combined with the physical characteristics of the UWB signal (such as frequency and transmit power), to perform ray tracing or channel impulse response simulations. The purpose of the analysis is to evaluate the problems encountered by the UWB signal within this specific space, such as: signal obstruction by tunnel walls or large metal structures; signal broadening and distortion after multiple reflections before reaching the receiver; and energy loss of the signal over long distances or after passing through specific media (such as humid air inside a tunnel). The simulation output is a heatmap of the signal coverage quality in the area and a series of potential optimal observation points.

[0053] The system integrates the analysis results of the first two sub-steps and uses an optimization algorithm to determine the marking locations of one or more UWB base stations for each key area. These locations are theoretically optimal deployment points, with the following objectives: ensuring a direct signal propagation path between the tag and at least one base station; deploying base stations in locations that reduce strong reflection paths; providing richer geometric constraints for positioning calculations by deploying base stations at different heights or sides; and outputting marking locations typically accompanied by a priority or deployment scheme description to guide actual engineering deployment or dynamic data interaction strategies.

[0054] Specifically, for the identified key area S1-12 (tunnel entrance), the system extracted the three-dimensional centerline coordinates of the S1-12 section and the 50-meter sections before and after it from the track map database; the system retrieved the building information model (BIM) of the area and constructed a three-dimensional spatial envelope, clearly representing the transition space from the open ground to the narrow tunnel interior; the system identified the key structural boundaries within this space, including the large metal information sign at the entrance, the concrete arch structure at the tunnel entrance, and the sidewalls.

[0055] The system performs simulation analysis on the orbital space of S1-12 (tunnel entrance): The system runs ray tracing simulation to simulate UWB signals emitted from different potential base station locations; the simulation results show that the signal coverage is good outside the tunnel entrance, but the signal strength drops sharply once inside the tunnel, and there is a serious first reflection path; the system identifies a signal blind spot or weak zone 10 meters inside the tunnel entrance due to the obstruction of the arch; the simulation heat map shows that 15 meters outside the tunnel entrance is a relatively stable anchor point location.

[0056] Based on the analysis of S1-12 (tunnel entrance), the system determines the marking positions of its trackside UWB base stations: Marking position 1 (high priority): located 15 meters outside the tunnel entrance. Simulation shows that this position has good line of sight with the front tag of the train about to enter the tunnel and can serve as a bridge base station for signals to enter from the outside; Marking position 2 (high priority): located 10 meters inside the tunnel entrance, above the tunnel sidewall. Simulation shows that this position can effectively cover the weak signal area at the tunnel entrance and provide the first internal positioning anchor point for trains entering the tunnel. The system generates a suggestion that the two marking positions should work together to form a bridging signal coverage pattern to smoothly handle the signal transition when trains enter the tunnel from the open area.

[0057] Therefore, the marking positions of onboard UWB tags are determined based on train detection. The data interaction system is established according to the marking positions of various key areas, trackside UWB base stations, and onboard UWB tags. This system takes into account the overall consideration of the marking positions of various key areas, trackside UWB base stations, and onboard UWB tags, ensuring the accuracy of the data interaction system. At the same time, the introduction of various key areas and the overall consideration of trackside UWB base stations and onboard UWB tags improve the accuracy of the data interaction system.

[0058] At this point, the system directly reads the configured UWB tag information from the vehicle model database built in S11. This process confirms the following key parameters: the tag's unique identifier (Tag_ID), used to distinguish different tags in communication; the tag's installation position on the train (Install_Position), such as the front or rear of the train; and the tag's physical installation parameters, including the installation height from the rail surface and the offset from the train's centerline. These marked positions are not only physical locations but also key calibration parameters for subsequently converting UWB ranging results into the actual position and length of the train.

[0059] The key areas output from the preceding steps, the marked locations of trackside UWB base stations, and the marked locations of vehicle-mounted UWB tags are associated, and a data interaction system between them is defined. The data interaction system includes three core elements: Entity List: a list containing all base stations and tags participating in the interaction, each entity has a unique ID, coordinates, and status; Spatial Topology: describes the spatial relationships between entities, such as which base station signal coverage overlaps with which key area; Interaction Protocol and Strategy: defines the rules for data collection, such as sampling frequency, communication timing, data filtering criteria, and special interaction modes for different key areas (e.g., when entering a key area, a high-precision mode is enabled, prioritizing communication with specific base stations).

