Track geometric parameter high-precision detection method and device

Abstract

The invention discloses a rail geometric parameter high-precision detection method and device, and the device comprises an optical fiber sensing module which comprises an optical fiber sensor laid on the surface of a steel rail and an optical signal demodulation device connected with the optical fiber sensor, and is used for collecting the strain distribution data and vibration data of the steel rail; the force detection module comprises a force sensor and a signal conditioning circuit which are mounted on a train wheel set and is used for collecting wheel-rail contact force data; the vibration monitoring module comprises an acceleration sensor and a data acquisition unit which are arranged on a train body and is used for acquiring train body vibration data; and the time synchronization module comprises a precise clock source and a timestamp marking unit and is used for performing time synchronization on data acquired by the optical fiber sensing module, the force detection module and the vibration monitoring module. Through cooperative use of the optical fiber sensing module and the data processing module, the effect of directly measuring the steel rail strain and avoiding integral error accumulation is achieved, and the problem that the detection precision is reduced along with distance increase in the existing inertial measurement technology is solved.

Description

Technical Field

[0001] This invention relates to the field of track inspection, specifically to a high-precision method and apparatus for detecting track geometric parameters. Background Technology

[0002] With the continuous expansion of my country's rail transit network and the sustained increase in train operating speeds, accurate detection of track geometry parameters is crucial for ensuring operational safety. Traditional track inspection mainly relies on manual inspections or large inspection vehicles, which suffer from problems such as long inspection cycles, poor real-time performance, and insufficient data continuity. Especially in the operating environment of high-speed railways, the track geometry dynamically evolves with factors such as train load and temperature changes, necessitating the development of technologies that can achieve real-time, continuous, and high-precision monitoring of track conditions.

[0003] Currently, the most advanced track inspection technologies primarily employ inertial measurement unit (IMU)-based systems. These systems calculate track geometry parameters by measuring vehicle acceleration. However, because IMUs need to perform integration calculations to convert acceleration data into displacement information, errors accumulate over long periods of operation. This causes the accuracy of track geometry parameter calculations to gradually decrease with increasing inspection distance. This error accumulation effect is particularly pronounced in sections with frequent train vibrations, severely impacting the reliability of the inspection results. Summary of the Invention

[0004] The purpose of this invention is to provide a high-precision detection method and apparatus for track geometry parameters to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A high-precision detection method for orbital geometric parameters includes the following steps:

[0007] Step S1: Collect strain distribution data and vibration data of the rail using fiber optic sensors laid on the rail.

[0008] Step S2: Collect wheel-rail contact force data using force sensors installed on the train wheelsets;

[0009] Step S3: Collect vehicle vibration data using acceleration sensors installed on the train body;

[0010] Step S4: Synchronize the data collected in steps S1 to S3 in time.

[0011] Step S5: Based on the strain distribution data, calculate the longitudinal and lateral displacements of the rail using a finite element inversion algorithm.

[0012] Step S6: Identify the characteristic frequencies of track irregularities based on the vibration data;

[0013] Step S7: Correct the displacement calculated in step S5 using the wheel-rail contact force data;

[0014] Step S8: Calculate the gauge change rate and track irregularity parameters based on the corrected displacement data.

[0015] Preferably, in step S1, the fiber optic sensor employs a Bragg grating array or distributed fiber optic sensing technology, wherein:

[0016] When using a Bragg grating array, the grating spacing is no more than 10 cm, and the strain measurement accuracy is no less than 1 με.

[0017] When using distributed fiber optic sensing, the spatial resolution should be no less than 0.5m and the sampling frequency should be no less than 100Hz.

[0018] Preferably, in step S2, the force sensor is a spoke-type strain sensor or a piezoelectric force sensor, and during installation:

[0019] The horizontal distance between the center of the sensor and the wheel-rail contact point shall not exceed 5% of the wheel diameter;

[0020] The angle between the sensor axis and the direction of the wheel-rail contact force shall not exceed 3°.

