CN112308858A - Multidimensional intelligent monitoring method and system for states of railway track and track slab - Google Patents
Multidimensional intelligent monitoring method and system for states of railway track and track slab Download PDFInfo
- Publication number
- CN112308858A CN112308858A CN202011586882.6A CN202011586882A CN112308858A CN 112308858 A CN112308858 A CN 112308858A CN 202011586882 A CN202011586882 A CN 202011586882A CN 112308858 A CN112308858 A CN 112308858A
- Authority
- CN
- China
- Prior art keywords
- track
- displacement
- image
- rail
- scale
- 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.)
- Granted
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 84
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000006073 displacement reaction Methods 0.000 claims abstract description 122
- 230000009193 crawling Effects 0.000 claims abstract description 29
- 238000001514 detection method Methods 0.000 claims abstract description 29
- 238000005259 measurement Methods 0.000 claims abstract description 28
- 238000012545 processing Methods 0.000 claims abstract description 19
- 238000005516 engineering process Methods 0.000 claims abstract description 12
- 238000013135 deep learning Methods 0.000 claims abstract description 11
- 230000000007 visual effect Effects 0.000 claims abstract description 9
- 230000002159 abnormal effect Effects 0.000 claims abstract description 6
- 238000004806 packaging method and process Methods 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 13
- 239000000126 substance Substances 0.000 claims description 12
- 238000003709 image segmentation Methods 0.000 claims description 10
- 238000012423 maintenance Methods 0.000 claims description 8
- 230000007246 mechanism Effects 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000004891 communication Methods 0.000 claims description 7
- 238000007689 inspection Methods 0.000 claims description 7
- 239000013589 supplement Substances 0.000 claims description 7
- 238000011161 development Methods 0.000 claims description 6
- 230000018109 developmental process Effects 0.000 claims description 6
- 238000007789 sealing Methods 0.000 claims description 6
- 230000007704 transition Effects 0.000 claims description 5
- 230000005059 dormancy Effects 0.000 claims description 4
- 230000001502 supplementing effect Effects 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 230000015556 catabolic process Effects 0.000 claims description 3
- 230000009194 climbing Effects 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 238000006731 degradation reaction Methods 0.000 claims description 3
- 238000005286 illumination Methods 0.000 claims description 3
- 238000007619 statistical method Methods 0.000 claims description 3
- 230000002194 synthesizing effect Effects 0.000 claims description 3
- 230000007774 longterm Effects 0.000 abstract description 3
- 238000009434 installation Methods 0.000 description 5
- 238000010276 construction Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000010191 image analysis Methods 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000009529 body temperature measurement Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000001681 protective effect Effects 0.000 description 1
- 230000001012 protector Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Geometry (AREA)
- Quality & Reliability (AREA)
- Length Measuring Devices By Optical Means (AREA)
- Machines For Laying And Maintaining Railways (AREA)
- Train Traffic Observation, Control, And Security (AREA)
Abstract
The invention provides a multidimensional intelligent monitoring method and a multidimensional intelligent monitoring system for states of a railway track and a track slab, wherein the monitoring method comprises the following steps: s1, acquiring the surface temperature of the rail and the rail plate by the multi-dimensional intelligent monitoring system, measuring the distance of the rail, converting the distance into the eccentric distance of the rail, and acquiring an observation image; s2, transmitting the surface temperature and the rail eccentric distance to a control module, and transmitting the observation image to a processing module, wherein the processing module calculates the rail plate displacement and the rail crawling by combining a deep learning network and a visual detection technology; s3, packaging and transmitting the measurement results of the surface temperature, the rail eccentric distance, the rail plate displacement and the rail crawling to a management center; and the S4 management center analyzes the temperature and displacement situation of each monitoring point position and pushes the railway track and the track slab to alarm the abnormal displacement. The multidimensional intelligent monitoring method and the multidimensional intelligent monitoring system realize comprehensive, long-term and intelligent monitoring of the states of the railway track and the track plate and improve the safety of railway operation.
Description
Technical Field
The invention relates to a multidimensional intelligent monitoring method and system for states of a railway track and a track slab, belonging to the field of safety monitoring of railway tracks and track slabs.
Background
Under the load effects of temperature, train braking, concrete shrinkage and the like, the railway tracks and the track slabs bear complex transverse force and longitudinal force. When the expansion plate of the track plate reaches more than 3mm due to high temperature and the height of a circuit or the direction deviation exceeds 3mm, timely arrangement and treatment are needed to avoid potential safety hazards of driving; at present, the railway track is climbed and moved, the track slab is displaced, parameters such as the joint size and the deformation condition of the connecting part of the transition plate and the supporting layer, the relative displacement of the transition plate and the track slab, the relative displacement of the supporting layer and the track slab and the like are manually measured and verified by a protector and a line patrol worker in the skylight time.
The railway track side equipment installation space is limited, the requirements on the size, the protective performance and the like of monitoring equipment are high, the interferences of train vibration, rain fog, wind, snow, mud and the like exist on site, the problems of difficulty in on-site electricity taking, high construction difficulty, high hardware and construction cost and the like exist when monitoring means such as common video monitoring, fiber bragg grating technology, acceleration displacement sensing and the like are used for on-site construction, and a new monitoring scheme and a new monitoring system are designed to comprehensively, long-term and intelligently monitor the states of railway tracks and track slabs and still have large requirements and exploration space.
Disclosure of Invention
In order to solve the defects in the prior art, the invention provides a multidimensional intelligent monitoring method and a multidimensional intelligent monitoring system for the states of a railway track and a track slab.
In order to achieve the purpose, the invention adopts the following technical scheme:
the invention discloses a multidimensional intelligent monitoring method for states of railway tracks and track slabs, which comprises the following steps:
s1, acquiring the surface temperature of the rail and the rail plate, measuring the distance of the rail, converting the surface temperature into the eccentric distance of the rail and acquiring an observation image by the multi-dimensional intelligent monitoring system;
s2, transmitting the surface temperature and the rail eccentric distance obtained in the step S1 to a multi-dimensional intelligent monitoring system control module, transmitting the observation image obtained in the step S1 to a multi-dimensional intelligent monitoring system processing module, and calculating the rail plate displacement and the rail crawling by combining a deep learning network and a visual detection technology through the processing module;
s3, packaging and transmitting the surface temperature, the rail eccentric distance, the rail plate displacement and the rail crawling measurement result to a management center;
s4, the management center analyzes the temperature and displacement situation of each monitoring point, pushes the railway track and the track slab abnormal displacement alarm, performs monthly statistics and analysis on the displacement development condition and the alarm frequency of each monitoring point, and combines the monitoring information of all monitoring points in the monitoring section to form a driving plan suggestion and a maintenance suggestion.
In one embodiment, a high-precision observation scale is installed at the observation position as a reference point for track slab displacement and track crawling measurement.
In one embodiment, in step S2, the calculating, by the processing module, the rail plate displacement and the rail crawling specifically includes:
s21, improving the image resolution by adopting image super-resolution and contrast enhancement to achieve the aim of high-precision measurement;
s22, positioning the boundary position of the scale by adopting an image segmentation method, acquiring a mask image of the observation scale, detecting key points, and calculating the coordinates of the displacement observation reference points by matching with the corner points of the mask image;
and S23, calculating the self-calibration of the camera parameters and the actual displacement of the scale by combining the rotation of the rotating mechanism and the standard size of the scale.
In one embodiment, step S21 specifically includes: designing a super-resolution network for observing scale edge and corner contrast enhancement, and measuring image texture by adopting correlation between different characteristic channels of an image in order to ensure that the image after super-resolution and a target image have the same contrast and texture, namely:
wherein the content of the first and second substances,
and
features of an ith channel and a jth channel of the image respectively;
to minimize the difference between the target image correlation and the generated image correlation, the loss function for image super-resolution and contrast enhancement training is:
wherein the content of the first and second substances,
for the super-resolution image predicted by the network,
in order to supervise the image(s),
is the number of channels of the image texture feature.
In one embodiment, the image super-resolution training data of step S21 is generated by the following two methods:
(1) synthesizing a bicubic interpolation image degradation method;
(2) and respectively using the low-resolution subcode stream and the high-resolution main stream to acquire images as input images and supervision images.
In one embodiment, step S22 specifically includes:
aiming at track slab displacement and track crawling monitoring scenes and calculation characteristics, a multitask deep learning network comprising target detection branches, image segmentation branches and key point detection branches is designed, the target detection branches carry out scale bounding box detection, the image segmentation branches carry out scale mask pixel level positioning, the key point detection branches carry out scale corner point detection and positioning, and a loss function for network training is as follows:
wherein the content of the first and second substances,
and
respectively object box class and object box coordinate loss,
for the loss function for the keypoint training:
wherein the content of the first and second substances,
n key points representing the scale, R being the real field,
，
is the output of the last layer of the network,
w is the input of the last layer of the network and is the weight;
after the monitoring equipment is installed, acquiring pixel coordinates of initial reference point of the scale
，
Wherein N is the number of reference points, the system regularly inspects the states of the track and the track slab to obtain the pixel coordinates of the reference point of the scale at the inspection moment
Calculating the displacement of pixel coordinates in horizontal and vertical directions corresponding to the reference point
And
：
in one embodiment, step S23 specifically includes:
after the displacement of the pixel coordinates of the scale reference point is obtained, calibrating the camera for calculating the displacement of the world coordinates, using a scale image to replace a checkerboard calibration plate for calibration, changing the relative position between the camera and a measuring surface by controlling the rotation of the rotating mechanism to obtain a calibration image, and solving the intrinsic parameters, distortion parameters and extrinsic parameters of the camera to obtain the intrinsic parameters of the camera: focal length of camera
Optical center coordinate
And distortion parameters including radial distortion parameters
And
tangential distortion parameter
And
，
calculating actual displacement, using camera internal parameters, and firstly displacing pixel coordinate
Converting into a camera coordinate displacement:
and carrying out distortion correction on the coordinate displacement of the camera by using the calibrated distortion parameters.
In one embodiment, step S4 specifically includes:
s41, acquiring a monitoring real-time image;
s42, calculating the absolute displacement of the track slab, the relative displacement of the track slab and the transition slab, and the relative displacement of the track slab and the roadbed sealing layer for the track slab; for the track, calculating the absolute creeping of the track and the relative creeping of the track-track plate;
s43, judging whether the absolute displacement of the track slab exceeds an alarm threshold T1;
s44, judging whether one of the relative displacement of the track slab-transition slab and the relative displacement of the track slab-roadbed sealing layer exceeds an alarm threshold T2 or not for the track slab; for the track, judging whether the relative creep of the track-track plate exceeds an alarm threshold T3;
s45, if the continuous M times of measurement S43-S44 are met, pushing track crawling or track slab displacement alarm, otherwise, recording inspection records, and dividing the track slab displacement into various alarm types according to the temperature and the displacement situation;
and S46, carrying out statistical analysis on the displacement development condition and the alarm frequency of each monitoring point location within a period of time, and combining the monitoring information of all the point locations in the monitoring section to form a driving plan suggestion and a maintenance suggestion.
The invention also provides a monitoring system adopting the multi-dimensional intelligent monitoring method for the states of the railway track and the track slab, which comprises the following steps:
the system comprises a measuring assembly, a control assembly and a control assembly, wherein the measuring assembly comprises infrared temperature measuring equipment, laser distance measuring equipment, an image sensor and a near-infrared light supplement lamp, the infrared temperature measuring equipment measures the surface temperature of a rail and a rail plate, the laser distance measuring equipment measures the distance of the rail and converts the distance into the eccentric distance of the rail, and the image sensor is switched to a night mode and triggers the near-infrared light supplement lamp to be turned on at low illumination;
the control assembly comprises a control module and a processing module, wherein the processing module receives an observation image collected by an image sensor, calculates displacement and track crawling of a track plate by combining a deep learning network and a visual detection technology, and the image sensor, a light supplement lamp and a laser ranging sensor are arranged on the same rotating platform which is provided with a plurality of preset point positions and rotates to an observation position during working and rotates to a hiding position during dormancy;
the control module transmits the surface temperature, the rail eccentric distance, the rail plate displacement and the rail crawling measurement result to the management center through the communication assembly;
and the power supply assembly is used for supplying power to the monitoring system.
In one embodiment, the power supply assembly comprises a solar panel, a solar controller, and a DC power supply interface, and provides solar panel power supply, DC power supply and storage battery power supply modes.
The multidimensional intelligent monitoring method and the multidimensional intelligent monitoring system have the following beneficial effects:
(1) the invention combines a plurality of intelligent sensing technologies of infrared temperature measurement, laser distance measurement and vision measurement, and senses and measures multidimensional data such as images, track/track plate temperature, atmospheric temperature, track plate displacement, track crawling and the like by designing monitoring equipment with the functions of power management control, automatic avoidance of severe weather interference, system dormancy/awakening, automatic control of camera working posture and the like, and realizes alarming of abnormal displacement and classification of alarming levels.
(2) In the aspect of displacement measurement algorithm, firstly, a super-resolution network is adopted to improve the resolution of an original image, and the optimization of network parameters is carried out by taking the feature correlation difference of multiple channels of the image as a loss function of training; then, in order to eliminate the interference of the complex background of the track and the track slab, a multi-task learning depth network with target detection, segmentation and key point detection branches is designed, the bounding box detection, the pixel level segmentation and the detection of a scale reference point of the scale position are carried out, and then the calculation and the measurement of the displacement are carried out; and finally, calibrating the camera parameters by using the observation scale image instead of the checkerboard calibration plate, obtaining different reference images by controlling the rotation of the rotating mechanism to change the relative position between the camera and the measuring surface, and solving the internal parameters, distortion parameters and external parameters of the camera so as to calculate the track crawling and the track slab displacement.
(3) In the aspect of displacement alarm logic, the abnormal state of the track slab is timely pre-warned by combining the mutual influence of the temperature and the displacement of the track/track slab, the mutual displacement of the absolute displacement of the track/track slab and the displacement of the track, the track slab, the supporting layer and the roadbed, and the health state of the railway track and the track slab in the section is comprehensively analyzed by combining multi-point monitoring information, so that the requirement on site inspection personnel is reduced, the time required by maintenance and overhaul is shortened, the operation efficiency of overhaul is enhanced, and the operation safety coefficient of the train is improved.
Drawings
The present invention will be further described and illustrated with reference to the following drawings.
Fig. 1 is a schematic diagram of a multi-dimensional intelligent monitoring system for the state of a railway track and a railway plate in a preferred embodiment of the invention.
Fig. 2 is a flow chart of a processing module of the monitoring system of fig. 1 calculating railway track, track slab displacement.
Fig. 3 is a schematic diagram of the alarm logic of the vertical displacement of the track slab in the monitoring method of the invention.
Detailed Description
The technical solution of the present invention will be more clearly and completely explained by the description of the preferred embodiments of the present invention with reference to the accompanying drawings.
As shown in fig. 1, a multidimensional intelligent monitoring system for the status of a railway track and a railway plate according to a preferred embodiment of the present invention comprises: the device comprises a measuring component 1, a control component, a communication component 3 and a power supply component 4.