[0060] Specifically, when the train initiates the dynamic positioning process, the system performs the following operations: The system retrieves the UWB tag information table from the train model database; confirms the existence of two valid tags: Tag_Head_A01 and Tag_Tail_A01; reads their installation locations: Tag_Head_A01 is located in CabA (front of the train), and Tag_Tail_A01 is located in CabB (rear of the train); reads their physical installation parameters: the installation height of both tags is 3.8 meters, and the offset from the train's centerline is 0.1 meters; at this point, the marking positions of the onboard UWB tags are completely determined, and the system knows the precise physical reference points of these two signal sources on the train.

[0061] The system constructs a complete data interaction system for the train: the system includes all conventional UWB base stations along the line, and specifically marks the base station locations optimized for two key areas: S1-12 (tunnel entrance) and S1-08 (curve); simultaneously, the list includes the train's onboard tags Tag_Head_A01 and Tag_Tail_A01; the system establishes spatial associations, for example, when Tag_Head_A01 enters the S1-12 area 500 meters before entering the area, it will enter the pre-communication range of the optimized base station in that area; the system is configured... The following rules apply: Normal mode: The tag communicates with the three or more base stations with the strongest signals at a frequency of 10Hz; Critical area mode: When the train initially enters the S1-12 or S1-08 area, it automatically switches to this mode; In this mode, Tag_Head_A01 and Tag_Tail_A01 increase the communication frequency to 20Hz and prioritize ranging with the optimized base stations marked for this area in S132. At the same time, the system will enable a more stringent data quality assessment algorithm to eliminate abnormal ranging values ​​contaminated by multipath effects.

[0062] refer to Figure 5 In step S14, the specific steps are as follows: S141: Real-time monitoring data interaction system, detect the data interaction system, determine multiple data during the detection process, determine the corresponding data combination based on multiple data, the train's driving position and corresponding overall shape, and determine the dynamic length of the train based on the identification of the data combination; S142: Based on the dynamic detection of the train, determine the on-board UWB tags at the front and rear of the train, and mark the positioning data of the on-board UWB tags at the front and rear of the train. Determine the UWB ranging length based on the positioning data of the on-board UWB tags at the front and rear of the train.

[0063] In the embodiments of this application, a real-time monitoring data interaction system is used to detect the data interaction system and determine multiple data during the detection process. Based on the multiple data, the train's driving position, and the corresponding overall shape, a corresponding data combination is determined. The dynamic length of the train is determined based on the identification of the data combination, which takes into account the overall consideration of the identification of the data combination and ensures the accuracy of the dynamic length of the train.

[0064] At this time, the system monitors the data interaction system defined by S13 in real time at a high-frequency sampling rate (e.g., 10-20Hz). Within each sampling period, the system captures and integrates a data combination, which is a data packet containing various heterogeneous data, and its composition is as follows: UWB raw ranging data: Time-of-flight (TOF) measurements from the front and rear tags and multiple trackside base stations, which are the basis for positioning calculations; Train status parameters: Train dynamics and status information acquired in real time through the onboard control network (such as MVB, Ethernet), including instantaneous speed, acceleration, current axle load, car body temperature, etc.; Track geometry data: Track feature data obtained from S12 that precisely matches the current position of the train, including the radius of curvature and gradient of the current section, as well as preview information of the preceding and following sections. This data combination is the complete spatiotemporally synchronized data content for all subsequent calculations.

[0065] The system utilizes the acquired data combination to perform multi-dimensional corrections on the train's designed length in the database to calculate the current instantaneous dynamic length (L_actual). This calculation strictly follows the core idea in the document, namely, considering factors such as curves, gradients, and elastic expansion / contraction. The calculation model is as follows: Curve compensation (∆L_curve): Based on the radius of curvature in the track geometry data and the train's designed length, the system applies geometric principles to calculate the change in the train's length projected onto the track centerline due to the path difference between the outer and inner rails when the train is on a curve; the smaller the radius, the greater the compensation. Gradient compensation (∆L_gradient): Based on the gradient in the track geometry data, the system calculates the slight change in the train's projected length in the horizontal direction due to the car body tilt when on a gradient. Elastic expansion / contraction compensation (∆L_coupling): Based on the real-time temperature and axle load in the train's state parameters, as well as the elastic expansion / contraction coefficient in the model database, the system calculates the physical length changes caused by thermal expansion and contraction and suspension system compression.