[0021] Preferably, in step S4, the time synchronization method adopts any of the following:

[0022] Hardware timestamp synchronization based on the IEEE 1588 precision time protocol, with a synchronization accuracy of over 100ns;

[0023] Software time synchronization based on GPS / BeiDou satellite time synchronization has a synchronization accuracy of more than 1ms;

[0024] The time-division multiplexing synchronization technology based on optical fiber transmission has a synchronization accuracy of over 10μs.

[0025] Preferably, in step 5, the finite element inversion algorithm specifically includes:

[0026] Establish a three-dimensional finite element model of the rail-fastener-sleeper system;

[0027] The strain-displacement inverse problem is solved using the Tikhonov regularization method;

[0028] The matrix solution process is accelerated by using the Lanczos iterative method.

[0029] Preferably, in step S6, when identifying the characteristic frequency of track irregularities:

[0030] A short-time Fourier transform is performed on the vibration data. The time window length is dynamically adjusted according to the current train speed v. The window length T = 0.5 / v, where L is the typical non-roughness wavelength.

[0031] Wavelet packet decomposition was used to extract energy features in the 1-100Hz frequency band;

[0032] Corrugation and low-joint defects were identified using a support vector machine classifier.

[0033] Preferably, in step S7, the correction process includes:

[0034] Establish the transfer function H(s) = U(s) / F(s) between the wheel-rail contact force F and the rail displacement u;

[0035] Update the transfer function parameters online using the recursive least squares method;

[0036] When the residual between the force sensor data and the displacement prediction value exceeds a threshold, the displacement result is recalculated.

[0037] Preferably, in step S8, when calculating the track irregularity parameters:

[0038] The gauge change rate is calculated using the standard deviation of a moving window, with the window length corresponding to 5 times the sleeper spacing.

[0039] Cubic spline interpolation is applied to the unevenness amplitude to ensure that the spatial sampling interval is no greater than 0.25m;

[0040] Based on the TB / T 3355-2014 standard, a graded early warning system is implemented for over-limit sections.

[0041] A high-precision detection device for track geometry parameters, comprising:

[0042] The fiber optic sensing module includes a fiber optic sensor laid on the surface of the rail and an optical signal demodulation device connected thereto, used to collect strain distribution data and vibration data of the rail.

[0043] The force detection module includes force sensors and signal conditioning circuits installed on the train wheelsets, used to collect wheel-rail contact force data;

[0044] The vibration monitoring module includes an acceleration sensor and a data acquisition unit installed on the train body to acquire vibration data of the train body;

[0045] The time synchronization module includes a precision clock source and a timestamp unit, used to synchronize the data collected by the fiber optic sensing module, force detection module and vibration monitoring module in time.

[0046] The data processing module includes:

[0047] The displacement calculation unit is configured to execute the finite element inversion algorithm to calculate the longitudinal and lateral displacements of the rail based on the strain distribution data.

[0048] The frequency analysis unit is configured to analyze vibration data to identify characteristic frequencies of track irregularities.

[0049] The data correction unit is configured to correct the displacement calculation results using wheel-rail contact force data;

[0050] The parameter calculation unit is configured to calculate the gauge change rate and track irregularity parameters based on the corrected displacement data.

[0051] Compared with the prior art, the beneficial effects of the present invention are:

[0052] 1. This invention achieves the effect of directly measuring rail strain and avoiding the accumulation of integral errors by using a fiber optic sensing module and a data processing module in combination, thus solving the problem that the detection accuracy of existing inertial measurement technology decreases with increasing distance;

[0053] 2. This invention achieves the effect of multi-source data collaborative optimization by using a time synchronization module and a data correction unit in combination, thus solving the problem of insufficient reliability of single sensor data under complex vibration environments. Attached Figure Description

[0054] Figure 1 This is a diagram of the fiber-optic-mechanical collaborative sensing system architecture of the present invention;

[0055] Figure 2 This is a flowchart of the dynamic calculation method for orbital parameters of the present invention;

[0056] Figure 3 This is a schematic diagram of an embodiment of the fiber optic sensing module of the present invention;

[0057] Figure 4 This is an installation configuration diagram of the wheelset force detection module of the present invention;

