A rail transit inspection device and an inspection method

Abstract

The application discloses a kind of track traffic inspection equipment and inspection method, belong to track detection technical field;Inspection equipment includes inspection car, which is equipped with mobile acoustic vibration acquisition array, mileage positioning unit, edge computing unit and wireless communication module, wheel-rail noise and vibration signal are synchronously collected in the process of driving inspection car;Edge computing unit utilizes vibration signal to carry out adaptive filtering noise reduction to acoustic signal, extracts track disease characteristics by time-frequency domain transformation and coherence analysis, determines fault type;When determining that there is a fault suspicion and abnormal sound source is located between sensors, time difference positioning algorithm is used, according to the time delay difference of at least two acoustic sensor signals and pre-stored sound velocity, the offset mileage of fault point relative to inspection car is accurately calculated, the absolute position of fault is obtained in combination with real-time position information, and warning information is sent to ground operation and maintenance center through wireless communication module.

Description

Technical Field

[0001] This invention relates to the field of track inspection technology, and in particular to an inspection device and corresponding inspection method for monitoring the health status of track structures. Background Technology

[0002] With the rapid development of urban rail transit and high-speed railways, ensuring the safe and stable operation of track structures has become a top priority in operation management. Under the long-term dynamic load of trains, tracks are prone to various defects such as rail wear, loose fasteners, sleeper cracks, and ballast bed settlement. If these defects are not detected and addressed in a timely manner, they will seriously threaten train operation safety.

[0003] Currently, track inspections are mainly divided into manual inspections, large flaw detection vehicle inspections, and online monitoring systems based on fixed sensors. Manual inspections are inefficient, highly subjective, and greatly affected by weather and lighting conditions; large flaw detection vehicles are expensive and require valuable "track windows" for operation, making routine inspections impossible; while fixed sensor networks can achieve real-time monitoring, they require the deployment of a large number of sensors along the entire line, resulting in a complex system, huge investment, and difficult maintenance. In recent years, passive monitoring methods based on noise and vibration have been proposed, but they still rely on pre-setting multiple fixed sound pressure sensing points along the track, failing to achieve mobile, low-cost, wide-area coverage.

[0004] Therefore, how to reduce system deployment costs and complexity while ensuring detection accuracy, and achieve flexible and efficient track inspection, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] (a) Technical problems to be solved In view of the above-mentioned deficiencies of the prior art, the present invention provides an inspection device and method for rail transit. One of the technical problems to be solved is to provide a mobile rail inspection device with high integration, flexible deployment and no need to pre-set a large number of ground sensors.

[0006] Another technical problem to be solved by the present invention is to provide an inspection method that can accurately locate, identify the type of defects and assess the level of defects along the track by using the acoustic and vibration signals collected by the mobile platform itself.

[0007] (II) Technical Solution To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: An inspection device for rail transit, used for traveling along a track, includes an inspection vehicle that can movably straddle an existing rail transit track. The inspection vehicle includes: a vehicle frame as a mounting base; a driving unit mounted on the vehicle frame for driving the inspection vehicle to move along the track; a mobile acoustic and vibration acquisition array mounted on the vehicle frame, including M acoustic sensors and N vibration sensors spaced apart along the direction of travel of the inspection vehicle, for acquiring wheel-rail noise signals and track structure vibration signals generated during the inspection vehicle's movement, where M and N are both integers greater than or equal to 2; a mileage positioning unit for acquiring the inspection vehicle's position information on the track in real time; and an edge computing unit electrically connected to the driving unit, the mobile acoustic and vibration acquisition array, and the mileage positioning unit, for controlling the inspection vehicle to travel at a preset speed and simultaneously receiving acoustic and vibration signals from the mobile acoustic and vibration acquisition array and position signals from the mileage positioning unit.