[0066] The system algebraically adds the designed vehicle length to the total compensation calculated in the previous step to obtain the final, high-precision dynamic vehicle length L_actual. This value represents the most accurate physical length of the train at the current time, current position, and current operating conditions, and serves as the benchmark for subsequent UWB accuracy verification.

[0067] Specifically, within a 100-millisecond sampling period, the system determined the following data combinations for the train (M6 type, designed length 120.5 meters): UWB ranging data: the ranging values ​​of the front tag Tag_Head_A01 with base stations BS01, BS02, and BS03 are [45.21m, 102.55m, 150.10m], respectively; the ranging values ​​of the rear tag Tag_Tail_A01 with base stations BS01, BS02, and BS03 are [165.88m, 213.40m, 260.95m], respectively; train status parameters: instantaneous speed 60km / h, acceleration 0m / s². 2 The real-time vehicle body temperature is 35°C, and the current axle load is 15 tons; track geometry data: the train is currently located in section S1-08, with a radius of curvature of 300 meters and a gradient of 0‰.

[0068] The system uses the above data combination to calculate the dynamic length compensation of the train: Curve compensation (∆L_curve): Based on the geometric model, the system calculates that on a curve with a radius of 300 meters, a 120.5-meter-long train will experience a length projection increment of approximately 0.02 meters due to the difference in the inner and outer rail paths; Gradient compensation (∆L_gradient): If the current section is a flat slope, this compensation is 0 meters; Elastic expansion compensation (∆L_coupling): Thermal expansion compensation: The real-time temperature of 35°C is 1 meter higher than the standard temperature of 20°C. 5°C; Based on the elastic coefficient in the vehicle model database, the thermal expansion is calculated as 120.5m x 0.0008 / °C x 15°C = 1.446 meters; Suspension compression compensation: Based on the current axle load of 15 tons and the suspension stiffness model, the vehicle body is estimated to be compressed by 0.05 meters; Total elastic compensation: 1.446 - 0.05 = 1.396 meters; Total compensation amount: ∆L_curve + ∆L_gradient + ∆L_coupling = 0.02m + 0m + 1.396m = 1.416 meters.

[0069] The system performs the final dynamic vehicle length synthesis: Design vehicle length (L_nominal): obtained from the vehicle model database, is 120.5 meters; Final dynamic vehicle length (L_actual): L_actual = L_nominal + total compensation = 120.5m + 1.416m = 121.916 meters; Through the complete process of S141, the system successfully transforms a static design length into a dynamic physical reference that is highly coupled with the actual operating environment. This value of 121.916 meters will serve as a benchmark for measuring the accuracy of UWB ranging.

[0070] Furthermore, based on the dynamic detection of the train, the on-board UWB tags at the front and rear of the train are determined, and the positioning data of the on-board UWB tags at the front and rear of the train are marked. The UWB ranging length is determined based on the positioning data of the on-board UWB tags at the front and rear of the train. This overall consideration takes into account the positioning data of the on-board UWB tags at the front and rear of the train, ensuring the accuracy of the UWB ranging length.

[0071] At this point, the system confirms that both the front and rear UWB tags are in normal working order through onboard detection mechanisms (such as network heartbeat and power status monitoring); the system assigns a logical identifier to each tag and marks its output positioning data in real time in the data stream. This process ensures that: the system can distinguish whether any positioning data point was generated by the front or rear tag; the data streams of the front and rear tags are precisely timestamped to ensure that subsequent calculations use snapshot data from the same moment; and the system checks whether each tag has successfully calculated valid 3D coordinates in each sampling period.

[0072] Based on the three-dimensional coordinates of the two tags at the same moment, the straight-line distance between them is calculated. At this time, the system uses the three-dimensional Euclidean distance formula to calculate the straight-line distance between the front and rear tags in space. This distance is the result directly calculated by the UWB system through multi-base station ranging and positioning algorithms (such as TDOA), and is the original measurement value of the distance between the tags by the UWB system. Since the UWB tags are not installed at the very front and rear of the train body, their measured distance needs to be corrected to more accurately represent the train length. The system reads the pre-calibrated tag installation offset (∆L_install) from the vehicle model database built by S11, which is the sum of the distance from the front tag to the front end face and the distance from the rear tag to the rear end face. This offset is added to the original spatial distance to obtain the final UWB ranging length.

