CN114581422A - Catenary contact line anomaly detection method and system based on image processing - Google Patents

Catenary contact line anomaly detection method and system based on image processing Download PDF

Info

Publication number
CN114581422A
CN114581422A CN202210231053.9A CN202210231053A CN114581422A CN 114581422 A CN114581422 A CN 114581422A CN 202210231053 A CN202210231053 A CN 202210231053A CN 114581422 A CN114581422 A CN 114581422A
Authority
CN
China
Prior art keywords
layer
contact line
image
cable
catenary contact
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
Application number
CN202210231053.9A
Other languages
Chinese (zh)
Other versions
CN114581422B (en
Inventor
杜长军
钱平慎
张栋波
张毅
杨才
王德圣
胡鼐
郑军
李越
朱广玉
彭松
孟凡金
武春波
李亮
李伟
沈翔
孙淼
张野
曹国欣
范宇
宋世禹
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jinzhou Power Supply Section Of China Railway Shenyang Bureau Group Co ltd
Shenyang Railway Science And Technology Research Institute Co ltd
Original Assignee
Jinzhou Power Supply Section Of China Railway Shenyang Bureau Group Co ltd
Shenyang Railway Science And Technology Research Institute Co ltd
Chengdu Nuobikan Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Jinzhou Power Supply Section Of China Railway Shenyang Bureau Group Co ltd, Shenyang Railway Science And Technology Research Institute Co ltd, Chengdu Nuobikan Technology Co ltd filed Critical Jinzhou Power Supply Section Of China Railway Shenyang Bureau Group Co ltd
Priority to CN202210231053.9A priority Critical patent/CN114581422B/en
Publication of CN114581422A publication Critical patent/CN114581422A/en
Application granted granted Critical
Publication of CN114581422B publication Critical patent/CN114581422B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration using local operators
    • G06T5/30—Erosion or dilatation, e.g. thinning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding
    • 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
    • G06T2207/20024—Filtering details
    • G06T2207/20032—Median filtering
    • 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
    • Y04S10/52—Outage or fault management, e.g. fault detection or location

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Quality & Reliability (AREA)
  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)

Abstract

The invention discloses a catenary contact line anomaly detection method and system based on image processing, which belong to the technical field of catenary contact line anomaly detection and comprise the following steps: acquiring a plurality of catenary contact line images shot by a contact line suspension state detection and monitoring device 4C; preprocessing each catenary contact line image to obtain a cable abnormal defect image set and a cable foreign matter suspension image set; respectively constructing a catenary contact line foreign body suspension detection network and a catenary contact line abnormal defect detection network; obtaining a trained foreign body suspension detection network and an abnormal defect detection network; carrying out anomaly detection on the catenary contact line image by using the trained foreign body suspension detection network and the trained abnormal defect detection network to obtain a cable abnormal defect detection result and a cable foreign body suspension detection result; the problem of messenger wire contact wire anomaly detection is realized through the high definition image that image processing utilized contact net suspension state to detect and monitoring devices 4C to shoot to this scheme.

Description

Catenary contact line anomaly detection method and system based on image processing
Technical Field
The invention belongs to the technical field of catenary contact line anomaly detection, and particularly relates to a catenary contact line anomaly detection method and system based on image processing.
Background
The high-speed railway contact net is the main power supply equipment of electrified railway, also is one of the most easy equipment that takes place the problem. In the operation of a high-speed rail, the pantograph is mainly used for getting electricity after contacting with a contact network, when the contact network is used for power transmission, a catenary cable is connected with the contact network line through pile heads, the pile heads are directly clamped on grooves on two sides of the contact network line, and in order to ensure driving safety, the contact network must be regularly detected and maintained to be in a good operation state.
The elastic simple suspension system with compensation device is equipped with elastic compensation device at the lower anchor of contact line to regulate the change of tension and sag of contact line of carrier cable. The method for carrying out abnormity detection on the catenary contact line through the image processing technology and the detection system corresponding to the method are urgently needed after the images shot by the catenary suspension state detection and monitoring device 4C with high efficiency and high definition are obtained.
Disclosure of Invention
Aiming at the defects in the prior art, the catenary contact line abnormity detection method and system based on image processing provided by the invention solve the problem of realizing catenary contact line abnormity detection by utilizing high-definition images shot by a catenary suspension state detection and monitoring device 4C through image processing.
In order to achieve the purpose of the invention, the invention adopts the technical scheme that:
the invention provides a catenary contact line abnormity detection method based on image processing, which comprises the following steps:
s1, acquiring a plurality of catenary contact line images shot by the contact line suspension state detection and monitoring device 4C;
s2, preprocessing each catenary contact line image to obtain a cable abnormal defect image set and a cable foreign matter hanging image set;
s3, respectively constructing a catenary contact line foreign body suspension detection network and a catenary contact line abnormal defect detection network;
s4, respectively training a catenary contact line foreign body suspension detection network and a catenary contact line abnormal defect detection network by using a cable abnormal defect image set and a cable foreign body suspension image set to obtain the trained catenary contact line foreign body suspension detection network and catenary contact line abnormal defect detection network;
and S5, respectively carrying out anomaly detection on the catenary contact line image subjected to the same pretreatment by using the trained catenary contact line foreign body suspension detection network and the trained catenary contact line anomaly defect detection network to obtain a cable anomaly defect detection result and a cable foreign body suspension detection result.
The invention has the beneficial effects that: the scheme can realize the line body abnormity detection and line body foreign body suspension detection of the catenary contact line by classifying the cable abnormity defects and processing the images, can detect the situations of line body cavities, stripping, bulges and large and small diameters and can detect the suspension situations of foreign bodies such as plastic bags, kites and the like at the same time, the method has the advantages of high detection precision, good detail sampling feature storage, low network construction cost and high retrieval efficiency.
Further, the step S2 includes the following steps:
s21, converting the contact line image format of each catenary into an RGB format by using a green interpolation method to obtain a catenary contact line image set in the RGB format;
s22, filtering the median of the catenary contact line images in the RGB format to obtain a filtered catenary contact line image set;
the expression of the filtered catenary contact line image is as follows:
g(x,y)＝Med{f(x-k,y-l),(k,l)∈W}
wherein g (x, y) represents a filtered catenary contact line image, f (x, y) represents a catenary contact line image in an RGB format, W represents an image calculation area, Med represents median filtering, k represents a filtered image window center abscissa, and l represents a filtered image window center ordinate;
s23, carrying out 2-magnification downsampling on each filtered catenary contact line image to obtain a sampled catenary contact line image set;
the calculation expression of the downsampling is as follows:
Figure BDA0003538377130000031
wherein, PkRepresenting sampled pixel point values, IiRepresenting the value of the original pixel point, s2Representing a square pixel area, win (k) representing the square pixel area where the original pixel point value is located, and i representing the ith original pixel point value;
s24, respectively obtaining a thick cable abnormal defect image set and a thick cable foreign matter hanging image set by carrying out threshold segmentation on each sampled catenary contact line image;
and S25, performing morphological processing on each thick cable abnormal defect image and each thick cable foreign body hanging image respectively to obtain a cable abnormal defect image set and a cable foreign body hanging image set.
The beneficial effect of adopting the further scheme is as follows: human eyes are most sensitive to green light reaction and weak to red light and blue light reaction, so that a catenary contact line collected by 4C is converted into an RGB format for image processing by a green interpolation method, salt and pepper noise in an image is filtered by median filtering, high detection rate and high detection accuracy are maintained by image downsampling processing, a cable line body is extracted from a background by threshold segmentation, discrete noise is filtered by morphological processing, and a cable abnormal defect image set and a cable foreign body suspension image set with clear characteristics are obtained.
Further, the step S25 includes the following sub-steps:
s251, performing expansion processing on each thick cable abnormal defect image and each thick cable foreign body hanging image to obtain an expanded thick cable abnormal defect image and a thick cable foreign body hanging image;
the computational expression of the dilation process is as follows:
Figure BDA0003538377130000041
wherein A and B respectively represent a set of elements to be processed and a set of structural elements of the region Z to be processed,
Figure BDA0003538377130000042
the structural element set B is represented to perform expansion operation on the element set A to be processed, z represents a displacement element, B' represents an origin reflection set of the structural element set,
Figure BDA0003538377130000043
representing an empty set;
s252, respectively carrying out corrosion treatment on each thick cable abnormal defect image and each thick cable foreign matter suspension image after expansion treatment to obtain a thick cable abnormal defect image set and a thick cable foreign matter suspension image set after corrosion treatment;
the calculation expression of the corrosion treatment is as follows:
Figure BDA0003538377130000044
wherein, theta represents that the structural element set B performs expansion operation on the element set A to be processed, and BzA set of structural displacement elements representing the region Z to be processed,
Figure BDA0003538377130000045
represents containing;
s253, respectively carrying out opening operation and closing operation on each thick cable abnormal defect image and each thick cable foreign matter hanging image after corrosion treatment to respectively obtain a cable abnormal defect image set and a cable foreign matter hanging image set;
the computational expressions of the opening operation and the closing operation are respectively as follows:
Figure BDA0003538377130000051
Figure BDA0003538377130000052
wherein,
Figure BDA0003538377130000053
indicates an open operation, and indicates a close operation.
The beneficial effect of adopting the further scheme is as follows: and the morphological processing filters the abnormal defect images of the thick cables and the discrete noise points of the foreign matter hanging images of the thick cables through expansion, corrosion, opening operation and closing operation to obtain the cable abnormal defect image set and the cable foreign matter hanging image set with clear characteristics.
Further, the catenary contact line anomaly defect detection network in step S3 includes:
the image processing device comprises a cable abnormal defect image input layer Inputs, a first Conv2D scrolling layer, a first BN layer, a first Relu active layer, a first residual module addlayer1, a first maximum pooling maxporoling 2D layer, a second Conv2D scrolling layer, a second BN layer, a second Relu active layer, a second residual module addlayer2, a second maximum pooling maxporoling 2D layer, a third Conv2D scrolling layer, a third BN layer, a third Relu active layer, a third maximum pooling maxporoling 2D layer, a third residual module addlayer3, a fourth Conv2D scrolling layer, a fourth BN layer, a regular fourth Relu active layer, a fourth residual module addlayer4, a fourth maximum pooling2 × flattening layer 2D, a fifth Conv2D scrolling layer, a fifth BN layer, a fifth Conv2 actuating layer, a regular fourth Relup active layer, a fifth residual module adduction 2 flattening layer 2D, a fifth Conv2 flattening layer, a fifth excess pooling layer, a first Relupo 2 flattening layer, a second Relu active layer, a second residual module flattening layer, a third residual flattening module flattening layer, a fifth residual flattening module flattening layer5, a fifth Conv2 flattening layer, a fifth Conv flattening layer, a fifth Conv2 flattening layer, a fifth flattening filter, a fifth filtering module flattening filter, a fifth filtering module flattening filter, a fifth filtering module flattening filter, a fifth filtering module flattening filter, a fifth filtering module flattening filter.
The beneficial effect of adopting the further scheme is as follows: the catenary contact line abnormal defect detection network learns through 5 BN layers, 5 relu activation functions, 4 pooling layers, 5 two-dimensional convolution layers and two Dense full-connection layers, prevents overfitting through a parameter regularization dropout layer, and can realize quick and high-precision detection of catenary contact line abnormal defects after training.
Further, the catenary contact line foreign body suspension detection network of step S3 includes:
the system comprises a cable foreign body hanging image input layer, a backbone network layer, a characteristic diagram acquisition layer, a region nomination network layer, a nomination region layer, an interested region pooling layer, a full connection layer and a category profile output layer;
the cable foreign matter suspension image input layer is connected with the backbone network layer; the backbone network layer is connected with the characteristic diagram acquisition layer; the characteristic diagram acquisition layer is respectively connected with the regional nomination network layer and the interested region pooling layer; the regional nomination network layer is connected with the nomination regional layer; the nomination region layer is connected with the interested region pooling layer; the region of interest pooling layer is connected with the full-link layer; the full connection layer is connected with the category profile output layer.
The beneficial effect of adopting the further scheme is as follows: the scheme readjusts all input cable foreign matter hanging images into a fixed size through a cable foreign matter hanging image input layer, directly outputs the cable foreign matter hanging images to an interested pooling layer through a characteristic image layer, and outputs the cable foreign matter hanging images to two paths of the interested pooling layer through a region nomination network layer and a nomination region layer to nominate a foreground target and a background target in a region, finally combines the nomination position, correspondingly finds out the position of a prediction target on a characteristic diagram, realizes regression of a prediction position frame through a full connection layer and a category contour output layer, and classifies the targets in the frame.
Furthermore, each convolution structure in the backbone network layer adopts a deformable convolution structure;
the convolution calculation expression of the deformable convolution structure is as follows:
Figure BDA0003538377130000061
wherein, yp(p0) Convolution output, p, representing a deformable convolution structure0Representing the convolution output position, pnRepresenting the convolution output position p0Corresponding integer offset, R represents a real number, Δ pnRepresenting the convolution output position p0Corresponding non-integer offset, w′(pn) Denotes the integer offset factor, x' (p)0+pn+Δpn) Representing the input to the deformable convolution structure.
The beneficial effect of adopting the further scheme is as follows:
further, the loss function L ({ p) in the area nomination network layer trainingi},{ti}) the expression is as follows:
Figure BDA0003538377130000071
Figure BDA0003538377130000072
Figure BDA0003538377130000073
Figure BDA0003538377130000074
wherein N isclsIndicates the number of classes, NregIndicating the number of nominated regions, i' indicating the ith anchor point, pi′Representing the probability that the prediction box is the foreground,
Figure BDA0003538377130000075
indicates the probability that the prediction block predicts correctly, ti′A parameter representing the boundary of the prediction box,
Figure BDA0003538377130000076
the boundary parameter of a real labeling frame of the foreground target is represented, the lambda represents a classification loss and prediction loss balance factor,
Figure BDA0003538377130000077
a function representing the loss of classification is represented,
Figure BDA0003538377130000078
represents a prediction loss function, x 'and y' represent the abscissa and ordinate, respectively, of the center point of the prediction box, w and h represent the width and height of the prediction box, respectively,
Figure BDA0003538377130000079
representing a smooth function.
The beneficial effect of adopting the further scheme is as follows: the comprehensive analysis of the classification loss and the prediction loss is realized through the classification loss and the prediction loss balance factor, and the high precision of the feature extraction of the regional nomination network layer is guaranteed.
The invention also provides a detection system of the catenary contact line abnormity detection method based on image processing, which comprises the following steps:
the catenary contact line image acquisition module is used for acquiring a plurality of catenary contact line images shot by the catenary suspension state detection and monitoring device 4C;
the image preprocessing module is used for preprocessing images of contact wires of each carrier cable to obtain a cable abnormal defect image set and a cable foreign matter hanging image set;
the detection network construction module is used for respectively constructing a catenary contact line foreign matter suspension detection network and a catenary contact line abnormal defect detection network;
the detection network training module is used for training a catenary contact line foreign matter suspension detection network and a catenary contact line abnormal defect detection network by utilizing the cable abnormal defect image set and the cable foreign matter suspension image set respectively to obtain the trained catenary contact line foreign matter suspension detection network and the trained catenary contact line abnormal defect detection network;
and the catenary contact line abnormity detection module is used for carrying out abnormity detection on the catenary contact line image subjected to the same pretreatment by utilizing the trained catenary contact line foreign matter suspension detection network and the trained catenary contact line abnormity defect detection network respectively to obtain a cable abnormity defect detection result and a cable foreign matter suspension detection result.
The beneficial effects of the invention are as follows: the detection system of the catenary contact line abnormity detection method based on image processing is a system correspondingly arranged to the catenary contact line abnormity detection method based on image processing and is used for realizing the catenary contact line abnormity detection method based on image processing.
Drawings
Fig. 1 is a flowchart of the steps of a catenary contact line anomaly detection method based on image processing in an embodiment of the present invention.
Fig. 2 is a schematic structural diagram of a catenary contact line anomaly defect detection network in the embodiment of the invention.
Fig. 3 is a schematic structural diagram of a catenary contact line foreign matter suspension detection network in the embodiment of the invention.
Fig. 4 is a structural diagram of a detection system of a catenary contact line anomaly detection method based on image processing in an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined by the appended claims, and all changes that can be made by the invention using the inventive concept are intended to be protected.
Example 1
As shown in fig. 1, in one embodiment of the present invention, the present invention provides a catenary contact line anomaly detection method based on image processing, which includes the following steps:
s1, acquiring a plurality of catenary contact line images shot by the contact line suspension state detection and monitoring device 4C;
s2, preprocessing each catenary contact line image to obtain a cable abnormal defect image set and a cable foreign matter hanging image set;
the step S2 includes the following steps:
s21, converting the format of each catenary contact line image into RGB format by using a green interpolation method to obtain a catenary contact line image set in RGB format;
s22, filtering median values of the catenary contact line images in the RGB formats to obtain filtered catenary contact line image sets;
the expression of the filtered catenary contact line image is as follows:
g(x,y)＝Med{f(x-k,y-l),(k,l)∈W}
wherein g (x, y) represents a filtered catenary contact line image, f (x, y) represents a catenary contact line image in an RGB format, W represents an image calculation area, Med represents median filtering, k represents a filtered image window center abscissa, and l represents a filtered image window center ordinate;
s23, carrying out 2-magnification downsampling on each filtered catenary contact line image to obtain a sampled catenary contact line image set;
the calculation expression of the downsampling is as follows:
Figure BDA0003538377130000091
wherein, PkRepresenting sampled pixel point values, IiRepresenting the value of the original pixel point, s2Representing a square pixel area, win (k) representing the square pixel area where the original pixel point value is located, and i representing the ith original pixel point value;
s24, respectively obtaining a thick cable abnormal defect image set and a thick cable foreign matter hanging image set by carrying out threshold segmentation on each sampled catenary contact line image;
s25, performing morphological processing on each thick cable abnormal defect image and each thick cable foreign body hanging image respectively to obtain a cable abnormal defect image set and a cable foreign body hanging image set;
the step S25 includes the following sub-steps:
s251, performing expansion processing on each thick cable abnormal defect image and each thick cable foreign body hanging image to obtain an expanded thick cable abnormal defect image and a thick cable foreign body hanging image;
the computational expression of the dilation process is as follows:
Figure BDA0003538377130000101
wherein A and B respectively represent a set of elements to be processed and a set of structural elements of the region Z to be processed,
Figure BDA0003538377130000102
representing the expansion operation of the structural element set B on the element set A to be processed, z representing a displacement element, B' representing an origin reflection set of the structural element set,
Figure BDA0003538377130000103
representing an empty set;
s252, respectively carrying out corrosion treatment on each thick cable abnormal defect image and each thick cable foreign matter suspension image after expansion treatment to obtain a thick cable abnormal defect image set and a thick cable foreign matter suspension image set after corrosion treatment;
the calculation expression of the corrosion treatment is as follows:
Figure BDA0003538377130000104
wherein, theta represents that the structural element set B performs expansion operation on the element set A to be processed, and BzA set of structural displacement elements representing the region Z to be processed,
Figure BDA0003538377130000105
represents containing;
s253, respectively performing opening operation and closing operation on each thick cable abnormal defect image and each thick cable foreign matter hanging image after corrosion treatment to respectively obtain a cable abnormal defect image set and a cable foreign matter hanging image set;
the computational expressions of the opening operation and the closing operation are respectively as follows:
Figure BDA0003538377130000111
Figure BDA0003538377130000112
wherein,
Figure BDA0003538377130000113
represents an open operation,. represents a close operation;
s3, respectively constructing a catenary contact line foreign body suspension detection network and a catenary contact line abnormal defect detection network;
as shown in fig. 3, the catenary contact line abnormal defect detection network in step S3 includes:
the system comprises a cable abnormal defect image input layer Inputs, a first Conv2D scrolling layer, a first BN layer, a first Relu activation layer, a first residual module addlayer1, a first maximum pooling maxporoling 2D layer, a second Conv2D scrolling layer, a second BN layer, a second Relu activation layer, a second residual module addlayer2, a second maximum pooling maxporoling 2D layer, a third Conv2D scrolling layer, a third BN layer, a third Relu activation layer, a third maximum pooling maxporoling 2D layer, a third residual module addlayer3, a fourth Conv2D scrolling layer, a fourth BN layer, a fourth Relu activation layer, a fourth residual module addlayer4, a fourth maximum pooling maxporoling 2 layer D, a fifth Conv2D scrolling layer, a fifth BN layer, a fifth Reuv activation layer, a fifth residual module adduction module addlayer4, a fourth maximum pooling2 flattening layer D, a fifth Conv2 scrolling layer, a fifth BN layer, a fifth Conv2, a fifth overflux activation layer, a fifth residual module flattening layer, a fifth recycling layer, a parameter input layer, a first Relu activation layer, a second residual module flattening module, a second residual module, a third residual module flattening 2 flattening layer, a fifth recycling module flattening layer, a fifth recycling parameter, a fifth recycling module flattening layer, a fifth recycling parameter, a fifth recycling module flattening layer, a fifth recycling module flattening layer, a fifth recycling parameter, a fifth recycling module flattening layer, a fifth recycling module flattening layer, a fifth recycling module flattening layer, a fifth recycling module flattening layer, a fifth recycling module flattening layer, a fifth recycling module flattening layer, a fifth recycling module flattening layer, a fifth recycling module flattening layer, a fifth recycling module flattening layer, a fifth recycling module flattening layer, a fifth recycling module flattening;
the abnormal cable defect images are 320 x 120 images, labels containing address information are added to the images respectively, the types of the labels are holes, stripping, bulges and sizes, and each image comprises more than 600 images;
as shown in fig. 4, the catenary contact line foreign body suspension detection network in step S3 includes:
the system comprises a cable foreign body hanging image input layer, a backbone network layer, a characteristic diagram acquisition layer, a regional nomination network layer, a nomination region layer, an interested region pooling layer, a full connection layer and a category profile output layer;
the cable foreign matter suspension image input layer is connected with the backbone network layer; the backbone network layer is connected with the characteristic diagram acquisition layer; the characteristic diagram acquisition layer is respectively connected with the regional nomination network layer and the region of interest pooling layer; the regional nomination network layer is connected with the nomination regional layer; the nomination region layer is connected with the interested region pooling layer; the region of interest pooling layer is connected with the full-link layer; the full connection layer is connected with the category profile output layer;
each convolution structure in the backbone network layer adopts a deformable convolution structure;
the convolution calculation expression of the deformable convolution structure is as follows:
Figure BDA0003538377130000121
wherein, yp(p0) Convolution output, p, representing a deformable convolution structure0Representing the convolution output position, pnRepresenting the convolution output position p0Corresponding integer offset, R represents a real number, Δ pnRepresenting the convolution output position p0Corresponding non-integer offset, w' (p)n) Denotes the integer offset factor, x' (p)0+pn+Δpn) An input representing a deformable convolution structure;
loss function L ({ p) in the training of the area nomination network layeri},{ti}) the expression is as follows:
Figure BDA0003538377130000122
Figure BDA0003538377130000123
Figure BDA0003538377130000124
Figure BDA0003538377130000125
wherein N isclsIndicates the number of classes, NregIndicating the number of nominated regions, i' indicating the ith anchor point, pi′Representing the probability that the prediction box is the foreground,
Figure BDA0003538377130000126
indicates the probability that the prediction block predicts correctly, ti′A parameter representing the boundary of the prediction box,
Figure BDA0003538377130000127
the boundary parameter of a real labeling frame of the foreground target is represented, the lambda represents a classification loss and prediction loss balance factor,
Figure BDA0003538377130000128
a function representing the loss of classification is represented,
Figure BDA0003538377130000129
represents a prediction loss function, x 'and y' represent the abscissa and ordinate, respectively, of the center point of the prediction box, w and h represent the width and height of the prediction box, respectively,
Figure BDA0003538377130000131
representing a smooth function.
S4, training a catenary contact line foreign matter suspension detection network and a catenary contact line abnormal defect detection network by utilizing the cable abnormal defect image set and the cable foreign matter suspension image set respectively to obtain the trained catenary contact line foreign matter suspension detection network and the trained catenary contact line abnormal defect detection network;
s5, respectively using the trained catenary contact line foreign matter suspension detection network and the trained catenary contact line abnormal defect detection network to perform abnormal detection on the catenary contact line image subjected to the same pretreatment, so as to obtain a cable abnormal defect detection result and a cable foreign matter suspension detection result;
the abnormal defect detection result of the cable comprises hole abnormality, stripping abnormality, bulge abnormality and size diameter abnormality, and the cable foreign matter suspension detection result is foreign matter suspension.
The invention has the beneficial effects that: the invention has the beneficial effects that: the scheme can realize the line body abnormity detection and line body foreign body suspension detection of the catenary contact line by classifying the cable abnormity defects and processing the images, can detect the situations of line body cavities, stripping, bulges and large and small diameters and can detect the suspension situations of foreign bodies such as plastic bags, kites and the like at the same time, the method has the advantages of high detection precision, good detail sampling feature storage, low network construction cost and high retrieval efficiency.
Example 2
As shown in fig. 4, in an embodiment of the present invention, the present invention provides a detection system of a catenary contact line anomaly detection method based on image processing, including:
the catenary contact line image acquisition module is used for acquiring a plurality of catenary contact line images shot by the catenary suspension state detection and monitoring device 4C;
the image preprocessing module is used for preprocessing images of contact wires of each carrier cable to obtain a cable abnormal defect image set and a cable foreign matter hanging image set;
the detection network construction module is used for respectively constructing a catenary contact line foreign body suspension detection network and a catenary contact line abnormal defect detection network;
the detection network training module is used for training a catenary contact line foreign matter suspension detection network and a catenary contact line abnormal defect detection network by utilizing the cable abnormal defect image set and the cable foreign matter suspension image set respectively to obtain the trained catenary contact line foreign matter suspension detection network and the trained catenary contact line abnormal defect detection network;
and the catenary contact line abnormity detection module is used for carrying out abnormity detection on the catenary contact line image subjected to the same pretreatment by utilizing the trained catenary contact line foreign matter suspension detection network and the trained catenary contact line abnormity defect detection network respectively to obtain a cable abnormity defect detection result and a cable foreign matter suspension detection result.
The detection system of the catenary contact line anomaly detection method based on image processing provided by the embodiment can execute the technical scheme shown in the catenary contact line anomaly detection method based on image processing in the embodiment of the method, and the implementation principle and the beneficial effects are similar, so that the details are not repeated here.
In the embodiment of the invention, the functional units can be divided according to the catenary contact line abnormality detection method based on image processing, for example, each function can be divided into each functional unit, and two or more functions can be integrated into one processing unit. The integrated unit may be implemented in the form of hardware, or may be implemented in the form of a software functional unit. It should be noted that the division of the cells in the present invention is schematic, and is only a logical division, and there may be another division manner in actual implementation.
In the embodiment of the invention, in order to realize the principle and the beneficial effect of the catenary contact line abnormity detection method based on image processing, the detection system of the catenary contact line abnormity detection method based on image processing comprises hardware structures and/or software modules corresponding to the execution of various functions. Those of skill in the art will readily appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as hardware and/or combinations of hardware and computer software, where a function is performed in a hardware or computer software-driven manner, and that the function described may be implemented in any suitable manner for each particular application depending upon the particular application and design constraints imposed on the technology, but such implementation is not to be considered as beyond the scope of the present application.

Claims (8)

1. A catenary contact line abnormity detection method based on image processing is characterized by comprising the following steps:
s1, acquiring a plurality of catenary contact line images shot by the contact line suspension state detection and monitoring device 4C;
s2, preprocessing each catenary contact line image to obtain a cable abnormal defect image set and a cable foreign matter hanging image set;
s3, respectively constructing a catenary contact line foreign body suspension detection network and a catenary contact line abnormal defect detection network;
s4, training a catenary contact line foreign matter suspension detection network and a catenary contact line abnormal defect detection network by utilizing the cable abnormal defect image set and the cable foreign matter suspension image set respectively to obtain the trained catenary contact line foreign matter suspension detection network and the trained catenary contact line abnormal defect detection network;
and S5, respectively carrying out anomaly detection on the catenary contact line image subjected to the same pretreatment by using the trained catenary contact line foreign body suspension detection network and the trained catenary contact line anomaly defect detection network to obtain a cable anomaly defect detection result and a cable foreign body suspension detection result.
2. The image processing-based catenary contact line anomaly detection method of claim 1, wherein said step S2 comprises the steps of:
s21, converting the contact line image format of each catenary into an RGB format by using a green interpolation method to obtain a catenary contact line image set in the RGB format;
s22, filtering the median of the catenary contact line images in the RGB format to obtain a filtered catenary contact line image set;
the expression of the filtered catenary contact line image is as follows:
g(x,y)＝Med{f(x-k,y-l),(k,l)∈W}
wherein g (x, y) represents a filtered catenary contact line image, f (x, y) represents a catenary contact line image in an RGB format, W represents an image calculation area, Med represents median filtering, k represents a filtered image window center abscissa, and l represents a filtered image window center ordinate;
s23, carrying out 2-magnification downsampling on each filtered catenary contact line image to obtain a sampled catenary contact line image set;
the downsampled computational expression is as follows:
Figure FDA0003538377120000021
wherein, PkRepresenting sampled pixel point values, IiRepresenting the value of the original pixel point, s2Representing a square pixel area, win (k) representing the square pixel area where the original pixel point value is located, and i representing the ith original pixel point value;
s24, respectively obtaining a thick cable abnormal defect image set and a thick cable foreign matter hanging image set by carrying out threshold segmentation on each sampled catenary contact line image;
and S25, performing morphological processing on each thick cable abnormal defect image and each thick cable foreign body hanging image respectively to obtain a cable abnormal defect image set and a cable foreign body hanging image set.
3. The image processing-based catenary contact line anomaly detection method of claim 2, wherein said step S25 comprises the sub-steps of:
s251, performing expansion processing on each thick cable abnormal defect image and each thick cable foreign body hanging image to obtain an expanded thick cable abnormal defect image and a thick cable foreign body hanging image;
the computational expression of the dilation process is as follows:
Figure FDA0003538377120000022
wherein A and B respectively represent a to-be-processed element set and a structural element set of a to-be-processed region Z,
Figure FDA0003538377120000023
the structural element set B is represented to perform expansion operation on the element set A to be processed, z represents a displacement element, B' represents an origin reflection set of the structural element set,
Figure FDA0003538377120000024
representing an empty set;
s252, respectively carrying out corrosion treatment on each thick cable abnormal defect image and each thick cable foreign matter suspension image after expansion treatment to obtain a thick cable abnormal defect image set and a thick cable foreign matter suspension image set after corrosion treatment;
the calculation expression of the corrosion treatment is as follows:
Figure FDA0003538377120000031
wherein, theta represents that the structural element set B performs expansion operation on the element set A to be processed, and BzA set of structural displacement elements representing the region Z to be processed,
Figure FDA0003538377120000032
represents a compound containing;
s253, respectively carrying out opening operation and closing operation on each thick cable abnormal defect image and each thick cable foreign matter hanging image after corrosion treatment to respectively obtain a cable abnormal defect image set and a cable foreign matter hanging image set;
the computational expressions of the opening operation and the closing operation are respectively as follows:
Figure FDA0003538377120000033
Figure FDA0003538377120000034
wherein,
Figure FDA0003538377120000035
indicates an open operation, and indicates a close operation.
4. The image processing-based catenary contact line anomaly detection method of claim 1, wherein the catenary contact line anomaly defect detection network of step S3 comprises:
the image processing device comprises a cable abnormal defect image input layer Inputs, a first Conv2D scrolling layer, a first BN layer, a first Relu active layer, a first residual module addlayer1, a first maximum pooling maxporoling 2D layer, a second Conv2D scrolling layer, a second BN layer, a second Relu active layer, a second residual module addlayer2, a second maximum pooling maxporoling 2D layer, a third Conv2D scrolling layer, a third BN layer, a third Relu active layer, a third maximum pooling maxporoling 2D layer, a third residual module addlayer3, a fourth Conv2D scrolling layer, a fourth BN layer, a regular fourth Relu active layer, a fourth residual module addlayer4, a fourth maximum pooling2 × flattening layer 2D, a fifth Conv2D scrolling layer, a fifth BN layer, a fifth Conv2 actuating layer, a regular fourth Relup active layer, a fifth residual module adduction 2 flattening layer 2D, a fifth Conv2 flattening layer, a fifth excess pooling layer, a first Relupo 2 flattening layer, a second Relu active layer, a second residual module flattening layer, a third residual flattening module flattening layer, a fifth residual flattening module flattening layer5, a fifth Conv2 flattening layer, a fifth Conv flattening layer, a fifth Conv2 flattening layer, a fifth flattening filter, a fifth filtering module flattening filter, a fifth filtering module flattening filter, a fifth filtering module flattening filter, a fifth filtering module flattening filter, a fifth filtering module flattening filter.
5. The image processing-based catenary contact line anomaly detection method of claim 1, wherein the catenary contact line foreign body suspension detection network of step S3 comprises:
the system comprises a cable foreign body hanging image input layer, a backbone network layer, a characteristic diagram acquisition layer, a regional nomination network layer, a nomination region layer, an interested region pooling layer, a full connection layer and a category profile output layer;
the cable foreign matter suspension image input layer is connected with the backbone network layer; the backbone network layer is connected with the characteristic diagram acquisition layer; the characteristic diagram acquisition layer is respectively connected with the regional nomination network layer and the region of interest pooling layer; the regional nomination network layer is connected with the nomination regional layer; the nomination region layer is connected with the interested region pooling layer; the region of interest pooling layer is connected with the full-link layer; the full connection layer is connected with the category profile output layer.
6. The method of claim 5, wherein each convolution structure in the backbone network layer is a deformable convolution structure;
the convolution calculation expression of the deformable convolution structure is as follows:
Figure FDA0003538377120000041
wherein, yp(p0) Convolution output, p, representing a deformable convolution structure0Representing the convolution output position, pnRepresenting the convolution output position p0Corresponding integer offset, R represents a real number, Δ pnRepresenting the convolution output position p0Corresponding non-integer offset, w' (p)n) Denotes the integer offset factor, x' (p)0+pn+Δpn) Representing the input to the deformable convolution structure.
7. The method for detecting catenary contact line anomaly based on image processing as claimed in claim 5, wherein the loss function L ({ p) in the area nomination network layer trainingi},{ti}) the expression is as follows:
Figure FDA0003538377120000042
Figure FDA0003538377120000043
Figure FDA0003538377120000044
Figure FDA0003538377120000051
wherein, NclsIndicates the number of classes, NregIndicating the number of nominated regions, i' indicating the ith anchor point, pi′Representing the probability that the prediction box is the foreground,
Figure FDA0003538377120000052
indicates the probability that the prediction block predicts correctly, ti′A parameter representing the boundary of the prediction box,
Figure FDA0003538377120000053
real labeling frame boundary parameters representing foreground targets, lambda represents a classification loss and prediction loss balance factor,
Figure FDA0003538377120000054
a function representing the loss of classification is represented,
Figure FDA0003538377120000055
represents a prediction loss function, x 'and y' represent the abscissa and ordinate, respectively, of the center point of the prediction box, w and h represent the width and height of the prediction box, respectively,
Figure FDA0003538377120000056
representing a smooth function.
8. An image processing-based catenary contact line anomaly detection system as claimed in claims 1-7, comprising:
the catenary contact line image acquisition module is used for acquiring a plurality of catenary contact line images shot by the catenary suspension state detection and monitoring device 4C;
the image preprocessing module is used for preprocessing images of contact wires of each carrier cable to obtain a cable abnormal defect image set and a cable foreign matter hanging image set;
the detection network construction module is used for respectively constructing a catenary contact line foreign body suspension detection network and a catenary contact line abnormal defect detection network;
the detection network training module is used for training a catenary contact line foreign matter suspension detection network and a catenary contact line abnormal defect detection network by utilizing the cable abnormal defect image set and the cable foreign matter suspension image set respectively to obtain the trained catenary contact line foreign matter suspension detection network and the trained catenary contact line abnormal defect detection network;
and the catenary contact line abnormity detection module is used for carrying out abnormity detection on the catenary contact line image subjected to the same pretreatment by utilizing the trained catenary contact line foreign matter suspension detection network and the trained catenary contact line abnormity defect detection network respectively to obtain a cable abnormity defect detection result and a cable foreign matter suspension detection result.
CN202210231053.9A 2022-03-09 2022-03-09 Carrier cable contact line anomaly detection method and system based on image processing Active CN114581422B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210231053.9A CN114581422B (en) 2022-03-09 2022-03-09 Carrier cable contact line anomaly detection method and system based on image processing