The measuring component 1 comprises infrared temperature measuring equipment, laser ranging equipment, an image sensor and a near-infrared light supplementing lamp. The infrared temperature measuring equipment measures the surface temperature of the rail and the track slab, and the angle is controlled through the rotating mechanism. The laser distance measuring device measures the distance of the rail and converts the distance into the change of the eccentric state of the rail, namely the eccentric distance. The image sensor has an automatic photosensitive function, and is switched to a night mode and triggers the near-infrared light supplement lamp to be turned on at low illumination. The near-infrared light supplement lamp adopts 940nm wavelength invisible to human eyes, and realizes normal collection and practicability of images at daytime and night.
The control assembly includes a control module 21 and a processing module 22. The control module 21 is a control circuit, the processing module 22 is an embedded image analysis core board, the embedded image analysis hardware is an AI core board, the processing module receives an observation image collected by the image sensor, measures the displacement of the track board and the track crawling by combining a deep learning network and a visual detection technology, and transmits the measurement parameters to the control circuit.
The image sensor, the light supplementing lamp and the laser ranging sensor are arranged on the same rotary platform, the horizontal rotation angle is 0-180 degrees, the image sensor, the light supplementing lamp and the laser ranging sensor can be arranged through preset point positions, rotate to an observation position during working and rotate to a hiding position during sleeping. The control circuit is communicated with the rotating mechanism and the sensor, output signals of the temperature measuring sensor and the laser ranging sensor are converted into network protocol signals through the control circuit, and are packed into data packets together with track slab displacement and track crawling measurement results, and the data packets are transmitted to the management center through the network, and the protocol format of the data packets is shown in the following table:
the communication component 3 comprises a 4G wireless router and a switch device, and supports two communication modes of an internet or a local area network wired network and a wireless connection. The control module transmits the surface temperature, the rail eccentric distance, the rail plate displacement and the rail crawling measurement result to the management center through the communication assembly.
The power supply assembly 4 comprises a solar cell panel and a solar controller, a 12V direct-current power supply interface is reserved, multiple power supply modes of solar cell panel power supply, direct-current power supply and storage battery power supply are provided, access of multiple power forms is facilitated, multiple modes of short-term work, long-term stable work and clean energy work are achieved, and the power management module is integrated to achieve effective control and use of electric quantity.
The multidimensional intelligent monitoring equipment consists of a power supply assembly, a communication assembly, a measuring assembly and a control processing assembly, can realize the functions of measuring the displacement of the track and the track plate, the temperature of the track and the track plate, the atmospheric temperature and the eccentricity of the track, simultaneously has the functions of power management control, automatic avoidance of severe weather interference, system dormancy/awakening, automatic control of the working posture of a camera and the like, realizes the automation and the intellectualization of data acquisition, and ensures the high reliability of data.
The invention also provides a multidimensional intelligent monitoring method for the states of the railway track and the railway plate, which comprises the following steps:
s1, acquiring the surface temperature of the rail and the rail plate, measuring the distance of the rail, converting the distance into the eccentric distance of the rail and acquiring an observation image by the multi-dimensional intelligent monitoring system;
step S2, transmitting the surface temperature and the rail eccentricity distance acquired in the step S1 to a control module; transmitting the observation image obtained in the step S1 to a processing module, and calculating the displacement of the track slab and the track crawling by combining a deep learning network and a visual detection technology through the processing module;
step S3, packaging and transmitting the surface temperature, the rail eccentric distance, the rail plate displacement and the rail crawling measurement result to a management center;
and step S4, the management center analyzes the temperature and the displacement situation of each monitoring point, pushes the railway track and the track slab abnormal displacement alarm, simultaneously carries out monthly statistic analysis on the displacement development condition and the alarm frequency of each monitoring point, and combines the monitoring information of all monitoring points in the monitoring section to form a driving plan suggestion and a maintenance suggestion.
Specifically, when the railway track crawling and the track plate displacement are calculated, the complex and variable shapes of the surfaces of the railway track and the track plate are considered, no uniform reference point is available for the comparison calculation of the displacement, the algorithm is high in complexity and poor in generalization performance due to the fact that direct measurement is conducted, a high-precision observation scale is installed at an observation position, the deep learning image segmentation and visual measurement technology is combined, the segmentation of the scale position and the positioning of the scale reference point are conducted, and the displacement of the scale reference point represents the displacement of the track and the track plate position.
The image analysis core plate collects images and carries out calculation of track crawling and track plate displacement, image resolution is guaranteed of measurement accuracy in visual measurement, measurement with higher accuracy is higher in requirement on image resolution, measurement requirements are met by generally configuring a high-resolution camera, but the imaging module size is increased, power is increased, cost is increased, and equipment installation and integration are not facilitated.
As shown in fig. 2, in order to avoid the disadvantage of large overall dimension of the imaging module, which is not beneficial to equipment installation and integration, the invention adopts the following method to calculate the displacement of the railway track and the track slab, and the calculating method comprises the following steps:
step S21, the image resolution is improved by adopting the image super-resolution and contrast enhancement technology, and the purpose of high-precision measurement is achieved;
s22, positioning the boundary position of the scale by adopting an image segmentation method to accurately measure track crawling and track slab displacement, acquiring a mask image of the observation scale, detecting key points of the scale to improve the positioning accuracy of reference points, and calculating coordinates of a displacement observation reference point by matching with angular points of the mask image;
and step S23, finally, performing camera parameter self-calibration and actual displacement calculation by combining the standard sizes of the rotating mechanism and the scale.
Specifically, in the image super-resolution and contrast enhancement of step S21, the edge and corner of the displacement observation scale reflect key information of scale positioning and key point detection, and the edge and corner belong to low-level semantic information, a super-resolution network for edge and corner contrast enhancement is designed, the image after super-resolution requires the same contrast and texture as the target image, and the image texture is measured by using the correlation between different characteristic channels of the image, that is:
wherein the content of the first and second substances,
and
the characteristics of the ith channel and the jth channel of the image are respectively represented by six-channel gray level co-occurrence matrixes of an RGB color space and an HSV color space, and the characteristics are one
L is a gray level, quantized to 32 bits.
The loss function for image super-resolution and contrast enhancement training requires minimizing the difference between the target image correlation and the generated image correlation:
the image super-resolution training data is generated by two methods, which are respectively: (1) synthesizing a bicubic interpolation image degradation method; (2) and respectively using the low-resolution subcode stream and the high-resolution main stream to acquire images as input images and supervision images. In practical use, the high-resolution main code stream image is collected and used as the input of the super-resolution network, and a super-resolution output image sampled by 2 times is obtained.
In the step S22, in the positioning of the image segmentation scale and the detection of the key points, the positioning of the reference points of the observation scale and the displacement measurement use super-resolution images as input, and in order to accurately measure the track crawling and the track slab displacement, the boundary position of the scale is positioned by adopting an image segmentation method, and a mask image of the observation scale is obtained, and the positioning of the reference points of the scale is performed by combining with the key point detection technology.
The design contains the target detection, cuts apart and the key point detects the branched multitask degree of deep learning network, and the target detection branch carries out the scale bounding box and detects, cuts apart the branch and carries out scale mask pixel level location, and the key point detects the branch and carries out the detection and the location of scale angular point for the loss function of network training is:
wherein the content of the first and second substances,
and
respectively object box class and object box coordinate loss,
for the loss function for the iterative training of the keypoints:
wherein the content of the first and second substances,
n key points representing the scale, R being the real field,
，
is the output of the last layer of the network,
w is the weight, which is the input to the last layer of the network.
After the monitoring equipment is installed, acquiring pixel coordinates of initial reference point of the scale
，
And N is the number of the reference points. System timing pair track, trackInspecting the board state to obtain the pixel coordinates of the reference point of the scale at the inspection time
Calculating the displacement of pixel coordinates in horizontal and vertical directions corresponding to the reference point
And
：
in the actual displacement calculation of the displacement observation scale in step S23, after the displacement of the pixel coordinates of the scale reference point is obtained, the camera is calibrated to calculate the displacement of the world coordinates, and the camera internal parameters (the focal length of the camera) are obtained
Optical center coordinate
) And distortion parameter (radial distortion parameter)
,
Tangential distortion parameter
And
）。
the parameter matrix A in the camera has 5 parameters, and the general Zhangyingyou calibration method needs to shoot at least three chessboard grids calibration boards to solve the parameters. The railway scene installation equipment generally has skylight time, the field shooting operability is not high, and higher requirements are provided for installation personnel.
Calculating actual displacement, using camera internal parameters, and firstly displacing pixel coordinate
Converting into a camera coordinate displacement:
and carrying out distortion correction on the coordinate displacement of the camera by using the calibrated distortion parameters to obtain the actual displacement of the scale:
wherein the content of the first and second substances,
。
after the displacement of the rail climbing and the rail plate of the railway is calculated, according to the displacement disease grade and the maintenance suggestion, 3-grade alarming is carried out on the displacement exceeding 1mm, 2-grade alarming is carried out on the displacement exceeding 2mm, and 1-grade alarming is carried out on the displacement exceeding 3 mm.
Specifically, the steps of climbing the railway track, alarming the displacement of the track slab and judging the grade classification comprise the following steps:
step S41, acquiring a monitoring real-time image;
step S42, calculating the absolute displacement of the track slab, the relative displacement of the track slab and the transition slab, and the relative displacement of the track slab and the roadbed sealing layer for the track slab; for the track, calculating the absolute creeping of the track and the relative creeping of the track-track plate;
step S43, judging whether the absolute displacement exceeds an alarm threshold T1;
step S44, judging whether one of the relative displacement of the track slab-transition slab and the relative displacement of the track slab-roadbed sealing layer exceeds an alarm threshold T2 or not for the track slab; for the track, judging whether the relative creep of the track-track plate exceeds an alarm threshold T3;
s45, if the continuous M times of measurement S43-S44 are met, pushing track crawling or track slab displacement alarm, otherwise, recording inspection record, and dividing the track slab displacement into alarm types such as high-temperature slab expansion and low-temperature shrinkage according to temperature and displacement situations;
and S46, carrying out statistical analysis on the displacement development condition and the alarm frequency of each monitoring point location within a period of time, and combining the monitoring information of all the point locations in the monitoring section to form a driving plan suggestion and a maintenance suggestion.
As shown in fig. 3, it is a schematic diagram of a vertical displacement alarm logic of the track slab.
The above detailed description merely describes preferred embodiments of the present invention and does not limit the scope of the invention. Without departing from the spirit and scope of the present invention, it should be understood that various changes, substitutions and alterations can be made herein by those skilled in the art without departing from the spirit and scope of the invention as defined by the appended claims and their equivalents. The scope of the invention is defined by the claims.
Claims (10)
1. A multidimensional intelligent monitoring method for the states of railway tracks and track slabs is characterized by comprising the following steps:
s1, acquiring the surface temperature of the rail and the rail plate, measuring the distance of the rail, converting the distance into the eccentric distance of the rail and acquiring an observation image by the multi-dimensional intelligent monitoring system;
s2, transmitting the surface temperature and the rail eccentricity distance acquired in the step S1 to a control module of a multidimensional intelligent monitoring system, transmitting the observation image acquired in the step S1 to a processing module of the multidimensional intelligent monitoring system, and calculating the track slab displacement and the track crawling by combining a deep learning network and a visual detection technology through the processing module;
s3, packaging and transmitting the surface temperature, the rail eccentric distance, the rail plate displacement and the rail crawling measurement result to a management center;
s4, the management center analyzes the temperature and displacement situation of each monitoring point, pushes the railway track and the track slab abnormal displacement alarm, performs monthly statistics and analysis on the displacement development condition and the alarm frequency of each monitoring point, and combines the monitoring information of all monitoring points in the monitoring section to form a driving plan suggestion and a maintenance suggestion.
2. The monitoring method of claim 1, further comprising:
and a high-precision observation scale is arranged at the observation position and is used as a reference point for track slab displacement and track crawling measurement.
3. The monitoring method according to claim 2, wherein in step S2, the calculating of the rail plate displacement and the rail climbing by the processing module specifically includes:
s21, improving the image resolution by adopting image super-resolution and contrast enhancement to achieve the aim of high-precision measurement;
s22, positioning the boundary position of the scale by adopting an image segmentation method, acquiring a mask image of the observation scale, detecting key points, and calculating the coordinates of the displacement observation reference points by matching with the corner points of the mask image;
and S23, calculating the self-calibration of the camera parameters and the actual displacement of the scale by combining the rotation of the rotating mechanism and the standard size of the scale.
4. The monitoring method according to claim 3, wherein the step S21 specifically includes:
designing a super-resolution network for observing scale edge and corner contrast enhancement, and measuring image texture by adopting correlation between different characteristic channels of an image in order to ensure that the image after super-resolution and a target image have the same contrast and texture, namely:
wherein the content of the first and second substances,
and
features of an ith channel and a jth channel of the image respectively;
to minimize the difference between the target image correlation and the generated image correlation, the loss function for image super-resolution and contrast enhancement training is:
wherein the content of the first and second substances,
for the super-resolution image predicted by the network,
in order to supervise the image(s),
is the number of channels of the image texture feature.
5. The monitoring method according to claim 4, wherein the image super-resolution training data of step S21 is generated by using two methods:
(1) synthesizing a bicubic interpolation image degradation method;
(2) and respectively using the low-resolution subcode stream and the high-resolution main stream to acquire images as input images and supervision images.
6. The monitoring method according to claim 3, wherein the step S22 specifically includes:
aiming at track slab displacement and track crawling monitoring scenes and calculation characteristics, a multitask deep learning network comprising target detection branches, image segmentation branches and key point detection branches is designed, the target detection branches carry out scale bounding box detection, the image segmentation branches carry out scale mask pixel level positioning, the key point detection branches carry out scale corner point detection and positioning, and a loss function for network training is as follows:
wherein the content of the first and second substances,
and
respectively object box class and object box coordinate loss,
for the loss function for the keypoint training:
wherein the content of the first and second substances,
n key points representing the scale, R being the real field,
，
is the output of the last layer of the network,
w is the input of the last layer of the network and is the weight;
after the monitoring equipment is installed, acquiring pixel coordinates of initial reference point of the scale
，
Wherein N is the number of reference points, the system regularly inspects the states of the track and the track slab to obtain the pixel coordinates of the reference point of the scale at the inspection moment
Calculating the displacement of pixel coordinates in horizontal and vertical directions corresponding to the reference point
And
：
7. the monitoring method according to claim 3, wherein the step S23 specifically includes:
after the displacement of the pixel coordinates of the scale reference point is obtained, calibrating the camera for calculating the displacement of the world coordinates, using a scale image to replace a checkerboard calibration plate for calibration, changing the relative position between the camera and a measuring surface by controlling the rotation of the rotating mechanism to obtain a calibration image, and solving the intrinsic parameters, distortion parameters and extrinsic parameters of the camera to obtain the intrinsic parameters of the camera: focal length of camera
Optical center coordinate
And distortion parameters including radial distortion parameters
And
tangential distortion parameter
And
，
calculating actual displacement, using camera internal parameters, and firstly displacing pixel coordinate
Converting into a camera coordinate displacement:
and carrying out distortion correction on the coordinate displacement of the camera by using the calibrated distortion parameters.
8. The monitoring method according to claim 1, wherein the step S4 specifically includes:
s41, acquiring a monitoring real-time image;
s42, calculating the absolute displacement of the track slab, the relative displacement of the track slab and the transition slab, and the relative displacement of the track slab and the roadbed sealing layer for the track slab; for the track, calculating the absolute creeping of the track and the relative creeping of the track-track plate;
s43, judging whether the absolute displacement of the track slab exceeds an alarm threshold T1;
s44, judging whether one of the relative displacement of the track slab-transition slab and the relative displacement of the track slab-roadbed sealing layer exceeds an alarm threshold T2 or not for the track slab; for the track, judging whether the relative creep of the track-track plate exceeds an alarm threshold T3;
s45, if the continuous M times of measurement S43-S44 are met, pushing track crawling or track slab displacement alarm, otherwise, recording inspection records, and dividing the track slab displacement into various alarm types according to the temperature and the displacement situation;
and S46, carrying out statistical analysis on the displacement development condition and the alarm frequency of each monitoring point location within a period of time, and combining the monitoring information of all the point locations in the monitoring section to form a driving plan suggestion and a maintenance suggestion.
9. A monitoring system using the multidimensional intelligent monitoring method for the states of the railway track and the railway plate as claimed in claim 1, which comprises:
the system comprises a measuring assembly, a control assembly and a control assembly, wherein the measuring assembly comprises infrared temperature measuring equipment, laser distance measuring equipment, an image sensor and a near-infrared light supplement lamp, the infrared temperature measuring equipment measures the surface temperature of a rail and a rail plate, the laser distance measuring equipment measures the distance of the rail and converts the distance into the eccentric distance of the rail, and the image sensor is switched to a night mode and triggers the near-infrared light supplement lamp to be turned on at low illumination;
the control assembly comprises a control module and a processing module, the processing module receives an observation image collected by the image sensor, and calculates displacement and track crawling of the track slab by combining a deep learning network and a visual detection technology, the image sensor, the light supplementing lamp and the laser ranging sensor are arranged on the same rotating platform, the rotating platform is provided with a plurality of preset point positions, rotates to an observation position during working and rotates to a hiding position during dormancy;
the control module transmits the surface temperature, the rail eccentric distance, the rail plate displacement and the rail crawling measurement result to a management center through the communication assembly;
and the power supply assembly is used for supplying power to the monitoring system.
10. The monitoring system of claim 9, wherein the power supply assembly includes a solar panel, a solar controller having a dc power interface providing solar panel power, dc power and battery power modes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011586882.6A CN112308858B (en) | 2020-12-29 | 2020-12-29 | Multidimensional intelligent monitoring method and system for states of railway track and track slab |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011586882.6A CN112308858B (en) | 2020-12-29 | 2020-12-29 | Multidimensional intelligent monitoring method and system for states of railway track and track slab |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112308858A true CN112308858A (en) | 2021-02-02 |
| CN112308858B CN112308858B (en) | 2021-05-04 |
Family
ID=74487600
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011586882.6A Active CN112308858B (en) | 2020-12-29 | 2020-12-29 | Multidimensional intelligent monitoring method and system for states of railway track and track slab |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112308858B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112949482A (en) * | 2021-03-01 | 2021-06-11 | 浙江大学 | Non-contact type rail sleeper relative displacement real-time measurement method based on deep learning and visual positioning |
| CN112949479A (en) * | 2021-03-01 | 2021-06-11 | 浙江大学 | Non-contact type rail sleeper relative displacement real-time measurement method based on deep learning and perspective transformation |
| CN113525455A (en) * | 2021-07-22 | 2021-10-22 | 中国铁道科学研究院集团有限公司电子计算技术研究所 | Train-following inspection communication system and method and train dynamic condition index estimation method |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104401360A (en) * | 2014-11-18 | 2015-03-11 | 北京交通大学 | Multi-means fusion-based safety real-time monitoring method and multi-means fusion-based safety real-time monitoring system for railway track system |
| CN105667540A (en) * | 2015-12-11 | 2016-06-15 | 中国铁道科学研究院铁道建筑研究所 | Rail state detection system |
| KR20170142696A (en) * | 2016-06-20 | 2017-12-28 | (주)호승이앤씨 | Measuring apparatus for rail and measuring method using the same |
| CN107609486A (en) * | 2017-08-16 | 2018-01-19 | 中国地质大学（武汉） | To anti-collision early warning method and system before a kind of vehicle |
| CN107858883A (en) * | 2017-11-29 | 2018-03-30 | 北京交通大学 | A kind of rail system safe condition comprehensive monitoring and intelligent analysis method |
| CN107928675A (en) * | 2017-11-22 | 2018-04-20 | 王华锋 | A kind of trunk measuring method being combined based on deep learning and red dot laser |