[0073] Specifically, during train operation, the system performs the following operations: The system detects that both the front tag (Tag_Head_A01) and the rear tag (Tag_Tail_A01) are online and in normal condition; the system continuously monitors the positioning results published by the onboard computer on the data bus; when a coordinate data packet is received, the system checks its source address; if the source address is Tag_Head_A01, the system marks the data packet as front positioning data; if the source address is Tag_Tail_A01, the system marks the data packet as rear positioning data; the system adds the same high-precision timestamp to both data packets, for example, T0=16988064012.500, to ensure that they represent the tag position at the same instant T0.

[0074] At the same time T0, the system performs UWB ranging length calculation: The system extracts the positioning data at time T0 from the data stream marked S142-1: the positioning data of the front tag Tag_Head_A01 (x_head, y_head, z_head) = (100.0, 200.0, 3.8); the positioning data of the rear tag Tag_Tail_A01 (x_tail, y_tail, z_tail) = (-20.8, 198.5, 3.8); L_UWB_raw ≈ 120.819 meters; the system queries the train model database to find that the distance between the front tag and the front tag is... The distance from the end face to the rear label is 0.1 meters, and the distance from the rear label to the end face of the vehicle is 0.1 meters; the total installation offset ∆L_install = 0.1m + 0.1m = 0.2m; the final UWB ranging length is: L_UWB = L_UWB_raw + ∆L_install = 120.819m + 0.2m = 121.019m; through S142, the system obtains a length value purely measured and calculated by the UWB system: 121.019 meters. This value will be directly compared with the dynamic vehicle length calculated by S141, which takes into account various physical factors. The difference will directly reflect the ranging accuracy of the UWB system in the current operating environment.

[0075] refer to Figure 6 In step S15, the specific steps are as follows: S151: Collect the UWB ranging length and the dynamic length of the train, determine the corresponding difference based on the comparison between the UWB ranging length and the dynamic length of the train, and determine the UWB ranging accuracy based on the difference, the mapping relationship of UWB ranging accuracy and the train. S152: Based on the tracing of the train, determine the train's past error data and mark the corresponding driving scenarios. Construct a corresponding error control model based on the train's past error data, the marked driving scenarios, and the train's current driving scenario. S153: Collect UWB ranging accuracy and real-time train position data, determine dynamic compensation amount based on UWB ranging accuracy and error control model, and determine dynamic positioning data of train based on dynamic compensation amount and real-time train position data.

[0076] In the embodiments of this application, the UWB ranging length and the dynamic length of the train are collected. The corresponding difference is determined by comparing the UWB ranging length and the dynamic length of the train. The UWB ranging accuracy is determined based on the difference, the mapping relationship of UWB ranging accuracy and the train. This approach takes into account the difference, the mapping relationship of UWB ranging accuracy and the overall consideration of the train, thus ensuring the accuracy of UWB ranging.

[0077] At this point, the system collects data from two parallel calculation results in S14 at a precise timestamp: UWB ranging length (L_UWB): This is the calculation result from S142, representing the distance between tags obtained by the UWB system through wireless signal measurement and geometric calculation; Dynamic vehicle length (L_actual): This is the calculation result from S141, representing the actual physical length of the train under the current operating conditions after compensation by the physical model. The comparison of these two values ​​is the core innovation of the entire method, as it directly compares a radio measurement result with a highly reliable physical benchmark.

[0078] The system calculates the absolute difference Δ between the UWB ranging length and the dynamic vehicle length. This difference Δ is a raw, unprocessed error indicator that directly quantifies the measurement deviation of the UWB system at the current time and location. The calculation formula is: Δ = |L_UWB - L_actual|. The magnitude of this difference Δ directly reflects the severity of the impact on the UWB signal in the current environment. A small Δ value means that the signal propagation environment is good (such as an open straight road), while a large Δ value strongly suggests the existence of serious non-line-of-sight (NLOS) propagation or multipath interference.

[0079] While the raw difference Δ is intuitive, it lacks statistical significance. The system transforms Δ into a more professional accuracy metric through a pre-defined mapping relationship or transformation function. The mapping can be a simple lookup table or a complex statistical model. It considers multiple factors, such as the magnitude of the difference Δ, its duration, and the historical error distribution of the current scene. The final output can take various forms, such as: Instantaneous Error Estimation: directly using Δ as the error estimate at the current moment; Root Mean Square Error (RMSE): calculating the average accuracy over a period of time using multiple Δ values ​​within a sliding window; Confidence Interval Error (CEP): providing an indicator that the true location has a 95% probability of falling within a circle of radius R centered on the measurement point. This process transforms a raw physical deviation into a statistically significant UWB ranging accuracy indicator that can be used to evaluate system performance.

