A metro train speed measurement system, speed measurement equipment and speed measurement method based on multi-source information

Abstract

The application discloses a subway train speed measurement system, a speed measurement device and a speed measurement method based on multi-source information, a trackside perception layer, including N microwave sensors linearly arranged along a single track at a fixed interval D to cover a required monitoring section; a synchronization and data transmission layer, including a master clock and a slave clock based on an IEEE1588 precision time protocol, the master clock is synchronously arranged in a central processing module, provides a transmission channel with a determined bounded delay for phase data, and has a link redundancy capability; a vehicle-mounted self-consistent observation source, an auxiliary observation unit based on a visual odometer is arranged at the head and tail of each train, and a visual calculated speed of the train is obtained through image sequence analysis; the central processing module extracts a rigid translation speed by constructing a space-time matrix, separates deformation and vibration from the rigid translation speed, calculates a time when each bogie passes each microwave sensor by tracking the shape of components in the space-time matrix, and then outputs a final speed with a comprehensive confidence by speed cross verification.

Description

Technical Field

[0001] This invention belongs to the field of electronic information technology, specifically, it relates to a subway train speed measurement system and speed measurement equipment based on multi-source information. Background Technology

[0002] The existing subway train speed measurement technology system is mainly built around two paths: "direct / indirect speed measurement" and "position differential calculation". Its technology combination has become solidified and faces the following systemic bottlenecks caused by its principles and architecture: 1.1 Measurement techniques relying on the wheel-rail contact interface and their inherent problems Wheel and axle speed encoders: As the most widely used core sensor, they indirectly calculate speed by measuring the angular velocity of wheel and axle rotation. The fundamental flaw of this method is that its measurement reference is parasitic on the wheel-rail contact interface. In actual operation, factors such as wheel spin (during traction), skidding (during braking), wheel wear, and road surface contamination can all lead to significant deviations between the "measured wheel circumference speed" and the "actual center of gravity speed of the train," which is more pronounced in rainy or snowy weather or during emergency braking.

[0003] 1.2 Non-contact speed measurement technology and its application limitations Doppler radar: It emits microwaves towards the rail surface and receives the echoes, measuring speed through frequency shift. While it avoids wheel-rail contact problems, it is hampered by the complex electromagnetic reflection environment. Multipath effects in tunnels, water stains on the rail surface, and oil contamination cause changes in scattering characteristics, all of which can lead to signal fading and speed jumps.

[0004] Transponder + Odometer Estimation: This is the mainstream positioning and speed measurement mode. The transponder provides an absolute position reference, but the information is discontinuous; the odometer (usually an encoder) estimates the position between the transponders. This mode is highly dependent on the integrity of the transponder network and the initial accuracy of the encoder. Failure in any link (such as transponder loss or encoder initialization error) will lead to the accumulation of positioning errors.

[0005] 1.3 The Current State of Superficial Multi-Sensor Information Fusion Technology To improve reliability, current advanced solutions generally employ "multi-sensor fusion," but the fusion level typically remains at the shallow backend fusion stage, such as the data layer or decision layer. For example: Weighted average fusion: The outputs of encoders, radar, and inertial navigation systems (INS) are overlapped with fixed or adaptive weights. In essence, it seeks statistical optimality in unreliable data and cannot identify and eliminate the root cause of error at the physical level.

[0006] Loosely coupled filtering fusion: Algorithms such as Kalman filters are used to stitch together data from various sensors. However, the observation models of each sensor are independent of each other, lacking a unified physical model that can describe the actual motion state of the train as an "anchor point" for fusion. When multiple sensors simultaneously degrade in performance due to common reasons (such as severe weather), the fusion system will collectively fail, lacking an effective "gatekeeper" mechanism. Summary of the Invention

[0007] To address the shortcomings of existing technologies, the present invention aims to provide a subway train speed measurement system, speed measurement equipment, and speed measurement method based on multi-source information.

[0008] To achieve the aforementioned objectives, the technical solution adopted by this invention includes: A subway train speed measurement system based on multi-source information includes: The trackside sensing layer consists of N microwave sensors linearly deployed along one side of the track at a fixed spacing D (e.g., 3 meters) to cover the required monitoring area. Each microwave sensor has a built-in beam differential microwave transceiver module (capable of emitting two conical waves with a fixed small angle difference), a high-precision phase demodulation unit, a local slave clock, and a data preprocessing unit. The synchronization and data transmission layer employs a master-slave clock synchronization scheme based on the IEEE 1588 Precision Time Protocol (PTP). A master clock, deployed at the central processing module or the beginning of the line, distributes synchronization trigger signals to all microwave sensors in the trackside sensing layer via dedicated optical fiber. Each microwave sensor's local clock acts as a slave clock, receiving and locking onto the master clock, achieving strict synchronization of the entire array's sampling time. Preprocessed phase data is appended with a unified timestamp and uploaded via another independent or multiplexed optical fiber data bus. The master clock synchronization signal and data communication share the same TSN industrial Ethernet ring network. This network provides a transmission channel with a defined bounded delay for phase data and possesses link redundancy capabilities. The onboard self-consistent observation source deploys auxiliary observation units based on visual odometry at the head and tail of each train. Each unit includes a high-performance global shutter camera and processing module. By recognizing tunnel wall textures or pre-set QR code arrays along the trackside, and combining this with inertial measurement unit (IMU) data, it outputs visually estimated speed using a tightly coupled visual inertial odometry (VIO) algorithm. and confidence level The data is uploaded asynchronously via the 5G-R vehicle-to-ground wireless network; the onboard camera continuously images the preset optical features of the tunnel wall or trackside, and the visual speed of the train is obtained through image sequence analysis. The data is then uploaded in real time to the central processing module via vehicle-to-ground wireless communication (such as 5G-R) for asynchronous fusion and cross-verification with microwave sensing information.