[0058] Figure 5 This is a timing diagram of the time synchronization mechanism based on IEEE 1588v2 of the present invention. Detailed Implementation

[0059] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] Example 1

[0061] This invention addresses the problems of dynamic error accumulation, insufficient synchronization accuracy of multi-source data, and poor adaptability to complex working conditions in track geometry parameter detection. It achieves high-precision dynamic monitoring of track status through distributed fiber optic sensing and multimodal data fusion technology. Since it involves an electromechanical integrated measurement system, the following technical objectives must be achieved based on the hardware architecture:

[0062] Objective 1: To achieve μm-level accuracy measurement of rail strain and vibration signals;

[0063] Objective 2: Ensure μs-level time synchronization of multi-source heterogeneous data;

[0064] Objective 3: To establish a multiphysics coupling model of strain, displacement, and wheel-rail forces.

[0065] Based on the above technical requirements, this embodiment adopts the following... Figure 1 The fiber-mechanical collaborative sensing architecture shown:

[0066] The sensing and acquisition layer consists of high-temperature resistant optical fibers (withstanding temperatures up to 300℃) laid longitudinally along the rail and wheelset force sensors. The optical fibers are fixed to both sides of the rail web using a V-groove encapsulation process, with a spacing of 50mm, and are connected to the demodulator via armored optical cables. The force sensors are mounted on the inside of the wheel axle via flanges, maintaining an eccentricity of 15±1mm with the wheel-rail contact point.

[0067] The signal processing layer contains a three-channel data acquisition unit, wherein:

[0068] Fibre Channel employs balanced detection technology, achieving a signal-to-noise ratio ≥60dB;

[0069] The force signal channel is equipped with a charge amplifier, with an adjustable gain range of 60-100dB;

[0070] The vibration channel has a built-in anti-aliasing filter with a cutoff frequency of 1kHz.

[0071] The computing control layer, which uses the Xilinx Zynq UltraScale+ MPSoC chip, integrates:

[0072] Real-time processing kernel, used for finite element inversion calculations;

[0073] The application processing core is used to run the disease identification algorithm;

[0074] Programmable logic is used to implement hardware-level time synchronization.

[0075] like Figure 2 As shown, the specific process of the dynamic orbital parameter calculation method executed on this architecture includes:

[0076] Step S1, strain-displacement conversion calculation:

[0077] A three-dimensional finite element model of the rail is established, and the displacement field is solved using the strain energy principle:

[0078]

[0079] Where D is the elasticity matrix and f is the nodal load vector. Tikhonov regularization is used to handle the ill-conditioned problem, with regularization parameter α = 0.1||K||2, where K is the stiffness matrix.

[0080] Step S2, vibration feature extraction:

[0081] Wavelet packet decomposition is performed on the acceleration signal a(t):

[0082]

[0083] in, Represented as db4 wavelet basis functions, with a decomposition level j=6, and a frequency band energy E j,k As feature vectors, input them into the SVM classifier.

[0084] Step S3, Multi-source data fusion:

[0085] Establish the force-displacement transfer function:

[0086]

[0087] The parameters are updated online using the recursive least squares method, with a forgetting factor λ = 0.95. Recalibration is triggered when the residual ||F-Hu|| > 0.05F_max.

[0088] Step S4, Comprehensive Parameter Evaluation:

[0089] The rate of change of track gauge is calculated using the sliding standard deviation:

[0090]

[0091] The window length N corresponds to 5 times the sleeper spacing (2.5m), and the over-limit threshold is dynamically adjusted according to σ(n)=1.2+0.003v (v is the vehicle speed, km / h).

[0092] A high-precision detection device for track geometry parameters, comprising:

[0093] The fiber optic sensing module includes a fiber optic sensor laid on the surface of the rail and an optical signal demodulation device connected thereto, used to collect strain distribution data and vibration data of the rail.

