CN110789566B - Track defect monitoring method and monitoring equipment based on axle box acceleration signal - Google Patents
Track defect monitoring method and monitoring equipment based on axle box acceleration signal Download PDFInfo
- Publication number
- CN110789566B CN110789566B CN201911092672.9A CN201911092672A CN110789566B CN 110789566 B CN110789566 B CN 110789566B CN 201911092672 A CN201911092672 A CN 201911092672A CN 110789566 B CN110789566 B CN 110789566B
- Authority
- CN
- China
- Prior art keywords
- axle box
- gps
- positioning
- signal
- vehicle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61K—AUXILIARY EQUIPMENT SPECIALLY ADAPTED FOR RAILWAYS, NOT OTHERWISE PROVIDED FOR
- B61K9/00—Railway vehicle profile gauges; Detecting or indicating overheating of components; Apparatus on locomotives or cars to indicate bad track sections; General design of track recording vehicles
- B61K9/08—Measuring installations for surveying permanent way
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Machines For Laying And Maintaining Railways (AREA)
Abstract
The invention discloses a track defect monitoring method and monitoring equipment based on axle box acceleration signals, which are applied to the field of track traffic and aim to solve the problem of low monitoring efficiency of the existing track health state monitoring, the invention adopts a mode of measuring the vertical vibration acceleration of a vehicle axle box to detect track faults and defects and combines GPS signals and vehicle speed integral to position a track, thereby realizing the accurate positioning of the defects of a multi-tunnel road section in a mountainous area, the invention can obtain the distribution characteristics of the axle box vibration acceleration caused by different track defects on a frequency domain by using wavelet transformation, can give consideration to the precision of a time domain and the frequency domain to obtain a high-resolution time-frequency diagram, simultaneously a convolutional neural network model can quickly and accurately identify and classify the time-frequency diagram, and only needs to input the axle box vertical acceleration signals and the GPS signals synchronous with the axle box vertical acceleration signals in the whole course of the running of a vehicle after the neural, the method and apparatus of the present invention can automatically identify the type of track defect and the GPS location on the line.
Description
Technical Field
The invention belongs to the field of rail transit, and particularly relates to a rail health state monitoring technology and equipment.
Background
The production life style of people is developed from a extensive mode to intensification and from pollution to environmental protection. In the field of transportation, electrified rail transit has been rapidly developed in recent years due to its high energy utilization rate and zero emission. The main reason for the high energy utilization rate of rail transit is the smaller rolling friction force caused by the larger contact rigidity of the steel wheel and the steel rail, so that the energy loss caused by overcoming the friction is reduced. However, the large contact rigidity brings about small friction force and greatly deteriorates the stress environment of the wheel and the steel rail. Therefore, in practical application, structural defects of the steel rail occur frequently, such as rail corrugation, lung lobe defects (squares) of the steel rail and even fatigue fracture. Larger rail faults tend to develop from smaller defects, which require timely discovery and timely disposal of rail defects during operation. Therefore, the monitoring and maintenance of the health state of the steel rail is an essential important guarantee for the safe operation of a railway system.
Over the past decades, our country has built very large scale rail transportation networks including general speed railways, high speed railways, subways, and emerging urban trams. With the gradual saturation of lines, the gravity center of the railway traffic infrastructure in China is changed from new construction to maintenance and modification. The huge track holding capacity presents a huge challenge to track condition monitoring. The traditional rail maintenance operation needs visual inspection by rail workers, and field measurement is carried out on key road sections by special instruments. The method has extremely low efficiency, wastes a large amount of manpower and material resources, and is difficult to control the influence of artificial uncertain factors on the detection effect. Therefore, the efficient, accurate and automatic rail health state monitoring system has huge market demands and important economic and safety significance. However, there are difficulties associated with automatic monitoring of the track condition. One is the enormous amount of data, the size of the track defect can be as small as a centimeter scale (10)-2) And the length of a main railway line can reach thousands of kilometers (10)6) So that the order of magnitude of the samples that the line needs to analyze is 108This is clearly a difficult task to manually inspect. The second criterion is how to determine the measured data from the measured signals (image, laser sensor, acceleration) for the defect and what type of defect. It is easy to subjectively judge due to the irregularity of the defect, but it is a great challenge to find a programmed mathematical discriminant rule for automatic monitoring.