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210231053.9A CN114581422B (en) 2022-03-09 2022-03-09 Carrier cable contact line anomaly detection method and system based on image processing

Publications (2)

Publication Number Publication Date
CN114581422A true CN114581422A (en) 2022-06-03
CN114581422B CN114581422B (en) 2024-06-28

Family

ID=81778957

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210231053.9A Active CN114581422B (en) 2022-03-09 2022-03-09 Carrier cable contact line anomaly detection method and system based on image processing

Country Status (1)

Country Link
CN (1) CN114581422B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115272831A (en) * 2022-09-27 2022-11-01 成都中轨轨道设备有限公司 Transmission method and system for monitoring images of suspension state of contact network
CN117576635A (en) * 2024-01-17 2024-02-20 中国石油集团川庆钻探工程有限公司 Method for judging linear target tensioning state in video identification

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20190004173A (en) * 2017-07-03 2019-01-11 한국철도기술연구원 Apparatus and method for monitoring of pantograph in electric railway
US20210272266A1 (en) * 2020-02-27 2021-09-02 North China Institute of Aerospace Engineering Automatic Interpretation Method for Winter Wheat Based on Deformable Fully Convolutional Neural Network
CN113870230A (en) * 2021-09-30 2021-12-31 湖南大学 Surface anomaly detection method based on hybrid supervised learning
CN114030395A (en) * 2021-11-05 2022-02-11 中铁工程设计咨询集团有限公司 Method and system for detecting foreign matters in contact suspension string area

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20190004173A (en) * 2017-07-03 2019-01-11 한국철도기술연구원 Apparatus and method for monitoring of pantograph in electric railway
US20210272266A1 (en) * 2020-02-27 2021-09-02 North China Institute of Aerospace Engineering Automatic Interpretation Method for Winter Wheat Based on Deformable Fully Convolutional Neural Network
CN113870230A (en) * 2021-09-30 2021-12-31 湖南大学 Surface anomaly detection method based on hybrid supervised learning
CN114030395A (en) * 2021-11-05 2022-02-11 中铁工程设计咨询集团有限公司 Method and system for detecting foreign matters in contact suspension string area

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115272831A (en) * 2022-09-27 2022-11-01 成都中轨轨道设备有限公司 Transmission method and system for monitoring images of suspension state of contact network
CN115272831B (en) * 2022-09-27 2022-12-09 成都中轨轨道设备有限公司 Transmission method and system for monitoring images of suspension state of contact network
CN117576635A (en) * 2024-01-17 2024-02-20 中国石油集团川庆钻探工程有限公司 Method for judging linear target tensioning state in video identification
CN117576635B (en) * 2024-01-17 2024-03-29 中国石油集团川庆钻探工程有限公司 Method for judging linear target tensioning state in video identification

Also Published As

Publication number Publication date
CN114581422B (en) 2024-06-28

Similar Documents

Publication Publication Date Title
CN110598736B (en) Power equipment infrared image fault positioning, identifying and predicting method
CN114581422A (en) Catenary contact line anomaly detection method and system based on image processing
Yan et al. YOLO V3+ VGG16-based automatic operations monitoring and analysis in a manufacturing workshop under Industry 4.0
CN110427871B (en) Fatigue driving detection method based on computer vision
US20210182615A1 (en) Alexnet-based insulator self-explosion recognition method
CN112950576B (en) Power transmission line defect intelligent identification method and system based on deep learning
CN111598855B (en) 2C equipment high-speed rail contact net dropper defect detection method based on deep learning and transfer learning
CN114202540B (en) Intelligent detection method for split pin defect of high-speed rail contact network
CN111008632A (en) License plate character segmentation method based on deep learning
CN111488805A (en) Video behavior identification method based on saliency feature extraction
CN111767826B (en) Timing and fixed-point scene anomaly detection method
CN111199255A (en) Small target detection network model and detection method based on dark net53 network
CN110599458A (en) Underground pipe network detection and evaluation cloud system based on convolutional neural network
CN116958786A (en) Dynamic visual identification method for chemical waste residues based on YOLOv5 and ResNet50 neural network
CN116680624B (en) Classification method, system and storage medium for metadata of power system
CN112613564A (en) Target detection post-processing method for eliminating overlapped frames
CN111401139A (en) Method for obtaining position of underground mine equipment based on character image intelligent identification
CN114140731B (en) Traction substation abnormality detection method
CN116363411A (en) Method and device for judging opening and closing states of isolating switch based on event camera
CN114372556A (en) Driving danger scene identification method based on lightweight multi-modal neural network
CN112529881B (en) Power control cabinet cable anomaly identification method and device
CN115240163A (en) Traffic sign detection method and system based on one-stage detection network
CN109614898B (en) Intelligent judging method for train running direction detection
CN117830633B (en) Railway scene segmentation method based on feature alignment and edge constraint
Guo et al. Defect detection method of high-speed rail catenary based on few-shot learning

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
TA01 Transfer of patent application right
TA01 Transfer of patent application right

Effective date of registration: 20230316

Address after: No. 3 Zhengdali, Linghe District, Jinzhou City, Liaoning Province, 121000

Applicant after: Jinzhou power supply section of China Railway Shenyang Bureau Group Co.,Ltd.

Applicant after: Shenyang Railway Science and Technology Research Institute Co.,Ltd.

Address before: 610015 No. 1534, floor 15, building 1, No. 88, Guanghua South Third Road, Qingyang District, Chengdu, Sichuan Province

Applicant before: CHENGDU NUOBIKAN TECHNOLOGY Co.,Ltd.

Applicant before: Jinzhou power supply section of China Railway Shenyang Bureau Group Co.,Ltd.

Applicant before: Shenyang Railway Science and Technology Research Institute Co.,Ltd.

GR01 Patent grant
GR01 Patent grant