[0009] The central processing module deploys an edge computing server cluster equipped with high-performance CPUs and GPUs, runs the core calculation algorithm, extracts the rigid translation velocity by constructing a spatiotemporal matrix, separates the deformation and vibration of the rigid translation velocity, calculates the time when each bogie passes each microwave sensor by tracking the shape of the components in the spatiotemporal matrix, and then outputs the final velocity with comprehensive confidence through velocity cross-validation.

[0010] A method for measuring subway train speed based on multi-source information, employing a subway train speed measurement system based on multi-source information, including: A1. Full array synchronous sampling and original phase field construction, describing the original phase field vector of the microscopic displacement on the one-dimensional contour line of the illuminated side of the train body; A2. Coordinate transformation from phase field to displacement field: Transform the beam directions of all sensing columns to obtain the micro displacement field. A3. The displacement field is deconstructed and partitioned for fine calculation and velocity fusion. The optimal continuous velocity of the carriage is calculated and weighted averaged to obtain the continuous velocity output based on the microwave displacement field and its comprehensive confidence level. A4. Virtual axis counting and reference speed generation; A5. Multi-source information fusion and final output; establishing a two-level fusion decision-making logic.

[0011] Furthermore, step A1 specifically includes: Each microwave sensor operates on a unified synchronization clock. At any given time, samples are taken from the surface area of ​​the train body under its responsibility, and the echo phase value of that sampling point is obtained. ,in For at any given moment Based on the sensor column number, the central processing module obtains an original phase field vector describing the microscopic displacement along the one-dimensional contour line of the illuminated side of the train body:

[0012] Furthermore, step A2 specifically includes: using the self-calibrated... Perform coordinate transformation:

[0013]

[0014] in, Let be the actual effective working angle of the i-th microwave sensing column. The wavelength of the continuous wave signal emitted by the microwave sensing column is [wavelength value]. In the first At the sampling time, the phase change of the echo signal measured by the i-th sensing column is... For the first At the sampling time, the line-of-sight displacement of the i-th sampling point on the train body surface along the microwave beam axis is... For the first At the sampling time, the microscopic displacement of the i-th sampling point on the train body surface in the horizontal direction (X-axis, i.e., the direction of train travel) of the track coordinate system is obtained; the microscopic displacement field after installation error compensation is obtained. .

[0015] Furthermore, step A3 specifically includes: A3.1 Coarse Velocity Estimation and Carriage Segmentation: Constructing the spatiotemporal matrix M. Performing a two-dimensional FFT on M, and detecting energy bands through Radon transform, yields a coarse estimate of the overall speed of the entire train. . use Transform M to the moving coordinate system and calculate the spatial curvature of the displacement field. , , representing the horizontal microscopic displacement of the train body surface at track position coordinate x and time t. It is the mathematical representation of the continuous field output from step A2 above. For the second-order partial derivative operator of the displacement field U in spatial dimension x, in calculus, the first-order derivative... The slope of the contour line, the second derivative This represents the curvature of the outline, i.e., the degree of bending of the shape. The larger the absolute value of the curvature, the more severe the bending at that point. By identifying points of abrupt curvature change, the initial segmentation of the carriages is completed, resulting in the sub-matrices of each carriage. ; A3.2 Partitioned Fine-grained Solution and Speed ​​Fusion: For each The displacement field decomposition model applicable to a single carriage is applied:

[0016] in This represents the composite displacement observation of the seg-th carriage at position x and time t in the onboard coordinate system. Let be the centroid translational velocity of the segth car (assuming it is a constant or slow variable within a short-time analysis window); The term represents the linearly increasing displacement caused by the uniform motion of the entire carriage. Let be the (small) pitch angle of the seg car at time t, which is a time-varying quantity caused by vehicle dynamics (such as going over a ramp, nose-nodding motion); The observation point position x and the position of the centroid of the carriage are given. The distance between them, and the pitch angle Multiplication represents the additional displacement at a point farthest from the center of mass due to the rotation of the carriage; The dynamic vibration field of the seg-th carriage, Let be the quasi-static deformation field of the seg-th carriage. It is about... The function describes the inherent elastic bending shape of the car body under the support of the bogie. In a coordinate system that moves with the car body, this shape does not change with time. A joint estimation filter (simultaneously estimating...) is used. , ,and (using the parameters), the optimal continuous speed estimate of the carriage is calculated. For all carriages A weighted average is performed (the weights can be calculated based on the residuals of each carriage) to obtain the continuous velocity output based on the microwave displacement field. and its overall confidence level .

[0017] Furthermore, step A4 specifically includes: For each carriage By finding the zero-crossing point (inflection point) of the second derivative of the isolated quasi-static deformation field, the bogie support point can be precisely located. Track these Time of each sensor column Discrete reference velocities based on absolute spatial intervals are calculated. This speed is only valid for the instant the bogie is identified, but it has extremely high absolute accuracy.

[0018] Furthermore, step A5 specifically includes: Calibration and Trust Assessment Level: When When valid, use it as a reference to calculate the microwave continuous velocity. and visual speed The instantaneous deviation and long-term drift error are dynamically updated to determine their respective online calibration parameters and confidence factors. , ; Adaptive fusion output stage: Under normal system conditions, the final speed is:

[0019] in, , The raw velocity estimates are from the microwave displacement field solution channel and the visual inertial odometry channel, respectively. , They are respectively and The instantaneous confidence level typically ranges from [0, 1]. , These are the long-term trust factors for the microwave channel and the visual channel, respectively, with a typical value range of (0, +∞), and the initial value can be set to 1; When the confidence level of a certain information source (such as vision in the textureless segment of the tunnel) is... If the trust factor T falls below a threshold, the system automatically switches to a different source mode or downgrades to a mode based on... The interpolation is performed in a conservative mode, and an alarm is triggered.

[0020] A subway train speed measurement device based on multi-source information, comprising: The trackside sensing layer consists of N microwave sensors linearly deployed along one side of the track at a fixed interval D, covering the required monitoring area; each microwave sensor has a built-in beam differential microwave transceiver module, a high-precision phase demodulation unit, a local slave clock, and a data preprocessing unit. The synchronization and data transmission layer includes a master clock and slave clock based on the IEEE 1588 precision time protocol. The master clock is synchronously deployed in the central processing module. The master clock signal and the data preprocessing unit share the same TSN industrial Ethernet ring network to provide a transmission channel with a defined bounded delay for phase data and have link redundancy capabilities. The vehicle-mounted self-consistent observation source deploys auxiliary observation units based on visual odometry at the head and tail of each train. It uses onboard cameras to continuously image preset optical features on the tunnel wall or trackside, obtains the visual speed of the train through image sequence analysis, and uploads it to the central processing module in real time through vehicle-to-ground wireless communication for asynchronous fusion and cross-verification with microwave sensing information. The central processing module deploys an edge computing server cluster equipped with high-performance CPUs and GPUs, runs the calculation algorithm, extracts the rigid translation velocity by constructing a spatiotemporal matrix, separates the deformation and vibration of the rigid translation velocity, calculates the time when each bogie passes each microwave sensor by tracking the shape of the components in the spatiotemporal matrix, and then outputs the final velocity with comprehensive confidence through velocity cross-validation.

[0021] Compared with the prior art, the advantages of the present invention include: (1) The present invention provides a subway train speed measurement system, speed measurement equipment and speed measurement method based on multi-source information, which transforms the measurement of derived quantities into the direct deconstruction of the physical field of the vehicle body displacement, thereby realizing non-contact, high-dimensional perception. (2) The present invention provides a subway train speed measurement system, speed measurement equipment and speed measurement method based on multi-source information, and proposes a two-stage processing framework of "coarse segmentation + partition fine calculation" and a car-level displacement field model including rigid body rotation, which is more in line with real physics. (3) The present invention provides a subway train speed measurement system, speed measurement equipment and speed measurement method based on multi-source information, and proposes a "dual-beam differential self-calibration" method to compensate for microwave sensor installation errors online, solving the core problem of large-scale engineering deployment; (4) The present invention provides a subway train speed measurement system, speed measurement equipment and speed measurement method based on multi-source information, establishing a high-precision discrete... The calibration mechanism for the "moving reference" drives and The trust assessment and adaptive integration are logically clear and highly fault-tolerant; (5) The present invention provides a subway train speed measurement system, speed measurement equipment and speed measurement method based on multi-source information, using PTP over TSN network to provide a physical basis for deterministic synchronous transmission of massive data and ensure the premise of core algorithm. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of a subway train speed measurement method based on multi-source information according to the present invention. Detailed Implementation

[0024] In view of the shortcomings of the prior art, the inventors of this invention, through long-term research and extensive practice, have proposed the technical solution of this invention. The technical solution, its implementation process, and principles will be further explained below with reference to the accompanying drawings and specific implementation examples in the embodiments of this application.

[0025] It should be noted that the embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. The described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, the present invention covers any substitutions, modifications, equivalent methods and solutions made on the spirit, principles and scope of the present invention as defined by the claims. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] In the description of this application, the terms "first," "second," "third," and similar words do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms "a" or "one," and similar words, do not indicate a quantity limitation, but rather indicate the presence of at least one. The terms "comprising" or "including," and similar words, mean that the elements or objects preceding "comprising" or "including" encompass the elements or objects listed following "comprising" or "including," and their equivalents, but do not exclude other elements or objects. The terms "connected" or "linked," and similar words, are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect.

[0027] In the description of this application, the terms "center," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used solely for the convenience of describing this application and for simplification, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. Furthermore, when using positional terms such as "both sides," "outer side," and "upper and lower," it should be understood that they are used only for ease of understanding and description, taking into account that the structure may be oriented to other positions.

[0028] In the description of this application, unless otherwise expressly specified and limited, the technical or scientific terms used shall have the ordinary meaning understood by a person with ordinary skills in the art to which this application pertains. Terms such as “installation,” “connection,” and “joining” shall be interpreted broadly, for example, as fixed connection, detachable connection, mating connection, or integral connection. For a person skilled in the art, the specific meaning of the above terms in this application can be understood according to the specific circumstances.

[0029] The present invention aims to introduce and explain the structural composition of a subway train speed measurement system, speed measurement equipment and speed measurement method based on multi-source information, as well as the cooperation relationship between the various components. Unless otherwise specified, the dimensions, materials and manufacturing processes of the various components in the subway train speed measurement system, speed measurement equipment and speed measurement method based on multi-source information in the embodiments of the present invention can be selected according to specific circumstances, and no special limitations or explanations are made here.

[0030] Furthermore, to provide the public with a better understanding of the present invention, certain specific details are described in detail in the following description of the invention. However, those skilled in the art will fully understand the invention even without these detailed descriptions.

[0031] Example 1

[0032] Please see Figure 1 A subway train speed measurement system based on multi-source information includes: The trackside sensing layer consists of N microwave sensors linearly deployed along one side of the track at a fixed spacing D (e.g., 3 meters) to cover the required monitoring area. Each microwave sensor has a built-in beam differential microwave transceiver module (capable of emitting two conical waves with a fixed small angle difference), a high-precision phase demodulation unit, a local slave clock, and a data preprocessing unit. The synchronization and data transmission layer employs a master-slave clock synchronization scheme based on the IEEE 1588 Precision Time Protocol (PTP). A master clock source, deployed at the central processing module or the beginning of the line, distributes synchronization trigger signals to all microwave sensors in the trackside sensing layer via dedicated optical fiber. Each microwave sensor's local clock acts as a slave clock, receiving and locking onto the master clock, achieving strict synchronization of the entire array's sampling time. Preprocessed phase data is appended with a unified timestamp and uploaded via another independent or multiplexed optical fiber data bus. The master clock source is deployed in the central processing module. The master clock synchronization signal and data communication share the same TSN industrial Ethernet ring network. This network provides a transmission channel with a defined bounded delay for phase data and possesses link redundancy capabilities. The vehicle-mounted self-consistent observation source deploys auxiliary observation units based on visual odometry at the head and tail of each train. Each unit contains a high-performance global shutter camera and processing module. By recognizing tunnel wall textures or pre-set QR code arrays along the trackside, and combining this with inertial measurement unit (IMU) data, it outputs visually estimated speed V_vis and confidence level C_vis using a tightly coupled visual inertial odometry (VIO) algorithm, which is asynchronously uploaded via a 5G-R vehicle-to-ground wireless network. The vehicle-mounted camera continuously images pre-set optical features of the tunnel wall or trackside, and the visually estimated speed of the train is obtained through image sequence analysis. This image is then uploaded in real-time to the central processing module via vehicle-to-ground wireless communication (such as 5G-R) for asynchronous fusion and cross-validation with microwave sensor information.

[0033] The central processing module deploys an edge computing server cluster equipped with high-performance CPUs and GPUs, runs the core calculation algorithm, extracts the rigid translation velocity by constructing a spatiotemporal matrix, separates the deformation and vibration of the rigid translation velocity, calculates the time when each bogie passes each microwave sensor by tracking the shape of the components in the spatiotemporal matrix, and then outputs the final velocity with high confidence through velocity cross-validation.

[0034] A method for measuring subway train speed based on multi-source information, employing a subway train speed measurement system based on multi-source information, includes: A1. Full-array synchronous sampling and original phase field construction: Describes the original phase field vector of the microscopic displacement along the one-dimensional contour line of the illuminated side of the train body; Step A1 specifically includes: Each microwave sensor operates on a unified synchronization clock. At any given time, samples are taken from the surface area of ​​the train body under its responsibility, and the echo phase value of that sampling point is obtained. , where i represents the value at any given time. Based on the sensor column number, the central processing module obtains an original phase field vector describing the microscopic displacement along the one-dimensional contour line of the illuminated side of the train body:

[0035] A2. Coordinate transformation from phase field to displacement field: Transform the beam directions of all sensing columns to obtain the microscopic displacement field; Step A2 specifically includes: using the self-calibrated... Perform coordinate transformation:

[0036]

[0037] in, The actual effective working angle of the i-th microwave sensing column is the angle between the central axis of the microwave beam emitted by the sensing column and the normal direction of the side wall (ideal vertical plane) of the train body that is irradiated. The wavelength of the continuous wave signal emitted by the microwave sensing column is determined based on the selected microwave frequency f, with the relationship λ = c / f, where c is the speed of light (approximately 3 × 10⁻⁶). 8 (m / s). For example, if the 24 GHz band is used, then λ ≈ 0.0125 meters; In the first At the sampling time, the phase change of the echo signal measured by the i-th sensing column is a time-varying observation. Its reference (zero point) is the static background echo phase when "no train is passing". .therefore, It directly reflects the minute change in the microwave propagation path length caused by the change in the position of the train body surface; For the first At the sampling time, the line-of-sight displacement of the i-th sampling point on the train body surface along the microwave beam axis is an intermediate variable calculated by the first formula. It represents the change in "slant distance" of the point on the train body surface along the beam illumination direction; For the first At the sampling time, the microscopic displacement of the i-th sampling point on the train body surface in the horizontal direction (X-axis, i.e., the direction of train travel) of the track coordinate system is the final observation calculated by the second formula. It represents the slant displacement. The result projected onto the horizontal direction constitutes the microscopic displacement field U( The i-th component of ) is obtained; the microscopic displacement field U( after installation error compensation) is obtained. ).

[0038] A3. Displacement field spatiotemporal evolution deconstruction and partitioned fine calculation and velocity fusion are performed to calculate the optimal continuous velocity of the carriage. A weighted average is then applied to obtain the continuous velocity output based on the microwave displacement field and its comprehensive confidence level. Step A3 specifically includes: A3.1 Coarse Velocity Estimation and Carriage Segmentation: Constructing the spatiotemporal matrix M. Performing a two-dimensional FFT on M, and detecting energy bands through Radon transform, yields a coarse estimate of the overall speed of the entire train. . use Transform M to the moving coordinate system and calculate the spatial curvature of the displacement field. , , representing the horizontal microscopic displacement of the train body surface at track position coordinate x and time t. It is the mathematical representation of the continuous field output from step A2 above. For the second-order partial derivative operator of the displacement field U in spatial dimension x, in calculus, the first-order derivative... The slope of the contour line, the second derivative This represents the curvature of the outline, i.e., the degree of bending of the shape. The larger the absolute value of the curvature, the more severe the bending at that point. By identifying points of abrupt curvature change, the initial segmentation of the carriages is completed, resulting in the sub-matrices of each carriage. ; A3.2 Partitioned Fine-grained Solution and Speed ​​Fusion: For each The displacement field decomposition model applicable to a single carriage is applied: in This represents the composite displacement observation of the seg-th carriage at position x and time t in the onboard coordinate system. Let be the centroid translational velocity of the segth car (assuming it is a constant or slow variable within a short-time analysis window); The term represents the linearly increasing displacement caused by the uniform motion of the entire carriage. Let be the (small) pitch angle of the seg car at time t, which is a time-varying quantity caused by vehicle dynamics (such as going over a ramp, nose-nodding motion); The observation point position x and the position of the centroid of the carriage are given. The distance between them, and the pitch angle Multiplication represents the additional displacement at a point farthest from the center of mass due to the rotation of the carriage; The dynamic vibration field of the seg-th carriage, Let be the quasi-static deformation field of the seg-th carriage. It is about The function describes the inherent elastic bending shape of the car body under the support of the bogie; this shape does not change with time in a coordinate system that moves with the car body. A joint estimation filter (simultaneously estimating...) is used. , ,and (using the parameters), the optimal continuous speed estimate of the carriage is calculated. For all carriages A weighted average is performed (the weights can be calculated based on the residuals of each carriage) to obtain the continuous velocity output based on the microwave displacement field. and its overall confidence level .

[0039] A4. Virtual axle counter and reference speed generation; for each car... By finding the zero-crossing point (inflection point) of the second derivative of the isolated quasi-static deformation field, the bogie support point can be precisely located. Track these Time of each sensor column Discrete reference velocities based on absolute spatial intervals are calculated. This speed is only valid for the instant the bogie is identified, but it has extremely high absolute accuracy.