Therefore, the invention solves the important technical problem by using the automatic, efficient and accurate rail health state monitoring method.
Disclosure of Invention
In order to solve the technical problems, the invention provides a track defect monitoring method and monitoring equipment based on an axle box acceleration signal, which are used for detecting track faults and defects by measuring the vertical vibration acceleration of a vehicle axle box and reducing the data volume needing to be processed.
One of the technical schemes adopted by the invention is as follows: a rail defect monitoring method based on an axle box acceleration signal comprises the following steps:
s1, establishing a database corresponding to the track state and the time-frequency signal; the method comprises the following steps:
a1, collecting a vertical vibration acceleration signal of a vehicle axle box;
a2, extracting the distribution characteristics of the vertical vibration acceleration signals of the vehicle axle box collected in the step A1 on a frequency domain by adopting wavelet transformation;
a3, surveying and obtaining the track state corresponding to each vehicle axle box vertical vibration acceleration signal, thereby establishing a database corresponding to the track state and the time-frequency signal;
s2, training the deep convolutional neural network by adopting the database in the step S1;
and S3, inputting the currently collected vertical vibration acceleration signals of the vehicle axle box into the deep convolutional neural network trained in the step S2, and outputting a corresponding track state by the deep convolutional neural network.
Further, step S3 includes locating the currently acquired vertical vibration acceleration signal of the axle box of the vehicle.
Furthermore, the positioning mode is as follows: when the GPS signal is good, the positioning process is based on GPS positioning, and when the GPS is interrupted:
b1, adopting the latest GPS signal to carry out rough positioning;
and B2, combining the coarse positioning of the step B1 by calculating the running distance of the vehicle to perform fine positioning.
Further, the fine positioning in step B2 is specifically: firstly, acquiring a vertical vibration acceleration signal of a vehicle axle box and acquiring the rotation angular velocity of a wheel; then, multiplying the wheel rotation angular velocity by the wheel rolling circle radius to obtain the vehicle speed; and finally, integrating the vehicle speed according to the latest GPS positioning to obtain the vehicle running distance and obtain accurate positioning.
Further, the deep convolutional neural network is an improved AlexNet, and specifically includes: changing the number of output parameters in the FullyConnectedLayer and the output layer of the 23 rd layer into 8 types, wherein the 8 types are respectively 1: crushing and sinking; 2: wave-shaped abrasion; 3: stripping the surface of the steel rail; 4: bruise, bruise and other traumas; 5: rail joints; 6: a turnout; 7: welding seams; 8: and (5) normal steel rails.
The second technical scheme adopted by the invention is as follows: a rail defect monitoring apparatus based on axlebox acceleration signals, comprising: one-way piezoelectric acceleration sensor, rotatory pulse speed sensor, GPS receiver, AD analog-to-digital conversion circuit, digital signal collector, power supply module, industrial computer, wireless network subassembly and remote work station, one-way piezoelectric acceleration sensor gathers the vertical vibration acceleration signal of vehicle axle box, rotatory pulse speed sensor gathers the signal that the wheel rolls the round and produces the primary pulse, and GPS receiver is used for acquireing real-time GPS positioning signal, one-way piezoelectric acceleration sensor, rotatory pulse speed sensor link to each other with AD analog-to-digital conversion circuit respectively, AD analog-to-digital conversion circuit links to each other with the digital signal collector, the digital signal collector links to each other with the industrial computer, industrial computer honor light wireless network subassembly links to each other with remote work station, power supply module is GPS receiver, digital signal collector, power supply module, And the industrial personal computer supplies power.