[0080] Specifically, at timestamp T0, the system performs data acquisition and comparison: the system obtains the train's UWB distance measurement length from module S142: L_UWB=121.010 meters; the system obtains the train's dynamic length from module S141: L_actual=121.916 meters; the system sets these two values ​​side by side, preparing to calculate the difference; at this point, the system can preliminarily determine that the UWB measurement value is significantly smaller than the physical reference value.

[0081] The system calculates the difference based on the data collected in the previous step: Δ=|121.010m-121.916m|=|-0.906m|=0.906 meters; the system obtains an original difference of 0.906 meters. This error of nearly 1 meter is huge for a UWB system with centimeter-level accuracy. Based on this, the system judges that in the current scenario, UWB ranging is significantly affected by a negative bias (the measured value is less than the true value).

[0082] The system determines accuracy based on the original difference of 0.906 meters: the system queries the built-in mapping table, which associates historical difference data with confidence levels in the tunnel entrance scenario; the system finds that in the tunnel entrance scenario, the instantaneous difference of 0.906 meters corresponds to a relatively high error level; the system finally determines and outputs: the current UWB ranging accuracy is low, the instantaneous estimation error is -0.906 meters, and the positioning error radius at 95% confidence level exceeds 1.0 meter.

[0083] Furthermore, based on the train's tracing, the train's past error data is determined, and the corresponding driving scenarios are marked. Based on the train's past error data, the marked driving scenarios, and the train's current driving scenario, a corresponding error control model is constructed, which is compatible with the overall consideration of train tracing and ensures the accuracy of the train's past error data.

[0084] At this point, the system accesses the train's long-term operation database, which records the complete process of each positioning verification. The system retrieves all historical records, each containing the original difference Δ calculated at that time (i.e., the output of S151). This Δ value is the target variable (i.e., the true value) for model learning. For each historical error record, the system automatically labels the driving scenario in which it occurred. This scenario is a multi-dimensional feature vector used to describe the environment and state at the time of the error. Key feature dimensions include: environment type: such as single-track tunnel, multi-track tunnel, open section, platform area, parking lot / section, etc.; geometric features: such as curve radius of curvature, gradient; dynamic state: such as train speed, acceleration; spatial location: such as relative distance to the base station, specific location within the section. Through this process, the system constructs a large historical error dataset with precise scene labels.

[0085] The system uses the scenario-error dataset built in the previous step to train a model capable of predicting errors. The goal of this model is to learn the complex mapping relationship between driving scenario features and error values ​​Δ. Different types of models can be selected based on the amount and complexity of the data. The document mentions a regression model based on an exponential function: τᵢ=a·eᵇˣ. 1 +c·eᵈˣ 1, where xᵢ can be variables such as distance or speed; or more modern machine learning models can be used, such as decision trees / random forests, which can handle nonlinear relationships and feature interactions well; neural networks, which can learn very complex patterns and are especially suitable for massive amounts of data.

[0086] The system uses most of the historical data to train the model and uses a portion of reserved data to verify the model's accuracy. The goal of training is to make the model's output prediction error close to the actual error Δ in the historical records. The final error control model takes the feature vector of the current driving scenario as input and outputs a quantifiable prediction error or dynamic compensation amount specific to the scenario.

[0087] The system extracts the feature vector of the train's current driving scene in real time. The composition of this vector is completely consistent with the feature dimensions used when labeling historical data. The system inputs this current scene feature vector into the error control model constructed by S152-2. The model outputs a predicted error value Δ_predicted, which is the system's best estimate of the current UWB measurement error. To facilitate subsequent compensation, the system usually converts it into a dynamic compensation amount. The magnitude of the compensation amount is usually equal to the prediction error, but in the opposite direction. For example, if the predicted UWB measurement value is systematically 0.9 meters smaller, then the dynamic compensation amount is +0.9 meters.

[0088] Specifically, the system traces and marks historical data for the train: In the past six months of operation records, the system found thousands of positioning verification records for the Line 1 northbound route; the system marks each record with a scene label; for example, it found 15 records with the common characteristics of: {Environment type: tunnel entrance, speed range: 55-65 km / h, curve curvature: none}, and the original difference Δ for these 15 records were [0.88, 0.92, 0.85, 0.95, ..., 0.89] meters respectively; the system also found 50 records with the characteristics of: {Environment type: open straight road, speed range: 55-65 km / h, curve curvature: none}, and the Δ values ​​for these records were generally small, such as [0.05, 0.08, 0.04, ..., 0.06] meters.