[0040] A5. Multi-source information fusion and final output; establishing a two-level fusion decision-making logic: Calibration and Trust Assessment Level: When When valid, use it as a reference to calculate the microwave continuous velocity. and visual speed The instantaneous deviation and long-term drift error are dynamically updated to determine their respective online calibration parameters and confidence factors. , ; Adaptive fusion output stage: Under normal system conditions, the final speed is

[0041] in, , The raw velocity estimates are from the microwave displacement field solution channel and the visual inertial odometry channel, respectively. , They are respectively and The instantaneous confidence level typically ranges from [0, 1]. , These are the long-term confidence factors for the microwave and visual channels, respectively, typically ranging from (0, +∞), with an initial value of 1. The T factor is not a constant but is based on the channel velocity estimate and the "golden reference"—i.e., the virtual axis velocity. --The long-term compliance level is adjusted slowly. For example: When the speed of a certain channel With high precision The results consistently showed good agreement in multiple comparisons. The value will gradually increase, and it will have a higher weight in the fusion process.

[0042] If persistent deviation or Confidence level of the channel when it is valid But it is very low, The value will gradually decrease.

[0043] Function: The T factor introduces historical performance memory, enabling the system to "learn" and trust sensor sources that perform more stably and accurately over long-term operation, thereby suppressing the impact of single-point data mutations caused by transient interference on the fusion results.

[0044] Final speed yes and The weighted average is calculated. The weight of each velocity value is determined by its instantaneous confidence (C) and long-term confidence factor (T). The denominator is the normalization factor of the weights. When a source completely fails (e.g., C=0) or its long-term performance is extremely poor (T→0), the system will automatically and smoothly transition to a state of complete dependence on another source, achieving seamless degradation.

[0045] When the confidence level C or confidence factor T of a certain information source (such as vision in a textureless segment of a tunnel) falls below a threshold, the system automatically switches to a pure other information source mode, or degrades to a mode based on... The interpolation is performed in a conservative mode, and an alarm is triggered.

[0046] A subway train speed measurement device based on multi-source information, comprising: The trackside sensing layer consists of N microwave sensors linearly deployed along one side of the track at a fixed interval D, covering the required monitoring area; each microwave sensor has a built-in beam differential microwave transceiver module, a high-precision phase demodulation unit, a local slave clock, and a data preprocessing unit. The synchronization and data transmission layer includes a master clock and slave clock based on the IEEE 1588 precision time protocol. The master clock is synchronously deployed in the central processing module. The master clock signal and the data preprocessing unit share the same TSN industrial Ethernet ring network to provide a transmission channel with a defined bounded delay for phase data and have link redundancy capabilities. The vehicle-mounted self-consistent observation source deploys auxiliary observation units based on visual odometry at the head and tail of each train. It uses onboard cameras to continuously image preset optical features on the tunnel wall or trackside, obtains the visual speed of the train through image sequence analysis, and uploads it to the central processing module in real time through vehicle-to-ground wireless communication for asynchronous fusion and cross-verification with microwave sensing information. The central processing module deploys an edge computing server cluster equipped with high-performance CPUs and GPUs, runs the calculation algorithm, extracts the rigid translation velocity by constructing a spatiotemporal matrix, separates the deformation and vibration of the rigid translation velocity, calculates the time when each bogie passes each microwave sensor by tracking the shape of the components in the spatiotemporal matrix, and then outputs the final velocity with comprehensive confidence through velocity cross-validation.

[0047] This invention abandons the measurement of indirectly related physical quantities such as wheel speed and frequency shift, and instead directly measures the most fundamental physical state of the train body in space—displacement. By deploying a microwave sensing array, it acquires high-resolution microscopic displacement fields on the side of the train body in a non-contact manner. Speed ​​is no longer "measured," but rather a parameter "deconstructed" from this continuous spatiotemporal displacement field through physical laws. This fundamentally breaks free from the constraints of the wheel-rail contact interface and specific electromagnetic reflection surfaces.

[0048] The core capabilities of system perception and processing are shifted from the train to the track infrastructure. By deploying intelligent sensor networks along the trackside, a "smart track" is built that provides consistent, high-precision perception of any passing train. The role of onboard equipment (such as vision-assisted units) changes from "main force" to "auxiliary" and "verification," forming a train-track collaborative observation system. This architecture is easier to maintain and upgrade, and can provide services to all trains, exhibiting significant economies of scale.

[0049] A "train digital mirror" (i.e., a multi-rigid-body chain system model) is introduced as the center of information processing. All sensor data (microwave displacement field, visual data) are no longer raw materials for equal fusion, but rather observational inputs used to drive and update this digital mirror. The final output of the system (velocity, position, and even coupler state) is a physically consistent state simulated by this mirror. This allows the fusion process to be built on a solid physical foundation, possessing inherent consistency verification and fault diagnosis capabilities (when observational data cannot enable harmonious operation of the mirror, data or model anomalies can be determined).

[0050] In summary, this solution is not a simple combination or optimization of existing technologies such as encoders, radar, and vision, but rather proposes a novel speed measurement paradigm based on "trackside physical field perception and digital mirror calculation." It does not attempt to find a better alternative among existing unreliable measurement methods, but instead opens a third path—directly sensing the motion itself and performing calculations through first-principles calculations. The "source" of its "multi-source information" comes from observations of the motion of the same physical entity (the train itself) from different dimensions (microwave geometric changes, visual texture flow), which are deeply fused through a unified physical model. This differs from the post-processing correlation of indirect measurement results of different types or even different principles in existing technologies. Therefore, this solution possesses non-obviousness and significant substantial progress in its technical approach.

[0051] It should be understood that the above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. It should not be considered that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, several simple deductions or substitutions can be made without departing from the concept of the present invention. All equivalent changes or modifications made in accordance with the spirit and essence of the present invention should be covered within the protection scope of the present invention.

Claims

1. A subway train speed measurement system based on multi-source information, characterized in that: include: The trackside sensing layer consists of N microwave sensors linearly deployed along one side of the track at fixed intervals D, covering the required monitoring area. Each microwave sensor has a built-in beam differential microwave transceiver module, a high-precision phase demodulation unit, a local slave clock, and a data preprocessing unit; the trackside sensing layer is configured to perform full-array synchronous sampling of the train body surface area under a unified synchronous clock to obtain the echo phase value of each sampling point. The synchronization and data transmission layer includes a master clock and slave clock based on the IEEE 1588 precision time protocol. The master clock is synchronously deployed in the central processing module. The master clock signal and the data preprocessing unit share the same TSN industrial Ethernet ring network to provide a transmission channel with a defined bounded delay for phase data and have link redundancy capabilities. The vehicle-mounted self-consistent observation source deploys auxiliary observation units based on visual odometry at the head and tail of each train. It uses onboard cameras to continuously image preset optical features on the tunnel wall or trackside, obtains the visual speed of the train through image sequence analysis, and uploads it to the central processing module in real time through vehicle-to-ground wireless communication for asynchronous fusion and cross-verification with microwave sensing information. The central processing module is deployed with an edge computing server cluster equipped with high-performance CPUs and GPUs. The central processing module is configured to: construct an original phase field vector based on the echo phase values ​​obtained from each sampling point; perform coordinate transformation on the original phase field vector based on the self-calibrated actual effective working angle to obtain the micro displacement field after installation error compensation; construct a spatiotemporal matrix; perform coarse velocity estimation and car segmentation through two-dimensional FFT and Radon transform; apply a displacement field decomposition model to each car submatrix; use joint estimation filtering to solve for the optimal continuous velocity of each car; and obtain the continuous velocity output of the microwave displacement field and its comprehensive confidence level by weighted averaging; locate the bogie support point from the separated quasi-static deformation field; track the time when the support point passes through each microwave sensor; and calculate the reference velocity based on the absolute spatial interval. Based on the reference speed, the microwave continuous speed and visual speed are calibrated and trusted online, and the final speed is adaptively fused and output. When the confidence or trust factor of a certain source is lower than the threshold, the system switches to a pure other source mode and triggers an alarm.

2. A subway train speed measurement method based on multi-source information, applied to the speed measurement system as described in claim 1, characterized in that: include: A1. Full array synchronous sampling and original phase field construction, describing the original phase field vector of the microscopic displacement on the one-dimensional contour line of the illuminated side of the train body; A2. Coordinate transformation from phase field to displacement field: Transform the beam direction of all microwave sensors to obtain the micro displacement field. A3. The displacement field is deconstructed and partitioned for fine calculation and velocity fusion. The optimal continuous velocity of each carriage is calculated and weighted averaged to obtain the continuous velocity output based on the microwave displacement field and its comprehensive confidence level. A4. Virtual axis counting and reference speed generation, calculate discrete reference speeds based on absolute spatial intervals; A5. Multi-source information fusion and final output; Establish a two-level integrated decision-making logic.

3. The subway train speed measurement method based on multi-source information according to claim 2, characterized in that: Specifically, A1 includes: each microwave sensor operating on a unified synchronization clock. At any given time, samples are taken from the surface area of ​​the train body under its responsibility, and the echo phase value of that sampling point is obtained. , where i is the value at any time The microwave sensor number is used to obtain a raw phase field vector from the central processing module, which describes the microscopic displacement along the one-dimensional contour line of the illuminated side of the train body. .

4. The subway train speed measurement method based on multi-source information according to claim 3, characterized in that: Specifically, A2 includes: obtained using self-calibration. Perform coordinate transformation: in, Let be the actual effective working angle of the i-th microwave sensor. The wavelength of the continuous wave signal emitted by the microwave sensor. In the first At the sampling time, the phase change of the echo signal measured by the i-th microwave sensor. For the first At the sampling time, the line-of-sight displacement of the i-th sampling point on the train body surface along the microwave beam axis is... For the first At the sampling time, the microscopic displacement of the i-th sampling point on the train body surface in the horizontal direction of the track coordinate system is obtained; the microscopic displacement field after installation error compensation is obtained. .

5. The subway train speed measurement method based on multi-source information according to claim 4, characterized in that: A3 specifically includes: A3.1 Coarse Speed ​​Estimation and Carriage Segmentation: Construct a spatiotemporal matrix M, perform a two-dimensional FFT on M, and detect energy bands using Radon transform to obtain a coarse estimate of the overall speed of the entire train. ,use Transform M to the vehicle coordinate system and calculate the spatial curvature of the displacement field. For the displacement field U in the spatial dimension The second-order partial derivative operator on the first derivative, in calculus, is the first derivative. The slope of the contour line, the second derivative This represents the curvature of the outline, identifies points of abrupt curvature change, completes the initial segmentation of the carriages, and obtains the sub-matrices of each carriage. ; A3.2 Partitioned Fine-grained Solution and Speed ​​Fusion: For each The displacement field decomposition model applicable to a single carriage is applied: in This represents the composite displacement observation of the segth carriage at position x and time t in the onboard coordinate system. Let be the translational velocity of the center of mass of the seg-th car. This represents the linearly increasing displacement caused by the uniform motion of the entire carriage. Let be the pitch angle of the segth carriage at time t. Let x be the position of the observation point and the position of the centroid of the carriage. The distance between them The quasi-static deformation field of the segth carriage. For the dynamic vibration field of the segth carriage, a joint estimation filter is used, and the dynamic vibration field of the segth carriage is estimated simultaneously. , ,and Using the parameters, the optimal continuous speed estimate for the carriage is calculated. For all carriages A weighted average is performed to obtain the continuous velocity output based on the microwave displacement field. and its overall confidence level .

6. The subway train speed measurement method based on multi-source information according to claim 5, characterized in that: The A4 specifically includes: For each carriage By finding the zero-crossing point of the second derivative of the isolated quasi-static deformation field, the bogie support point can be precisely located. Track these Time of each microwave sensor Discrete reference velocities based on absolute spatial intervals are calculated. This speed is only valid for the instant the bogie is identified, but it has extremely high absolute accuracy.

7. The subway train speed measurement method based on multi-source information according to claim 6, characterized in that: A5 specifically includes: Calibration and Trust Assessment Level: When When valid, use it as a reference to calculate the microwave continuous velocity. and visual speed The instantaneous deviation and long-term drift error are dynamically updated to determine their respective online calibration parameters and confidence factors. , ; Adaptive fusion output stage: Under normal conditions, the final speed is: in, , The raw velocity estimates are from the microwave displacement field solution channel and the visual inertial odometry channel, respectively. , They are respectively and Instantaneous confidence level , These are the long-term trust factors for the microwave channel and the visual channel, respectively. When the confidence level C or trust factor T of a certain information source falls below a threshold, the system automatically switches to a different information source mode or downgrades to a mode based on... The interpolation is performed in a conservative mode, and an alarm is triggered.

8. A subway train speed measurement device based on multi-source information, used to implement the method described in any one of claims 2-7, characterized in that: include: The trackside sensing layer consists of N microwave sensors linearly deployed along one side of the track at fixed intervals D, covering the required monitoring area. Each microwave sensor has a built-in beam differential microwave transceiver module, a high-precision phase demodulation unit, a local slave clock, and a data preprocessing unit; the trackside sensing layer is configured to perform full-array synchronous sampling of the train body surface area under a unified synchronous clock to obtain the echo phase value of each sampling point. The synchronization and data transmission layer includes a master clock and slave clock based on the IEEE 1588 precision time protocol. The master clock is synchronously deployed in the central processing module. The master clock signal and the data preprocessing unit share the same TSN industrial Ethernet ring network to provide a transmission channel with a defined bounded delay for phase data and have link redundancy capabilities. The vehicle-mounted self-consistent observation source deploys auxiliary observation units based on visual odometry at the head and tail of each train. It uses onboard cameras to continuously image preset optical features on the tunnel wall or trackside, obtains the visual speed of the train through image sequence analysis, and uploads it to the central processing module in real time through vehicle-to-ground wireless communication for asynchronous fusion and cross-verification with microwave sensing information. The central processing module is deployed with an edge computing server cluster equipped with high-performance CPUs and GPUs. The central processing module is configured to: construct an original phase field vector based on the echo phase values ​​obtained from each sampling point; perform coordinate transformation on the original phase field vector based on the self-calibrated actual effective working angle to obtain the micro displacement field after installation error compensation; construct a spatiotemporal matrix; perform coarse velocity estimation and car segmentation through two-dimensional FFT and Radon transform; apply a displacement field decomposition model to each car submatrix; use joint estimation filtering to solve for the optimal continuous velocity of each car; and obtain the continuous velocity output of the microwave displacement field and its comprehensive confidence level by weighted averaging; locate the bogie support point from the separated quasi-static deformation field; track the time when the support point passes through each microwave sensor; and calculate the reference velocity based on the absolute spatial interval. Based on the reference speed, the microwave continuous speed and visual speed are calibrated and trusted online, and the final speed is adaptively fused and output. When the confidence or trust factor of a certain source is lower than the threshold, the system switches to a pure other source mode and triggers an alarm.

Images