Further, the remote workstation at least comprises an auxiliary positioning module which is used for assisting positioning according to the last GPS position signal and the data collected by the rotary pulse velocimetry sensor when the GPS is interrupted.
Further, the unidirectional piezoelectric acceleration sensor is installed in the vertical direction of the axle box of the vehicle.
Further, the rotary pulse velocimetry sensor is mounted in a vehicle axle box.
The invention has the beneficial effects that: the invention adopts a mode of measuring the vertical vibration acceleration of the axle box of the vehicle to detect the fault and the defect of the track, thereby reducing the data volume needing to be processed; in addition, a speed integral auxiliary GPS positioning mode is adopted, the advantages of large space range and local anti-interference are considered, and accurate positioning of track defect complex road conditions can be realized; the method uses wavelet analysis to obtain the time-frequency graph of the acceleration, can give consideration to the precision of time domain and frequency domain, and obtains the time-frequency graph with high resolution; meanwhile, the convolutional neural network model can quickly and accurately identify and classify the time-frequency diagram, and after the neural network is trained, only a vertical acceleration signal of an axle box in the whole running process of the vehicle and a GPS signal synchronous with the vertical acceleration signal need to be input, so that the system can automatically identify the type of the track defect and the GPS position on the line; the invention adopts remote control, has simple operation and low required labor cost, and is suitable for long-term large-scale application.
Drawings
FIG. 1 is a schematic diagram of a remote acquisition and transmission system for axle box vertical acceleration signals, GPS signals and rotary pulse velocimetry sensor signals;
the system comprises a bogie, a wheel, a unidirectional piezoelectric crystal acceleration sensor, a bogie frame, an axle box and a rotary pulse speed measurement sensor, wherein 1 is the wheel, 2 is the unidirectional piezoelectric crystal acceleration sensor, 3 is the bogie frame, 4 is the axle box, and 5 is the rotary pulse speed measurement sensor;
FIG. 2 is a time domain signal of vertical vibration acceleration of the same axle box at the same speed on different road sections;
wherein (a) shown in fig. 2 is a rail surface scuffing depression and (b) shown in fig. 2 is a rail corrugation;
FIG. 3 is a wavelet transform of the corresponding acceleration signal of FIG. 2;
wherein, (a) shown in fig. 3 is the wavelet transform time-frequency diagram of (a) shown in fig. 2, and (b) shown in fig. 3 is the wavelet transform time-frequency diagram of (b) shown in fig. 2;
fig. 4 is a flow chart of an automatic detection method for track defects.
Detailed Description
In order to facilitate the understanding of the technical contents of the present invention by those skilled in the art, the present invention will be further explained with reference to the accompanying drawings.
Regardless of the size of the rail defect, it causes the wheel 1 to vibrate for a short time after rolling over it. In other words, the wheel 1 will never continue to vibrate due to the time it takes for the energy to decay, although the wheel 1 has rolled over the track defect. Therefore, the track defect is indirectly monitored by monitoring the vibration acceleration, the influence range of the track defect on the time domain is enlarged, and compared with the requirement of directly monitoring the sampling frequency required by the defect geometry on the space, the requirement of the sampling frequency required by the defect geometry on the space is reduced, so that the monitoring efficiency is improved; therefore, the invention detects the track fault and defect by measuring the vertical vibration acceleration of the axle box of the vehicle, and reduces the data volume needing to be processed.
The problem of locating a specific orbit is addressed here, and GPS signals are suitable for locating over a large area span because of their accuracy and low cost to reach the global scale. In open plain, the GPS signals can be continuously recorded. However, when the vehicle travels in a mountainous area, it is difficult to keep the GPS signal constant due to interference of obstacles such as mountains, and thus it is difficult to simply perform GPS positioning for a track defect in the mountainous area. In order to solve the problem, the invention adopts the following solution:
the rotation angular velocity of the wheel is measured by measuring the interval time of each pulse signal, and the rotation angular velocity of the wheel is multiplied by the rolling circle radius of the wheel to be equal to the vehicle speed; the vehicle travel distance can be obtained by integrating the vehicle speed.