[0094] like Figure 3 As shown, specifically, the fiber optic sensing module can be implemented using two optional schemes:

[0095] Option 1: An FBG sensor array is laid longitudinally along the web of the rail. Ultraviolet laser writing technology is used to fabricate Bragg gratings with a wavelength spacing of 0.5 nm on single-mode optical fibers. The center-to-center spacing of each grating is set to 8 cm, and a wavelength demodulator (accuracy ±1 pm) is used to monitor the center wavelength offset of each grating in real time. During strain measurement, the strain value is calculated using the formula ΔλB / λB=(1-pe)ε, where pe is the effective elastic-optical coefficient, typically 0.22.

[0096] Option 2: Lay corrosion-resistant special single-mode optical fiber along the centerline of the rail base, using... The system performs monitoring. It is equipped with a 1550nm tunable laser (linewidth <1kHz) with a pulse width of 50ns, corresponding to a spatial resolution of 0.5m. The receiver uses a balanced detector (1GHz bandwidth) to acquire backscattered Rayleigh signals, and calculates strain changes using a cross-correlation algorithm, achieving a strain resolution of ±2με.

[0097] The force detection module includes force sensors and signal conditioning circuits installed on the train wheelsets, used to collect wheel-rail contact force data;

[0098] like Figure 4 As shown, the specific implementation process of the spoke-type strain sensor is as follows:

[0099] Four sets of 120Ω metal foil strain gauges are attached to the spokes of the train wheelset to form a full-bridge circuit. The horizontal distance between the sensor installation position and the wheel-rail contact point is controlled within 3% of the wheel diameter (for a 920mm wheel diameter, the installation radius deviation is <27.6mm). The signal conditioning circuit includes an instrumentation amplifier (1000x gain) and a 24-bit Σ-Δ ADC, with a sampling rate set to 1kHz.

[0100] Alternatively, a quartz triaxial force sensor (sensitivity 4 pC / N) can be embedded in the wheel axle bearing housing, and the force signal can be output through a charge amplifier (conversion factor 1 V / mN). The installation angle deviation is controlled within 1° using a laser level.

[0101] The vibration monitoring module includes an acceleration sensor and a data acquisition unit installed on the train body to acquire vibration data of the train body;

[0102] Specifically, a three-axis MEMS accelerometer (range ±50g, bandwidth 1kHz) is installed at both the front and rear ends of the bogie frame, with a sampling frequency set to 2kHz. The data acquisition unit uses an anti-aliasing filter (cutoff frequency 800Hz) and a 16-bit synchronous sampling ADC, and transmits data via a CAN bus.

[0103] The time synchronization module includes a precision clock source and a timestamp unit, used to synchronize the data collected by the fiber optic sensing module, force detection module and vibration monitoring module in time.

[0104] like Figure 5 As shown, specifically, the precision clock source uses an OCXO temperature-controlled crystal oscillator (stability ±0.01ppm), and synchronization of each node is achieved through the PTP protocol (IEEE 1588v2). The timestamp marking unit uses an FPGA to implement hardware-level time marking, with synchronization errors controlled within 80ns. The backup solution uses a BeiDou-3 receiver (1PPS accuracy 30ns) as the master clock source.

[0105] The data processing module includes:

[0106] The displacement calculation unit is configured to execute the finite element inversion algorithm to calculate the longitudinal and lateral displacements of the rail based on strain distribution data. Specifically, when establishing the finite element model of the rail, eight-node hexahedral elements are used for meshing, with element size not exceeding 5cm. Material parameters are set as follows: elastic modulus 210GPa, Poisson's ratio 0.3. The regularization parameter α is determined using the L-curve method, and the iteration termination condition is set to residual norm <0.1%.

[0107] The frequency analysis unit is configured to analyze vibration data to identify characteristic frequencies of track irregularities. Specifically, the short-time Fourier transform uses a Hanning window, with the window length T dynamically adjusted according to the formula T = 0.5L / v, where L is taken as the typical irregularity wavelength of 0.5m. Wavelet packet decomposition uses the db4 wavelet basis with 6 decomposition layers. The SVM classifier uses the RBF kernel function, and the training samples contain vibration spectra of 200 known track defects.

[0108] The data correction unit is configured to correct the displacement calculation results using wheel-rail contact force data. Specifically, the transfer function is modeled as a second-order system H(s)=ωn^2 / (s^2+2ζωns+ωn^2), with initial parameters ωn=200rad / s and ζ=0.3. The forgetting factor for the recursive least squares method is set to 0.98, and the residual threshold is taken as 5% of the measured force value.

[0109] The parameter calculation unit is configured to calculate the gauge change rate and track irregularity parameters based on the corrected displacement data. Specifically, when calculating the standard deviation of the moving window, the window length is set to 2.5m (corresponding to the spacing of 5 sleepers). The node spacing for cubic spline interpolation is set to 0.2m. The criteria for exceeding limits are based on Class III standards of TB / T 3355-2014: gauge change rate > 1.2mm / 2.5m, and unevenness > 6mm / 10m.

[0110] The above modules are interconnected via industrial Ethernet and adopt a layered chassis structure:

[0111] The bottom layer is the sensor interface layer, which includes fiber optic couplers, signal conditioning circuits, etc.

[0112] The middle layer is the data processing layer, configured with a multi-core processor (clock frequency ≥ 2GHz) and an FPGA accelerator. The upper layer is the display control layer, providing a touch screen human-machine interface.

[0113] The system is powered by a redundant 24VDC power supply, key components are equipped with UPS backup power supplies, all cables are shielded twisted-pair cables, and electromagnetic compatibility meets the EN 50121 standard.

[0114] In this embodiment, the fiber optic sensing module collects strain distribution data in real time through optical fibers laid on the rails, while the force detection module collects wheel-rail contact force data, the vibration monitoring module acquires vehicle vibration information, and the time synchronization module performs precise time alignment of the data collected by each module. The data processing module first converts the strain data into rail displacement through the built-in displacement calculation unit, then identifies the vibration spectrum characteristics through the frequency analysis unit, and then the data correction unit dynamically corrects the displacement using the wheel-rail force data. Finally, the parameter calculation unit outputs accurate gauge change rate and track irregularity parameters, realizing real-time monitoring of the track condition.

[0115] Example 2

[0116] Assuming a CRH380AL train passes through a curved section at a speed of 250 km / h, the system detection process is as follows:

[0117] Step 1: The fiber optic sensing module detects periodic strain fluctuations on the outer side of the rail web, with wavelength λ = 0.42 m and amplitude Δε = 85 με. Based on the strain-displacement conversion formula:

[0118]

[0119] The peak lateral displacement u was calculated. max =2.1mm ( p is a shape function e =0.22).

[0120] Step 2: Vibration monitoring showed a sudden increase in energy in the 315Hz frequency band, with wavelet packet energy accounting for 42%, exceeding the normal threshold of 15%. Combined with the SVM classifier, it was determined to be a wave-wrinkle disease (classification confidence level 92%).

[0121] Step 3: The wheel-rail force sensor measures the impact force F. peak =152KN, through the transfer function:

[0122]

[0123] After correction, the displacement was reduced to 1.8 mm, and the error was reduced by 14.3%.

[0124] Step four: The comprehensive assessment shows that the gauge change rate σ of this section is 1.35mm / 2.5m, which exceeds the dynamic threshold of 1.28mm. The system triggers a level two alarm and marks the location of the defect.

[0125] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0126] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A high-precision detection method for track geometric parameters, characterized in that, Includes the following steps: Step S1: Collect strain distribution data and vibration data of the rail using fiber optic sensors laid on the rail. Step S2: Collect wheel-rail contact force data using force sensors installed on the train wheelsets; Step S3: Collect vehicle vibration data using acceleration sensors installed on the train body; Step S4: Synchronize the data collected in steps S1 to S3 in time. Step S5: Based on the strain distribution data, calculate the longitudinal and lateral displacements of the rail using a finite element inversion algorithm. Step S6: Identify the characteristic frequencies of track irregularities based on the vibration data; Step S7: Correct the displacement calculated in step S5 using the wheel-rail contact force data; Step S8: Calculate the gauge change rate and track irregularity parameters based on the corrected displacement data.

2. The high-precision detection method for track geometric parameters according to claim 1, characterized in that: In step S1, the fiber optic sensor employs a Bragg grating array or distributed fiber optic sensing technology, wherein: When using a Bragg grating array, the grating spacing is no more than 10 cm, and the strain measurement accuracy is no less than 1 με. When using distributed fiber optic sensing, the spatial resolution should be no less than 0.5m and the sampling frequency should be no less than 100Hz.

3. The high-precision detection method for track geometric parameters according to claim 1, characterized in that: In step S2, the force sensor is a spoke-type strain sensor or a piezoelectric force sensor. During installation: The horizontal distance between the center of the sensor and the wheel-rail contact point shall not exceed 5% of the wheel diameter; The angle between the sensor axis and the direction of the wheel-rail contact force shall not exceed 3°.

4. The high-precision detection method for track geometric parameters according to claim 1, characterized in that: In step S4, the time synchronization method can be any of the following: Hardware timestamp synchronization based on the IEEE 1588 precision time protocol, with a synchronization accuracy of over 100ns; Software time synchronization based on GPS / BeiDou satellite time synchronization has a synchronization accuracy of more than 1ms; The time-division multiplexing synchronization technology based on optical fiber transmission has a synchronization accuracy of over 10μs.

5. The high-precision detection method for track geometric parameters according to claim 1, characterized in that: Step 5 of the finite element inversion algorithm specifically includes: Establish a three-dimensional finite element model of the rail-fastener-sleeper system; The strain-displacement inverse problem is solved using the Tikhonov regularization method; The matrix solution process is accelerated by using the Lanczos iterative method.

6. The high-precision detection method for track geometric parameters according to claim 1, characterized in that: In step S6, when identifying the characteristic frequency of track irregularities: A short-time Fourier transform is performed on the vibration data. The time window length is dynamically adjusted according to the current train speed v. The window length T = 0.5 / v, where L is the typical non-roughness wavelength. Wavelet packet decomposition was used to extract energy features in the 1-100Hz frequency band; Corrugation and low-joint defects were identified using a support vector machine classifier.

7. The high-precision detection method for track geometric parameters according to claim 1, characterized in that: In step S7, the correction process includes: Establish the transfer function H(s) = U(s) / F(s) between the wheel-rail contact force F and the rail displacement u; Update the transfer function parameters online using the recursive least squares method; When the residual between the force sensor data and the displacement prediction value exceeds a threshold, the displacement result is recalculated.

8. The high-precision detection method for track geometric parameters according to claim 1, characterized in that: In step S8, when calculating the track irregularity parameters: The gauge change rate is calculated using the standard deviation of a moving window, with the window length corresponding to 5 times the sleeper spacing. Cubic spline interpolation is applied to the unevenness amplitude to ensure that the spatial sampling interval is no greater than 0.25m; Based on the TB / T 3355-2014 standard, a graded early warning system is implemented for over-limit sections.

9. A high-precision detection device for track geometry parameters, characterized in that, include: The fiber optic sensing module includes a fiber optic sensor laid on the surface of the rail and an optical signal demodulation device connected thereto, used to collect strain distribution data and vibration data of the rail. The force detection module includes force sensors and signal conditioning circuits installed on the train wheelsets, used to collect wheel-rail contact force data; The vibration monitoring module includes an acceleration sensor and a data acquisition unit installed on the train body to acquire vibration data of the train body; The time synchronization module includes a precision clock source and a timestamp unit, used to synchronize the data collected by the fiber optic sensing module, force detection module and vibration monitoring module in time. The data processing module includes: The displacement calculation unit is configured to execute the finite element inversion algorithm to calculate the longitudinal and lateral displacements of the rail based on the strain distribution data. The frequency analysis unit is configured to analyze vibration data to identify characteristic frequencies of track irregularities. The data correction unit is configured to correct the displacement calculation results using wheel-rail contact force data; The parameter calculation unit is configured to calculate the gauge change rate and track irregularity parameters based on the corrected displacement data.

Images