| CN109443480A (en) * | 2018-11-02 | 2019-03-08 | 南京邮电大学 | Gauge positioning and water level measurement method based on image procossing |
| CN111935412A (en) * | 2020-10-19 | 2020-11-13 | 广东科凯达智能机器人有限公司 | Method, system and robot for automatically identifying and tracking inspection target |
-
2020
- 2020-12-29 CN CN202011586882.6A patent/CN112308858B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104401360A (en) * | 2014-11-18 | 2015-03-11 | 北京交通大学 | Multi-means fusion-based safety real-time monitoring method and multi-means fusion-based safety real-time monitoring system for railway track system |
| CN105667540A (en) * | 2015-12-11 | 2016-06-15 | 中国铁道科学研究院铁道建筑研究所 | Rail state detection system |
| KR20170142696A (en) * | 2016-06-20 | 2017-12-28 | (주)호승이앤씨 | Measuring apparatus for rail and measuring method using the same |
| CN107609486A (en) * | 2017-08-16 | 2018-01-19 | 中国地质大学（武汉） | To anti-collision early warning method and system before a kind of vehicle |
| CN107928675A (en) * | 2017-11-22 | 2018-04-20 | 王华锋 | A kind of trunk measuring method being combined based on deep learning and red dot laser |
| CN107858883A (en) * | 2017-11-29 | 2018-03-30 | 北京交通大学 | A kind of rail system safe condition comprehensive monitoring and intelligent analysis method |
| CN109443480A (en) * | 2018-11-02 | 2019-03-08 | 南京邮电大学 | Gauge positioning and water level measurement method based on image procossing |
| CN111935412A (en) * | 2020-10-19 | 2020-11-13 | 广东科凯达智能机器人有限公司 | Method, system and robot for automatically identifying and tracking inspection target |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112949482A (en) * | 2021-03-01 | 2021-06-11 | 浙江大学 | Non-contact type rail sleeper relative displacement real-time measurement method based on deep learning and visual positioning |
| CN112949479A (en) * | 2021-03-01 | 2021-06-11 | 浙江大学 | Non-contact type rail sleeper relative displacement real-time measurement method based on deep learning and perspective transformation |
| CN112949482B (en) * | 2021-03-01 | 2022-04-29 | 浙江大学 | Non-contact type rail sleeper relative displacement real-time measurement method based on deep learning and visual positioning |
| CN113525455A (en) * | 2021-07-22 | 2021-10-22 | 中国铁道科学研究院集团有限公司电子计算技术研究所 | Train-following inspection communication system and method and train dynamic condition index estimation method |
| CN113525455B (en) * | 2021-07-22 | 2023-08-25 | 中国铁道科学研究院集团有限公司电子计算技术研究所 | Communication system and method for inspection along with train and train condition index estimation method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112308858B (en) | 2021-05-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112308858B (en) | Multidimensional intelligent monitoring method and system for states of railway track and track slab | |
| CN109716108B (en) | Bituminous paving disease detecting system based on two mesh image analysis | |
| CN111595453A (en) | Infrared temperature measurement system and method based on face recognition | |
| AU2019390444B2 (en) | Tower clearance monitoring system and method therefor | |
| CN109208468B (en) | Automatic bridge inspection system based on video analysis | |
| CN103047930B (en) | A kind of vehicular subway tunnel disease data automatic acquisition system | |
| CN102564335A (en) | Method for measuring deformation of large-scale tunnel | |
| CN112945096B (en) | Tunnel disease monitoring system and method for high-speed train | |
| CN110310503B (en) | High-precision positioning system of miniature model car | |
| CN104410839A (en) | Mobile online monitoring system for mountain fire and landslide hazards in power transmission line pole and tower region and monitoring method | |
| JP3466169B2 (en) | Management system for roads and surrounding facilities | |
| CN103321129A (en) | Remote pavement construction diagnosis system and method based on infrared thermal image of third-generation (3G) network | |
| CN112854175A (en) | Foundation pit deformation monitoring and early warning method based on machine vision | |
| CN110213536B (en) | Monitoring method of shore-based digital image monitoring system for coastal zone | |
| CN110702343B (en) | Deflection measurement system and method based on stereoscopic vision | |
| CN110646434A (en) | Device and method for automatically monitoring dam damage deformation in real time through video | |
| CN113569956A (en) | Mountain fire disaster investigation and identification method based on AI algorithm | |
| CN115620239A (en) | Point cloud and video combined power transmission line online monitoring method and system | |
| CN116665125A (en) | Intelligent recognition system for bridge tunnel diseases | |
| Xiao et al. | Research on pose adaptive correction method of indoor rail mounted inspection robot in GIS Substation | |
| CN112326039B (en) | Photovoltaic power plant patrols and examines auxiliary system | |
| CN117704970A (en) | Building visual displacement monitoring system and monitoring method thereof | |
| CN114639047A (en) | System and method for detecting Hainan gibbon field image based on convolutional neural network | |
| CN111216763B (en) | Rail crawling observation device | |
| CN220472741U (en) | Down-pass channel ponding monitoring device |
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 | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: A Multidimensional Intelligent Monitoring Method and System for Railway Track and Track Plate Status Effective date of registration: 20230522 Granted publication date: 20210504 Pledgee: Nanjing Zidong sub branch of Bank of Nanjing Co.,Ltd. Pledgor: Nanjing Paiguang Intelligence Perception Information Technology Co.,Ltd. Registration number: Y2023980041122 |