Considering that each step of displacement calculation by velocity integration generates a tiny error, when the distance is too long, the accumulated error is not negligible, and a vehicle positioning method which only adopts velocity integration is not suitable for large-range positioning, so that a positioning method which is assisted by GPS signal positioning and velocity integration positioning is adopted; when GPS signals are good, GPS positioning is based, and when GPS is interrupted, the integral of a vehicle speed sensor is accumulated according to the last GPS position signal to assist positioning. The advantages of the two positioning modes are combined, and the defect accurate positioning of the multi-tunnel road section in the mountainous area can be realized.
The invention adopts monitoring equipment shown in figure 1 to collect vertical acceleration signals of an axle box, GPS signals and rotary pulse speed measurement sensor signals; the monitoring device includes: the device comprises a unidirectional piezoelectric acceleration sensor 2, a rotary pulse speed measurement sensor 5, a sensor matched signal line, a GPS receiver, an A/D (analog/digital) conversion circuit, a digital signal collector, power supply equipment and a matched part, an industrial personal computer and a wireless network connecting component; as shown in fig. 1, four unidirectional piezoelectric crystal acceleration sensors are mounted on the bogie 3 at vertical positions of four axle boxes 4. The sensors are arranged at the left end and the right end of the same wheel pair, so that data of two rails can be acquired, and the detection probability of the tiny rail defect can be increased by arranging the sensors at the front axle box and the rear axle box on the same side. Meanwhile, a pulse rotation speed measurement sensor is arranged in any axle box to acquire wheel rotation speed signals; the power supply and the industrial personal computer module are kept in a starting state at any time, when data are required to be collected, the remote workstation controls the industrial personal computer through the wireless network, and the industrial personal computer controls the digital signal collector to start collecting the data. The axle box vertical acceleration signal and the rotation speed signal are converted into digital signals through an A/D analog-to-digital circuit and then transmitted to the digital signal collector together with the GPS signals for storage. And after the acquisition process is finished, the tester controls the industrial personal computer to issue an acquisition stopping command to the data acquisition through the wireless network. The experimental data are uploaded to a magnetic disk of an industrial personal computer from a data acquisition memory card through a data line. And then the data is uploaded to a remote control workstation by an industrial personal computer through a wireless network. Therefore, the experimental data can be remotely operated and transmitted. The experimenter can collect the axle box vibration and the vehicle position data when the train runs in a laboratory.
In fig. 1, firstly, test signal data, secondly, a power supply path, and thirdly, a control signal are shown.
As shown in fig. 4, the track condition monitoring of the present invention is mainly divided into two processes:
and a first process, training a convolutional neural network model. This step requires a large number of rail defects and corresponding axle box vibration acceleration response data to train a convolutional neural network model that can be used for rail monitoring. The specific implementation method is that an acceleration sensor is installed on a vehicle axle box, and an axle box vertical acceleration signal is collected. As shown in fig. 2, different track defect forms have different axle box vibration responses, and the magnitude of vibration acceleration amplitude at a certain moment can be seen in the time domain, and is represented as local impact form as shown in (a) shown in fig. 2, and is mainly represented as overall larger amplitude vibration plus local impact as shown in (b) shown in fig. 2; the acceleration signal is a one-dimensional data sequence in the time domain, and it is difficult to qualitatively analyze whether the two impacts are the same defect in the time domain, and it is also difficult to determine the reason of the overall large-amplitude vibration of (b) shown in fig. 2; therefore, the acceleration signal in the time domain needs to be further processed; the processing mode adopted by the invention is as follows: the distribution characteristics of the axle box vibration acceleration caused by different track defects on the frequency domain are obtained by using wavelet transformation, and as shown in fig. 3, the time-frequency characteristic of the vibration acceleration within 1 second is obtained by performing wavelet transformation based on Complex Morlet wavelet on the time domain signal of fig. 2. Both the bandwidth parameter and the center frequency of the Complex Morlet wavelet are 3; it can be found that the frequency domain distribution of the signal in (a) shown in fig. 3 at the impact point is wide, and the energy is mainly concentrated around 300 Hz; while the signal of (b) shown in fig. 3 has a vibration component of concentrated energy around 550Hz in the entire range, which is particularly significant at the impact position; it can be found that the impact energies of (a) shown in fig. 3 and (b) shown in fig. 3 are distributed in the frequency domain with a significant difference.