[0008] The edge computing unit integrates a data fusion and noise reduction module, a multi-source signal processing module, and a fault location module. The data fusion and noise reduction module uses the vibration signal collected by the vibration sensor as a reference to adaptively filter the acoustic signal collected by the acoustic sensor to remove background noise, obtaining a denoised acoustic signal, which is then provided to the multi-source signal processing module. The multi-source signal processing module performs time-frequency domain transformation and coherence analysis on the denoised multi-channel acoustic signal and the vibration signal collected by the vibration sensor to extract characteristic parameters reflecting the health status of the track structure. The fault location module, when an anomaly is determined to exist in the track and the abnormal sound source is located in the rail area between at least two of the acoustic sensors, uses the time difference of arrival (TDOA) positioning algorithm based on the position signal from the mileage positioning unit and the analysis results from the multi-source signal processing module. It calculates the positional offset of the abnormal sound source relative to the inspection vehicle based on the time delay difference of the signals from the at least two acoustic sensors and the pre-stored sound velocity, and combines this with real-time position information to obtain the absolute position of the abnormal sound source, generating a fault warning message.

[0009] Furthermore, the inspection vehicle also includes a wireless communication module, which is electrically connected to the edge computing unit and is used to send the early warning information generated by the edge computing unit to the ground operation and maintenance center.

[0010] Furthermore, the edge computing unit also includes a synchronization trigger module, used to send synchronization trigger signals to the mobile acoustic and vibration acquisition array and the odometer positioning unit to synchronize the acoustic and vibration signals with the position signals in time. Specifically, the synchronization trigger module can generate synchronization trigger pulses based on a high-precision clock synchronization protocol (such as the PTP protocol) to achieve multi-channel synchronous acquisition at the microsecond or even nanosecond level.

[0011] Preferably, the inspection vehicle straddles the track via walking wheels and guide wheels, and the track is made of C-shaped steel or I-beam steel.

[0012] Preferably, the acoustic sensor is a miniature microphone array arranged in a line or cross shape, pointing towards the top surface of the rail and the fastener area of ​​the track; the vibration sensor is an accelerometer, closely attached to the axle of the inspection vehicle or the vehicle frame, used to pick up the wheel-rail coupled vibration signal.

[0013] The present invention also provides a rail transit inspection method based on the above-mentioned inspection equipment, comprising the following steps: Step S1: Control the inspection vehicle to travel along the track at a preset speed.

[0014] Step S2: During the journey, the location information is acquired in real time through the odometer positioning unit, and at the same time, the time-synchronized multi-channel sound pressure signal is acquired through the mobile acoustic vibration acquisition array. and vibration acceleration signal .

[0015] Step S3: Process the acquired sound pressure signal through the data fusion and noise reduction module. and vibration acceleration signal Preprocessing is performed to obtain the effective acoustic signal after noise reduction. The preprocessing specifically includes utilizing the vibration acceleration signal. As a noise reference source, the sound pressure signal Perform adaptive filtering.

[0016] Step S4: For the effective acoustic signal and vibration acceleration signal Feature extraction and anomaly detection are performed, specifically including: performing coherence analysis on the noise-reduced multi-channel effective acoustic signals and vibration acceleration signals, calculating the coherence coefficient at different frequencies, and determining that the acoustic and vibration signals in a certain frequency band are effective feature signals when the coherence coefficient in a certain frequency band exceeds a preset threshold; performing envelope spectrum analysis and time-frequency domain transformation on the effective feature signals to extract the feature frequencies and energy distribution patterns representing specific track defects; comparing the extracted feature parameters with a pre-stored normal track feature parameter library, and if they exceed the normal range, determining that there is a suspected fault at the current mileage location and determining the fault type through feature matching.

[0017] Step S5: When a suspected fault is identified and the abnormal sound source is located in the rail area between at least two acoustic sensors at different spatial locations, a time-of-arrival (TOA) localization algorithm is used. Based on the time delay difference between the effective acoustic signals collected by at least two acoustic sensors, and combined with the pre-stored sound wave propagation speed in the rail, the positional offset of the abnormal sound source relative to the inspection vehicle is calculated. And, combined with the current mileage location, accurately calculate the absolute mileage location L of the fault point. fault Specifically, the effective acoustic signals of the two acoustic sensors are calculated. and The cross-correlation function is used to obtain the time delay difference. : ; When an abnormal sound source is determined to be located in the rail area between the two acoustic sensors, the position offset can be calculated based on the following formula. : ; in The pre-stored speed of sound wave propagation in the rail, Let be the time delay difference between any two sensors. Then the absolute mileage location of the fault point is: ; in The mileage location of the inspection vehicle when the anomaly was detected.