[0089] The system constructs an error control model for the train: the system selects a random forest regression model because it can handle classification features and numerical features such as speed and curvature for different scene types well; the system inputs all labeled historical data (thousands of records) into the model for training; the model learns that when the environment type is a tunnel entrance and the speed is around 60km / h, the error Δ will increase significantly; while when the environment type is an open straight road, the error Δ is very small; after training, the system obtains a fixed error control model for the train. This model is like an error predictor. As long as the current scene features are input, it can output a predicted error value.

[0090] The system application model determines the dynamic compensation amount for the train: At time T0, the system extracts the current scene features of the train: {Environment type: tunnel entrance, speed: 60km / h, curve curvature: none, distance from entrance: 5m}; the system inputs this feature vector into the pre-trained random forest model; the model outputs a prediction error Δ_predicted = 0.88 meters; based on this, the system determines that the dynamic compensation amount in the current scene is +0.88 meters, which is used to correct the original positioning data in the next step S153.

[0091] Therefore, by collecting UWB ranging accuracy and real-time train position data, determining the dynamic compensation amount based on the UWB ranging accuracy and error control model, and then determining the train's dynamic positioning data based on this dynamic compensation amount and the train's real-time position data, the overall consideration of dynamic compensation amount and real-time train position data is taken into account, ensuring the accuracy of the train's dynamic positioning data. At the same time, the introduction of UWB ranging accuracy further controls the error control model, realizing the overall consideration of the error control model, UWB ranging accuracy, and real-time train position data, thus improving the accuracy of the train's dynamic positioning data.

[0092] At this point, the system acquires the output of S151, which is not just a numerical value, but also state information containing confidence levels. For example, a low-precision state will trigger the system to rely more on the model's prediction, while a high-precision state will cause the system to give higher weight to the raw UWB data. The system acquires the raw positioning data directly output by the UWB positioning solution unit without any compensation. This is usually the three-dimensional coordinates (x_raw, y_raw, z_raw) of the front and rear labels, or the coordinates of the train's center point and attitude angle calculated based on these coordinates. Based on the current timestamp and train position, the system extracts the current driving scene feature vector that is completely consistent with the feature dimensions used when constructing the error control model in S152. This vector is the benchmark for activating the error control model and obtaining the compensation amount.

[0093] The system inputs the current driving scene feature vector prepared in the previous step into the error control model constructed in S152; the model outputs a basic prediction error value or prediction compensation amount; the system adjusts the weights of the model's prediction results based on the UWB ranging accuracy status collected in S151. This is a key fusion step, and the logic is as follows: When the accuracy is low: the system gives high confidence to the model's prediction and will directly use the compensation amount output by the model, or even add a safety margin on this basis; when the accuracy is high: the system gives high confidence to the original UWB data and will proportionally reduce the compensation amount output by the model, or only compensate when the prediction error exceeds a certain threshold; when the accuracy is medium: the system adopts a weighted average method, for example, compensation amount = α × model prediction compensation amount + (1-α) × 0 (i.e., no compensation), where α is a coefficient related to the accuracy level; after dynamic adjustment, the system obtains a final, well-considered dynamic compensation amount, which is usually a three-dimensional vector (Δx, Δy, Δz), which precisely indicates the displacement that needs to be corrected in various directions in space.

[0094] The system adds the dynamic compensation vector to the real-time position data vector point by point. This is a precise vector addition operation used to correct the spatial offset of the original positioning data. The result of the operation is the dynamic positioning data, which represents the best estimate of the train's true physical position after multiple verifications and intelligent compensation. The system will output this high-precision positioning data to the Automatic Train Control (ATC) system, Automatic Train Operation (ATO) system, or other downstream applications that require high-precision position information.

[0095] Specifically, at time T0, the system performs data acquisition: the system obtains a low-precision status from S151, with an estimated instantaneous error of -0.897 meters. This status tells the system that compensation must be performed; the system obtains the original UWB positioning coordinates of the train's front tag: (x_raw, y_raw, z_raw) = (100.0, 200.0, 3.8); the system extracts the current scene feature vector: {environment type: tunnel entrance, speed: 60km / h, distance from entrance: 5m}, and this vector will be sent to the error control model constructed in S152.