Namely, the first process further comprises: and cutting the acceleration signal into a plurality of segments by taking one second as a unit, and performing wavelet analysis on each acceleration signal segment to obtain a wavelet transform time-frequency graph of the segment. And (4) surveying the track state corresponding to each signal segment, namely normal track, scratch, pit, corrugation, rail head joint, turnout, fastener failure, roadbed defect and the like. And (4) taking the investigated rail-shaped bodies as acceleration time-frequency signals to manufacture classification labels, and establishing a database corresponding to the rail states and the time-frequency signals. The database is then used to train a convolutional neural network model.
And secondly, automatically identifying the track health state by using the trained convolutional neural network model. The acceleration sensor is installed on the axle box vertical direction of the test vehicle, and the GPS receiver is synchronously installed on the vehicle. And acquiring an acceleration signal of the axle box and a GPS signal synchronized with the acceleration signal. The acceleration signal is divided into segments one second long and wavelet analysis is performed for each segment. And (4) judging the track state reflected by the signal of each time segment by using the convolutional neural network model trained in the previous step, and judging whether the signal has defects or not. And screening out the time segment with the track defect and outputting the defect type and the GPS information at the moment. Thus, automatic identification and position determination of track defects are achieved.
The deep convolutional neural network adopted by the invention is AlexNet, and is improved, wherein the number of output parameters in a FullyConnectedLayer (fully connected layer) of a 23 th layer and an output layer is changed into 8 types. 1: crushing and sinking; 2: wave-shaped abrasion; 3: stripping the surface of the steel rail; 4: bruise, bruise and other traumas; 5: rail joints; 6: a turnout; 7: welding seams; 8: normal steel rails; and establishing a deep learning database by adopting the axle box vibration video images of the eight steel rail state markers. Then training the modified AlexNet convolutional neural network model by using the database; classifying the time-frequency segments based on the trained AlexNet convolutional neural network model; then, if the video picture is classified into the 1 st to 7 th categories, the specific category and the corresponding time information are output. Searching GPS position information through the time information, and outputting the GPS position information if the GPS signal at the moment is normal; if the GPS signal is absent, the GPS signal at the position closest to the time is searched, and then the distance between the GPS signal and the nearest GPS position is calculated through the speed signal integration of the pulse rotation speed measurement sensor, so that the specific position of the track defect is calculated.
It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Various modifications and alterations to this invention will become apparent to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.
Claims (5)
1. A rail defect monitoring method based on an axle box acceleration signal is characterized by comprising the following steps:
s1, establishing a database corresponding to the track state and the time-frequency signal; the method comprises the following steps:
a1, collecting a vertical vibration acceleration signal of a vehicle axle box;
a2, extracting the distribution characteristics of the vertical vibration acceleration signals of the vehicle axle box collected in the step A1 on a frequency domain by adopting wavelet transformation;
a3, surveying and obtaining the track state corresponding to each vehicle axle box vertical vibration acceleration signal, thereby establishing a database corresponding to the track state and the time-frequency signal;
s2, training the deep convolutional neural network by adopting the database in the step S1;
s3, inputting the currently collected vertical vibration acceleration signal of the vehicle axle box to the deep convolutional neural network trained in the step S2, and outputting a corresponding track state by the deep convolutional neural network; step S3 also includes positioning the currently collected vertical vibration acceleration signal of the vehicle axle box; the specific positioning mode is as follows: when the GPS signal is good, the positioning process is based on GPS positioning, and when the GPS is interrupted:
b1, adopting the latest GPS signal to carry out rough positioning;
b2, fine positioning is carried out by calculating the running distance of the vehicle and combining the rough positioning of the step B1; the fine positioning specifically comprises the following steps: firstly, acquiring a vertical vibration acceleration signal of a vehicle axle box and acquiring the rotation angular velocity of a wheel; then, multiplying the wheel rotation angular velocity by the wheel rolling circle radius to obtain the vehicle speed; and finally, integrating the vehicle speed according to the latest GPS positioning to obtain the vehicle running distance and obtain accurate positioning.