[0018] Step S6: Output early warning information containing the location and type of the fault. Specifically, the early warning information can be sent to the ground operation and maintenance center via the wireless communication module to generate a maintenance work order.

[0019] (III) Beneficial Effects Compared with the prior art, the inspection equipment and method for rail transit provided by the present invention have the following beneficial effects: Low cost and easy deployment: No fixed sensors need to be laid along the entire track. Only one inspection device is needed to achieve dynamic inspection of the entire line, which greatly reduces the system hardware cost and engineering construction difficulty.

[0020] High-precision, real-time positioning: By adopting a mobile acoustic and vibration acquisition array and TDOA positioning algorithm, combined with high-precision mileage positioning and microsecond-level time synchronization technology, the positioning accuracy of fault points can be improved to the meter level or even the sub-meter level, making it easier for maintenance personnel to quickly find and deal with problems.

[0021] Strong anti-interference capability: By combining adaptive filtering and coherence analysis of vibration signals, it can effectively eliminate background noise in complex environments and accurately extract weak feature signals reflecting wheel-rail conditions, thereby improving fault identification rate and accuracy.

[0022] All-weather, routine inspection: The inspection equipment can be designed to be lightweight and low-power, enabling it to perform frequent inspections either by following operating trains or on its own outside of designated maintenance windows, thus achieving real-time monitoring of track health status and aligning with the development trend of smart operation and maintenance. Attached Figure Description

[0023] Figure 1 This is a first-angle isometric view of the structure of the rail transit inspection equipment of the present invention. Figure 2 This is a second-angle isometric view of the structure of the rail transit inspection equipment of the present invention; Figure 3 This is an isometric view of the internal structure of the inspection vehicle of the rail transit inspection equipment of the present invention. Figure 4 This is a structural side view of the inspection equipment for rail transit according to the present invention.

[0024] [Explanation of markings in the attached image]: 11- Track beam; 12- Running surface; 2-Inspection vehicle; 21-Vehicle frame; 22-Walking wheels; 23-Guide wheels; 24-Walking drive unit; 31-Acoustic sensor; 32-Vibration sensor; 4-Obstacle positioning unit; 5-Edge computing unit; 6-Wireless communication module; Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] Example 1 Please see Figure 1 and Figure 2 This embodiment provides an inspection device for rail transit, which travels along C-shaped steel rails laid along the rail transit line. The bottom of the rail is a rail beam 11, which is used to fix it to the tunnel wall or viaduct; the top of the rail has a horizontal walking surface 12.

[0027] The inspection vehicle 2 includes a closed vehicle frame 21. Multiple wheels 22 are symmetrically arranged on both sides of the bottom of the vehicle frame 21, and these wheels roll and support themselves on the track surface 12. Multiple guide wheels 23 are also arranged on the sides of the vehicle frame 21, making rolling contact with the sides of the track to ensure the stability of the inspection vehicle 2 when turning. One of the wheels 22 is connected to a drive unit 24, serving as the drive wheel. The drive unit 24 includes a DC servo motor and a matching driver, enabling precise speed control.

[0028] At the bottom of the vehicle frame 21, and on both the front and rear sides of the running wheels 22, a mobile acoustic and vibration acquisition array is installed. This array includes four linearly arranged miniature microphones as acoustic sensors 31, and two accelerometers mounted on different axles of the running wheels 22 as vibration sensors 32. The acoustic sensors 31 are aligned with the contact area between the running wheels 22 and the track surface 12 to collect rolling noise generated by the wheel-rail interaction. The vibration sensors 32 are used to collect the vertical and lateral vibration accelerations of the running wheels 22.

[0029] The vehicle frame 21 integrates a mileage positioning unit 4, an edge computing unit 5, and a wireless communication module 6. The mileage positioning unit 4 is connected to the axle of the driving wheels 22, calculating the real-time mileage and position of the inspection vehicle 2 by recording the number of revolutions, with a resolution of up to 1 mm. The edge computing unit 5 is electrically connected to the driving unit 24, acoustic sensor 31, vibration sensor 32, mileage positioning unit 4, and wireless communication module 6. The edge computing unit 5 integrates a data fusion and noise reduction module, a multi-source signal processing module, a fault location module, and a synchronization triggering module. This edge computing unit 5 can use a high-performance embedded industrial computer, such as the NVIDIA Jetson AGX Orin series, to meet the requirements of real-time signal processing and algorithm computation.

[0030] The synchronization trigger module generates synchronization trigger pulses based on the IEEE 1588PTP protocol and broadcasts them to the acquisition nodes of each acoustic sensor 31, vibration sensor 32, and odometer positioning unit 4 via an Ethernet interface. This ensures that the sampling clock edges of all channels are strictly aligned, keeping the time synchronization error between the acoustic / vibration signals and the position signals stably controlled within 1 microsecond, providing a time reference for subsequent TDOA high-precision positioning. The data fusion and noise reduction module receives the raw signals from the acoustic sensor 31 and vibration sensor 32 and performs adaptive filtering and noise reduction processing to obtain the denoised acoustic signal. The multi-source signal processing module performs coherence analysis, time-frequency domain transformation, and feature extraction on the denoised signal. The fault location module executes the TDOA positioning algorithm to determine the location of the fault point. The wireless communication module 6 uses a 4G / 5G or WiFi module for data interaction with the ground operation and maintenance center.

[0031] Example 2 This embodiment is based on the equipment described in Embodiment 1, and elaborates in detail the rail transit inspection method performed by it.

[0032] Step S101: Task Initiation. The ground operation and maintenance center sends an inspection command to the inspection vehicle 2 via the wireless network. After receiving the command, the edge computing unit 5 controls the walking drive unit 24 to start, driving the inspection vehicle 2 to travel at a constant speed of 15km / h along the track.

[0033] Step S102: Data Synchronization Acquisition. During driving, the synchronization trigger module sends a synchronization trigger signal to all sensors and the odometer positioning unit 4 at a frequency of 25.6 kHz. The odometer positioning unit 4 reports its position in real time by sending one pulse signal per revolution. The four acoustic sensors 31 synchronously acquire sound pressure signals at a sampling rate of 25.6 kHz. Two vibration sensors 32 synchronously acquire vibration acceleration signals at the same sampling rate. , .

[0034] Step S103: Preprocessing and Denoising. The data fusion and denoising module uses an adaptive filtering algorithm based on minimum mean square error (LMS) to process the vibration signal. As a reference input, the sound pressure signal Filtering is performed to remove noise components related to environmental vibrations, resulting in a noise-reduced effective acoustic signal. This step can improve the signal-to-noise ratio by more than 15dB.

[0035] Step S104: Anomaly Detection and Type Identification. The multi-source signal processing module extracts a signal window of approximately 1 second in length, with a window overlap rate of 50%. Calculate the signal within this window... and The coherence coefficient within the 1kHz-8kHz frequency band. The preset threshold was determined to be 0.7 through statistical analysis of 100km of normal track data. When the coherence coefficient exceeds 0.7, the acoustic-vibration signal in this frequency band is considered a valid characteristic signal.

[0036] Subsequently, short-time Fourier transform (STFT) and envelope spectrum analysis were performed on the effective signal segment to extract its characteristic frequencies and energy distribution patterns. The extracted characteristic parameters were compared with a pre-stored database of normal track characteristic parameters. If they exceeded the normal range, a fault was suspected at the current mileage location, and the fault type was determined by matching them with a fault feature template. The characteristic frequency ranges of common track defects are: rail corrugation 500Hz-2kHz, ballast bed unsupported 200Hz-500Hz, sleeper cracks 3kHz-6kHz, rail joint defects 100Hz-300Hz, and fastener elastic clip breakage 2kHz-3kHz.

[0037] In this embodiment, the calculation results show that at mileage K120+352, the coherence coefficient is as high as 0.92, which is far higher than the threshold of 0.7, indicating that the acoustic signal at this location is valid and the energy is abnormally concentrated. Its spectral energy is mainly concentrated around 2.5kHz, which highly matches the pre-stored characteristic frequency template of "fastener spring breakage". Therefore, it is determined that there is a fastener spring breakage fault near mileage K120+352.

[0038] Step S105: Precise location. The fault location module extracts the time of the fault. Effective acoustic signals from the first sensor (front end) and the fourth sensor (back end) and Since the abnormal sound source was determined to be located in the track section between the two sensors, the one-dimensional TDOA positioning formula was directly applied. The cross-correlation function of the two signals was calculated to obtain the time delay difference. = -0.00025 seconds (the negative sign indicates the fault point is closer to the fourth sensor). The speed of sound in the rail is known. =5000m / s, sensor spacing =1.5 meters. Calculate the positional offset of the fault point relative to the array center: Combined with the mileage location of the inspection vehicle at this time =K120+352.5 meters, finally the absolute location of the fault point is calculated: =K120+352.5+(-0.625)=K120+351.875 meters.

[0039] It should be noted that the above positioning calculation is based on the determination that the sound source is located in the region between the lines connecting the two acoustic sensors 31. When the sound source is determined to be outside the two sensors based on the positive or negative characteristics of the time delay difference or the signal energy distribution, the calculation of the position offset ΔL should combine the azimuth information of the sound source relative to the sensor array. Specifically, a one-dimensional hyperbolic equation system about the sound source position can be established based on the combination of time delay differences of at least three acoustic sensors 31 arranged in a straight line, and the sound source position can be solved using the least squares method. This solution process is a mature technique in this field, and will not be elaborated further in this embodiment.

[0040] As an alternative, when the sound source is determined to be located in the area between the two sensors, the simplified one-dimensional positioning formula described in this embodiment is used for rapid positioning; when the sound source is determined to be located outside the sensors, the offset is determined by a multi-sensor joint solution method based on hyperbolic equations.

[0041] Step S106: Early Warning Reporting. Edge computing unit 5 packages early warning information such as "Fault Type: Fastener spring clip broken; Fault Level: Severe based on energy ratio assessment; Precise Location: Uplink line K120+351.875 meters" and sends it to the ground operation and maintenance center via wireless communication module 6. The operation and maintenance center automatically generates a maintenance work order based on the received early warning information and dispatches it to the nearest maintenance team.

[0042] Through the above process, the present invention achieves mobile, high-precision, and automated inspection of track fastener defects.

[0043] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the invention by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the invention should be included within the scope of protection of the invention.

Claims

1. An inspection device for rail transit, used to travel along the track, characterized in that, Includes an inspection vehicle (2), which can be moved and straddle the existing rail transit tracks; The inspection vehicle (2) includes: Vehicle body frame (21); The walking drive unit (24) is installed on the vehicle frame (21) and is used to drive the inspection vehicle (2) to move along the track; A mobile acoustic and vibration acquisition array is installed on the vehicle frame (21). The mobile acoustic and vibration acquisition array includes M acoustic sensors (31) and N vibration sensors (32) arranged at intervals along the travel direction of the inspection vehicle. M and N are both integers greater than or equal to 2. Mileage positioning unit (4) is used to obtain the position information of the inspection vehicle (2) on the track in real time; The edge computing unit (5) is electrically connected to the walking drive unit (24), the mobile acoustic and vibration acquisition array and the mileage positioning unit (4), respectively; The edge computing unit (5) includes: The synchronization trigger module is used to send a synchronization trigger signal to the mobile acoustic and vibration acquisition array and the mileage positioning unit (4) so ​​that the acoustic and vibration signal and the position signal are synchronized in time; The data fusion and noise reduction module is used to adaptively filter the acoustic signal collected by the acoustic sensor (31) using the vibration signal collected by the vibration sensor (32) as a reference, so as to remove background noise. The multi-source signal processing module is used to perform time-frequency domain transformation and coherence analysis on the noise-reduced multi-channel acoustic signal and the vibration signal collected by the vibration sensor (32) to extract characteristic parameters reflecting the health status of the track structure. The fault location module is used to calculate the position offset of the abnormal sound source relative to the inspection vehicle (2) based on the position signal of the mileage positioning unit (4) and the analysis results of the multi-source signal processing module when it is determined that there is an abnormality in the track and the abnormal sound source is located in the rail area between at least two of the acoustic sensors (31). The module uses the time difference of arrival positioning algorithm to calculate the position offset of the abnormal sound source relative to the inspection vehicle (2) based on the time delay difference of the signals of the at least two acoustic sensors (31) and the pre-stored sound velocity, and combines the real-time position information to obtain the absolute position of the abnormal sound source and generate fault warning information.