[0096] The system determines the dynamic compensation amount for the train: The system inputs the scene feature vector into the model, and the model outputs a predicted compensation vector (Δx_model, Δy_model, Δz_model) = (0.0, +0.88, 0.0); Since the current accuracy is low, the system decides to fully trust the model prediction; the adjustment coefficient α = 1.0 is set; the final compensation amount is: Δx = α × Δx_model = 1.0 × 0.0 = 0.0 meters; Δy = α × Δy_model = 1.0 × 0.88 = 0.88 meters; Δz = α × Δz_model = 1.0 × 0.0 = 0.0 meters; the system determines the final dynamic compensation vector to be (0.0, +0.88, 0.0).

[0097] The system determines the train's final dynamic positioning data: The system performs vector addition: x_final=x_raw+Δx=100.0+0.0=100.0 meters; y_final=y_raw+Δy=200.0+0.88=200.88 meters; z_final=z_raw+Δz=3.8+0.0=3.8 meters; The system outputs the final dynamic positioning data of the train head after compensation as (100.0,200.88,3.8), with a highly reliable tag attached. This data point is closer to the real physical location than the original UWB positioning result and can be directly used for key decisions to ensure train operation safety.

[0098] Please see Figure 7 , Figure 7 This is a schematic diagram of the structural composition of the dynamic positioning system for a train according to an embodiment of the present invention; the dynamic positioning system for the train includes: The vehicle model database module 21 is used to construct the corresponding vehicle model database based on the vehicle basic information table, the UWB tag information table, and the vehicle dynamic parameter table. The track geometry module 22 is used to collect track map database and determine the track geometry of the section where the train is located based on the vehicle type database, track map database and the initial position of the train. The data interaction system module 23 is used to determine multiple key areas based on the identification of the track geometry, and to determine the data interaction system based on each key area, the trackside UWB base station and the vehicle-mounted UWB tag; UWB ranging length module 24 is used to determine the train's data combination based on the detection of the data interaction system in the data interaction system, and to determine the dynamic train length based on the identification of the data combination; and to determine the UWB ranging length based on the positioning data of the onboard UWB tag at the front of the train and the positioning data of the onboard UWB tag at the rear of the train. The dynamic positioning module 25 is used to determine the UWB ranging accuracy based on the difference between the UWB ranging length and the dynamic length of the train. At the same time, it collects the train's past error data, constructs a corresponding error control model based on the past error data and the train's current driving scenario, and determines the train's dynamic positioning data based on the error control model, the UWB ranging accuracy, and the train's real-time position data.

[0099] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all 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.

Claims

1. A dynamic positioning method for trains, characterized in that, include: A corresponding vehicle model database is constructed based on the vehicle basic information table, the UWB tag information table, and the vehicle dynamic parameter table; Collect track map database, and determine the track geometry of the section where the train is located based on the vehicle type database, track map database, and the initial location of the train; Based on the identification of the track geometry, multiple key areas are identified, and a data interaction system is determined based on each key area, trackside UWB base station, and vehicle-mounted UWB tag. The data interaction system includes an entity list, spatial topology, and interaction protocols and strategies; In the data interaction system, the train's data combination is determined based on the detection of the data interaction system, and the dynamic train length is determined based on the identification of the data combination; the UWB ranging length is determined based on the positioning data of the onboard UWB tag at the front of the train and the positioning data of the onboard UWB tag at the rear of the train; the UWB ranging length represents the distance between tags obtained by the UWB system through wireless signal measurement and geometric calculation. The UWB ranging accuracy is determined by the difference between the UWB ranging length and the dynamic train length. At the same time, historical error data of the train is collected, and a corresponding error control model is constructed based on this historical error data and the current train operating scenario. The dynamic positioning data of the train is determined based on this error control model, the UWB ranging accuracy, and the real-time position data of the train.

2. The dynamic positioning method for trains according to claim 1, characterized in that, The construction of the corresponding vehicle model database based on the vehicle basic information table, UWB tag information table, and vehicle dynamic parameter table includes: Collect train numbers and train databases, determine the train's information space based on the matching of the train database and the train number, and determine the vehicle basic information table, UWB tag information table and vehicle dynamic parameter table based on the traversal of the train's information space. The first level of vehicle model content is determined based on the vehicle basic information table and the UWB tag information table. The second level of vehicle model content is determined based on the vehicle basic information table and the vehicle dynamic parameter table. The corresponding vehicle model database is then determined based on the first level of vehicle model content and the second level of vehicle model content.