2. The method for monitoring the rail defect based on the axle box acceleration signal according to claim 1, wherein the deep convolutional neural network is an improved AlexNet convolutional neural network, and specifically comprises: changing the number of output parameters in the fully-connected layer and the output layer of the 23 rd layer into 8 types, wherein the 8 types are respectively 1: crushing and sinking; 2: wave-shaped abrasion; 3: stripping the surface of the steel rail; 4: bruise, bruise and trauma; 5: rail joints; 6: a turnout; 7: welding seams; 8: and (5) normal steel rails.
3. A rail defect monitoring apparatus based on an axlebox acceleration signal, comprising: the system comprises a one-way piezoelectric acceleration sensor, a rotary pulse speed measurement sensor, a GPS receiver, an A/D (analog-to-digital) conversion circuit, a digital signal collector, a power supply assembly, an industrial personal computer, a wireless network assembly and a remote workstation, wherein the one-way piezoelectric acceleration sensor collects vertical vibration acceleration signals of a vehicle axle box, the rotary pulse speed measurement sensor collects signals of primary pulses generated by rolling of wheels, the GPS receiver is used for acquiring real-time GPS positioning signals, the one-way piezoelectric acceleration sensor and the rotary pulse speed measurement sensor are respectively connected with the A/D conversion circuit, the A/D conversion circuit is connected with the digital signal collector, the digital signal collector is connected with the industrial personal computer, the industrial personal computer is connected with the remote workstation through the wireless network assembly, and the power supply assembly is a GPS receiver, a digital, The industrial personal computer supplies power;
the remote workstation comprises an auxiliary positioning module which is used for assisting positioning according to the last GPS position signal and the data collected by the rotary pulse speed measurement sensor when the GPS is interrupted.
4. The apparatus according to claim 3, wherein the unidirectional piezoelectric acceleration sensor is installed in a vertical direction of the axle box of the vehicle.
5. The apparatus according to claim 3, wherein the rotary pulse tacho sensor is mounted within a vehicle axle housing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911092672.9A CN110789566B (en) | 2019-11-11 | 2019-11-11 | Track defect monitoring method and monitoring equipment based on axle box acceleration signal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911092672.9A CN110789566B (en) | 2019-11-11 | 2019-11-11 | Track defect monitoring method and monitoring equipment based on axle box acceleration signal |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110789566A CN110789566A (en) | 2020-02-14 |
| CN110789566B true CN110789566B (en) | 2021-01-12 |
Family
ID=69443727
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911092672.9A Active CN110789566B (en) | 2019-11-11 | 2019-11-11 | Track defect monitoring method and monitoring equipment based on axle box acceleration signal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110789566B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111637964B (en) * | 2020-05-12 | 2021-10-26 | 西南交通大学 | Rail corrugation identification method |
| CN112298273B (en) * | 2020-11-03 | 2021-09-14 | 石家庄铁道大学 | Wheel scratch length measuring method and device and terminal equipment |
| CN112381027B (en) * | 2020-11-23 | 2022-08-19 | 西南交通大学 | Wheel polygon wave depth estimation method based on train axle box vertical acceleration signal |
| CN113624517A (en) * | 2021-08-02 | 2021-11-09 | 李俊 | Vibration testing method based on combination of track line equipment and GPS system |
| CN116223075B (en) * | 2023-05-05 | 2023-08-11 | 昆明轨道交通四号线土建项目建设管理有限公司 | Vibration stability detection system and method for rail transit vehicle |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104260754A (en) * | 2014-10-08 | 2015-01-07 | 南京理工大学 | Track height irregularity prediction system and method based on axle box vibration acceleration |