2. The inspection equipment according to claim 1, characterized in that, The inspection vehicle (2) straddles the track via the walking wheels (22) and guide wheels (23), and the track is a C-shaped steel or an I-beam.

3. The inspection equipment according to claim 1, characterized in that, The acoustic sensor (31) is a miniature microphone array arranged in a line or cross shape, pointing towards the top surface of the rail and the fastener area of ​​the track; the vibration sensor (32) is an accelerometer, which is closely attached to the axle of the inspection vehicle (2) or the vehicle frame (21).

4. The inspection equipment according to claim 1, characterized in that, It also includes a wireless communication module (6), which is electrically connected to the edge computing unit (5) and is used to send the early warning information generated by the edge computing unit (5) to the ground operation and maintenance center.

5. An inspection method for the inspection equipment based on any one of claims 1 to 4, characterized in that, Includes the following steps: Step S1: Control the inspection vehicle (2) to travel along the track at a preset speed; Step S2: During the driving process, the location information is obtained in real time through the mileage positioning unit (4), and at the same time, the time-synchronized multi-channel sound pressure signal is collected synchronously through the mobile acoustic vibration acquisition array. and vibration acceleration signal ; Step S3: Process the acquired sound pressure signal through the data fusion and noise reduction module. and vibration acceleration signal Preprocessing is performed to obtain the effective acoustic signal after noise reduction. ; Step S4: For the effective acoustic signal and vibration acceleration signal Feature extraction and anomaly detection are performed. If the extracted feature parameters exceed the normal range, the current mileage location is suspected of having a fault, and the fault type is determined. Step S5: When a fault is suspected and the abnormal sound source is located in the rail area between at least two acoustic sensors (31) at different spatial locations, the time difference of arrival (TDOA) positioning algorithm is used. Based on the time delay difference between the effective acoustic signals collected by at least two acoustic sensors (31) and combined with the pre-stored propagation speed of sound waves in the rail, the positional offset of the abnormal sound source relative to the inspection vehicle (2) is calculated. And, combined with the current mileage location, accurately calculate the absolute mileage location L of the fault point. fault ; Step S6: Output warning information containing the location of the fault and the type of fault.

6. The inspection method according to claim 5, characterized in that, The preprocessing in step S3 specifically includes: utilizing the vibration acceleration signal As a noise reference source, the sound pressure signal Perform adaptive filtering.

7. The inspection method according to claim 5, characterized in that, The feature extraction and anomaly detection in step S4 specifically include: For the noise-reduced multi-channel effective acoustic signal and vibration acceleration signal Coherence analysis is performed to calculate the coherence coefficient at different frequencies. When the coherence coefficient in a certain frequency band exceeds a preset threshold, the acoustic vibration signal in that frequency band is determined to be an effective characteristic signal. Envelope spectrum analysis and time-frequency domain transformation are performed on the effective feature signals to extract the characteristic frequencies and energy distribution patterns representing specific track defects; The extracted feature parameters are compared with a pre-stored library of normal track feature parameters. If they exceed the normal range, the current mileage position is suspected of having a fault, and the fault type is determined by feature matching.

8. The inspection method according to claim 5, characterized in that, In step S5, the time delay difference is obtained by calculating the cross-correlation function between the effective acoustic signals collected by at least two acoustic sensors (31) at different spatial locations. The calculation formula is: ； in ,and The effective acoustic signals collected by the i-th and k-th acoustic sensors (31) are respectively. This is a time delay variable.

9. The inspection method according to claim 8, characterized in that, In step S5, when the abnormal sound source is determined to be located in the rail area between the at least two acoustic sensors (31), the position offset is calculated. The formula is: ； in, The speed at which the pre-stored sound waves propagate in the rail; Calculate the absolute mileage location of the fault point. The formula is: ； in, The mileage location of the inspection vehicle when the anomaly was detected.

Images