3. The dynamic positioning method for trains according to claim 1, characterized in that, The process of collecting track map databases, determining the track geometry of the train's section based on the vehicle type database, track map database, and the train's preliminary location, includes: The system collects the train's travel trajectory, identifies the corresponding track area based on the identification of the travel trajectory, marks the track information corresponding to the track area, and determines the track map database based on the tracing of the track information; The system collects the initial location of the train, determines the train's operating section based on the initial location and the train model database, determines the overall shape of the track based on the track map database, and determines the track geometry of the section where the train is located based on the overall shape of the track and the train's operating section.

4. The dynamic positioning method for trains according to claim 1, characterized in that, The process involves identifying multiple key regions based on the track geometry, and determining a data interaction system based on these key regions, trackside UWB base stations, and vehicle-mounted UWB tags. This includes: The track geometry is dynamically identified, and multiple track regions are determined during the identification process. The grade coefficient of each track region is determined based on its location, corresponding shape, and past train operation events. Multiple key regions are determined based on the comparison of the grade coefficients of each track region.

5. The dynamic positioning method for trains according to claim 4, characterized in that, The process of identifying multiple key regions based on the recognition of the track geometry, and determining a data interaction system based on each key region, trackside UWB base station, and vehicle-mounted UWB tag, also includes: In multiple key areas, the corresponding orbital space is determined based on the identification of each key area, and the marking position of the trackside UWB base station is determined according to the spatial shape of the orbital space and the corresponding surrounding environment. The marking location of onboard UWB tags is determined based on train detection, and the data interaction system is determined based on the marking locations of various key areas, trackside UWB base stations, and onboard UWB tags.

6. The dynamic positioning method for trains according to claim 1, characterized in that, In the data interaction system, the train's data combination is determined based on the detection of the data interaction system, and the dynamic length of the train is determined based on the identification of the data combination. The UWB ranging length is determined based on the positioning data from the vehicle-mounted UWB tag at the front and rear of the vehicle, including: The system monitors the data interaction system in real time, detects the data interaction system, and identifies multiple data points during the detection process. Based on these multiple data points, the train's driving position, and its overall shape, the system determines the corresponding data combination. The dynamic length of the train is then determined based on the identification of this data combination.

7. The dynamic positioning method for trains according to claim 6, characterized in that, In the data interaction system, the train's data combination is determined based on the detection of the data interaction system, and the dynamic length of the train is determined based on the identification of the data combination. Determining the UWB ranging length based on positioning data from the vehicle's front and rear UWB tags also includes: Based on the dynamic detection of the train, the on-board UWB tags at the front and rear of the train are determined, and the positioning data of the on-board UWB tags at the front and rear of the train are marked. The UWB ranging length is determined based on the positioning data of the on-board UWB tags at the front and rear of the train.

8. The dynamic positioning method for trains according to claim 1, characterized in that, The UWB ranging accuracy is determined based on the difference between the UWB ranging length and the dynamic train length. Simultaneously, historical error data of the train is collected, and a corresponding error control model is constructed based on this historical error data and the train's current operating scenario. Based on this error control model, the UWB ranging accuracy, and the train's real-time position data, the train's dynamic positioning data is determined, including: The UWB ranging length and the dynamic length of the train are collected. The difference between the UWB ranging length and the dynamic length of the train is determined by comparing them. The UWB ranging accuracy is determined based on this difference, the mapping relationship between UWB ranging accuracy and the train. Based on the train's historical error data, the corresponding driving scenarios are identified and marked. Based on the train's historical error data, the marked driving scenarios, and the train's current driving scenario, a corresponding error control model is constructed.

9. The dynamic positioning method for trains according to claim 8, characterized in that, The process of determining UWB ranging accuracy based on the difference between UWB ranging length and the dynamic train length, while simultaneously collecting historical error data of the train, constructing a corresponding error control model based on this historical error data and the train's current operating scenario, and determining the train's dynamic positioning data based on this error control model, UWB ranging accuracy, and the train's real-time position data, also includes: Collect UWB ranging accuracy and real-time train position data, determine dynamic compensation amount based on UWB ranging accuracy and error control model, and determine dynamic positioning data of train based on dynamic compensation amount and real-time train position data.

10. A dynamic positioning system for trains, characterized in that, The dynamic positioning system of the train is applied to the dynamic positioning method of the train as described in any one of claims 1-9.

Images