| CN105292177A (en) * | 2015-11-26 | 2016-02-03 | 唐智科技湖南发展有限公司 | Method for measuring track corrugation by utilizing axle box vibration and impact information |
| CN109034192A (en) * | 2018-06-20 | 2018-12-18 | 西南交通大学 | A kind of track based on deep learning-body oscillating trend prediction method |
| CN110210555A (en) * | 2019-05-29 | 2019-09-06 | 西南交通大学 | Rail fish scale hurt detection method based on deep learning |
-
2019
- 2019-11-11 CN CN201911092672.9A patent/CN110789566B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104260754A (en) * | 2014-10-08 | 2015-01-07 | 南京理工大学 | Track height irregularity prediction system and method based on axle box vibration acceleration |
| CN105292177A (en) * | 2015-11-26 | 2016-02-03 | 唐智科技湖南发展有限公司 | Method for measuring track corrugation by utilizing axle box vibration and impact information |
| CN109034192A (en) * | 2018-06-20 | 2018-12-18 | 西南交通大学 | A kind of track based on deep learning-body oscillating trend prediction method |
| CN110210555A (en) * | 2019-05-29 | 2019-09-06 | 西南交通大学 | Rail fish scale hurt detection method based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 基于小波的列车加速度和轨道不平顺关系分析;曲铭等;《华 东 交 通 大 学 学 报》;20081031;第25卷(第5期);第36-41页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110789566A (en) | 2020-02-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110789566B (en) | Track defect monitoring method and monitoring equipment based on axle box acceleration signal | |
| CN109910947B (en) | System and method for dynamically monitoring and intelligently analyzing health state of loaded track | |
| Weston et al. | Perspectives on railway track geometry condition monitoring from in-service railway vehicles | |
| Hodge et al. | Wireless sensor networks for condition monitoring in the railway industry: A survey | |
| CN105416097B (en) | Contact Line Detection tool car and method based on laser radar | |
| CN103552579B (en) | Comprehensive detection train for freight heavy haul railway | |
| CN201429413Y (en) | Pantograph performance on-line automatic detection system for high-speed trains | |
| CN106192634A (en) | A kind of railroad track elastic bar fastener condition automatic detection device and method | |
| CN202011397U (en) | Multifunctional continuous inspection type track inspection vehicle | |
| CN111257415A (en) | Tunnel damage detection management system based on mobile train vibration signal | |
| CN110806324A (en) | Wheel polygon abrasion detection method based on rail displacement and data acquisition equipment | |
| CN105115605A (en) | Track train infrared detection system and detection method | |
| WO2019185873A1 (en) | System and method for detecting and associating railway related data | |
| US20210009175A1 (en) | System and method for extracting and processing railway-related data | |
| US20220017129A1 (en) | Onboard Railway Health Monitoring | |
| CN113466247B (en) | Rail weld detection method and system based on inertial technology and machine vision fusion | |
| CN203651812U (en) | Freight heavy railway comprehensive detection train | |
| Hoelzl et al. | On-board monitoring for smart assessment of railway infrastructure: A systematic review | |
| CN113988326A (en) | Subway equipment maintenance optimization method and system | |
| CN111332338A (en) | Track disease detection system | |
| Li et al. | Smart railway based on the Internet of Things | |
| CN107364467A (en) | A kind of fault monitoring device and method based on wheel to vibration signal | |
| CN112611442A (en) | Railway bridge health monitoring method and system based on track accompanying optical cable | |
| CN110171442B (en) | Detection system and detection method for wheel flat scars | |
| CN116331069A (en) | Portable bow net detection track inspection vehicle detection system with power |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |