CN114608801B - Automatic detection algorithm for falling off of connecting wire of locomotive shaft temperature probe - Google Patents
Automatic detection algorithm for falling off of connecting wire of locomotive shaft temperature probe Download PDFInfo
- Publication number
- CN114608801B CN114608801B CN202011441929.XA CN202011441929A CN114608801B CN 114608801 B CN114608801 B CN 114608801B CN 202011441929 A CN202011441929 A CN 202011441929A CN 114608801 B CN114608801 B CN 114608801B
- Authority
- CN
- China
- Prior art keywords
- temperature probe
- picture
- detection algorithm
- deep learning
- automatic detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000000523 sample Substances 0.000 title claims abstract description 64
- 238000001514 detection method Methods 0.000 title claims abstract description 45
- 230000003137 locomotive effect Effects 0.000 title claims abstract description 22
- 238000013135 deep learning Methods 0.000 claims abstract description 40
- 230000006870 function Effects 0.000 claims description 11
- 230000002159 abnormal effect Effects 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 4
- 230000001629 suppression Effects 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 2
- 238000000034 method Methods 0.000 abstract description 5
- 230000005856 abnormality Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 2
- 229920000742 Cotton Polymers 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 208000003464 asthenopia Diseases 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M11/00—Testing of optical apparatus; Testing structures by optical methods not otherwise provided for
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M17/00—Testing of vehicles
- G01M17/08—Railway vehicles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an automatic detection algorithm for the falling-off of a connecting wire of a locomotive shaft temperature probe, and relates to the technical field of train detection. The method comprises the following steps: s1, detecting a snap-shot image axle box and an axle temperature probe connector; s2, acquiring a grabbing picture at the bottom of the train and a historical archiving picture; s3, adopting a rectangular target detection algorithm based on deep learning for the two pictures; s4, positioning the rectangular area of the axle box and the probe connector; and S5, sending the rectangular frame coordinate information and the train picture into a rectangular target detection network based on deep learning for training. The automatic detection algorithm for the falling off of the connecting wire of the locomotive axle temperature probe judges whether the axle temperature probe connector has a loosening fault or not through the intelligent positioning and comparison algorithm by deep learning, has great advantages in accuracy and instantaneity compared with a method for manually checking pictures by utilizing the intelligent detection algorithm by deep learning, saves labor cost and improves train checking efficiency.
Description
Technical Field
The invention relates to the technical field of train detection, in particular to an automatic detection algorithm for locomotive shaft temperature probe connecting wire falling.
Background
The axle temperature detector is an important part on a train, the axle of the train adopts a bush type sliding bearing, an oil-containing cotton yarn roll is adopted in an axle box to lubricate the bearing, and because of the high-speed running of the train, if the bush lacks oil or is mixed with impurities, high temperature is easily generated until the axle box burns due to friction, commonly called as the axle burning, the serious axle burning can cause axle cutting and even the train overturns, and the axle temperature sensor can send an alarm signal when the axle temperature exceeds a certain temperature, so that accidents are prevented.
In the recent locomotive fault accidents, the release accident of the shaft temperature probe connector frequently happens, the detection of the release fault of the shaft temperature probe connector is generally completed by manually checking the car bottom picture, and the detection rate is lower due to the large data volume and the manual easiness of interference of visual fatigue, so that the development of the automatic release detection method of the shaft temperature probe connector is an important work.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an automatic detection algorithm for the falling-off of a connecting wire of a locomotive shaft temperature probe, so as to solve the problems in the prior art.
In order to achieve the above purpose, the present invention provides the following technical solutions: an automatic detection algorithm for the falling off of a connecting wire of a locomotive shaft temperature probe comprises the following steps:
s1, detecting a snap-shot image axle box and an axle temperature probe connector;
S2, acquiring a grabbing picture at the bottom of the train and a historical archiving picture;
s3, adopting a rectangular target detection algorithm based on deep learning for the two pictures;
S4, positioning the rectangular area of the axle box and the probe connector;
S5, sending the rectangular frame coordinate information and the train pictures into a rectangular target detection network based on deep learning for training;
S6, predicting a rectangular frame candidate region of the target picture by using the model obtained through training;
s7, performing NMS non-maximum suppression on the candidate areas, and filtering by using a threshold according to the set confidence level;
S8, outputting final rectangular frame target area information when the confidence coefficient of the predicted rectangular frame target area is greater than a set threshold value theta;
S9, acquiring an archive picture and a current snap shot picture;
s10, aligning the file pictures and the snap shots according to the axle box area;
S11, simultaneously transmitting the file map and the shaft temperature probe connector region of the snap map to a twin network extraction feature based on deep learning;
S12, comparing the positioning area of the archival picture shaft temperature probe connector with the positioning area of the snap shot picture shaft temperature probe connector, and judging whether the archival picture shaft temperature probe connector is abnormal or not.
Further optimizing the technical scheme, in the step S3, a rectangular target detection algorithm based on deep learning needs to collect pictures containing the axle box region of the train.
In the step S3, the rectangular target detection algorithm based on deep learning is used for marking out the areas of the axle box and the axle temperature probe connector by using rectangular frames.
Further optimizing the technical scheme, in 11, based on the deep learning twin network, the loss function during twin network training is:
Loss1＝error_same(x1，x2)；
Loss2＝error_different(x1，x2)；
Loss＝loss1-loss2+α。
further optimizing the technical scheme, in 11, based on the deep learning twin network, when the historical picture and the current picture have no difference, the loss function is required to be minimized, which is equivalent to minimizing loss1 as much as possible, and the network can be understood as having the strongest ability of identifying two pictures as the same picture.
Further optimizing the technical scheme, in 11, when the historical picture and the current picture are different based on the deep learning twin network, the loss2 is as large as possible, and the ability of the network to distinguish the picture from each other is understood to be as strong as possible.
Further, in the above-mentioned 11, the parameter α is set to avoid the loss function being 0 based on the deep learning twin network.
Further optimizing the technical scheme, in the 11, based on the deep learning twin network, similarity is calculated on the extracted features, a threshold is set, and whether the difference exists is compared.
Compared with the prior art, the invention provides an automatic detection algorithm for the falling off of the connecting wire of the locomotive shaft temperature probe, which has the following beneficial effects:
1. the automatic detection algorithm for the falling-off of the connecting wire of the locomotive shaft temperature probe adopts a rectangular target detection algorithm based on deep learning, compares the positioning area of the shaft temperature probe connector of the archival picture with the positioning area of the shaft temperature probe connector of the snap shot picture, judges whether the abnormality exists, and gives an alarm if the abnormality exists.
2. According to the automatic detection algorithm for the falling-off of the connecting line of the locomotive shaft temperature probe, a twin network based on deep learning is adopted for comparison, a shaft temperature probe connector area of an archive image and a shaft temperature probe connector area of a snap image are simultaneously sent to the twin network to extract characteristics, similarity is calculated for the extracted characteristics, a threshold value is set, whether the comparison is different or not is judged, and if the comparison is abnormal, an alarm is given.
3. The automatic detection algorithm for the falling off of the connecting wire of the locomotive axle temperature probe judges whether the axle temperature probe connector has a loosening fault or not through the intelligent positioning and comparison algorithm by deep learning, has great advantages in accuracy and instantaneity compared with a method for manually checking pictures by utilizing the intelligent detection algorithm by deep learning, saves labor cost and improves train checking efficiency.
Detailed Description
The technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention, and it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Embodiment one:
An automatic detection algorithm for the falling off of a connecting wire of a locomotive shaft temperature probe comprises the following steps:
s1, detecting a snap-shot image axle box and an axle temperature probe connector;
S2, acquiring a grabbing picture at the bottom of the train and a historical archiving picture;
S3, adopting a rectangular target detection algorithm based on deep learning for the two pictures, and firstly collecting pictures containing the axle box region of the train, and marking out the axle box and the region where the axle temperature probe connector is located by using a rectangular frame;
S4, positioning the rectangular area of the axle box and the probe connector;
S5, sending the rectangular frame coordinate information and the train pictures into a rectangular target detection network based on deep learning for training;
S6, predicting a rectangular frame candidate region of the target picture by using the model obtained through training;
s7, performing NMS non-maximum suppression on the candidate areas, and filtering by using a threshold according to the set confidence level;
S8, outputting final rectangular frame target area information when the confidence coefficient of the predicted rectangular frame target area is greater than a set threshold value theta;
S9, acquiring an archive picture and a current snap shot picture;
s10, aligning the file pictures and the snap shots according to the axle box area;
S11, simultaneously transmitting the file map and the shaft temperature probe connector region of the snap map to a twin network based on deep learning to extract characteristics, wherein a loss function during twin network training based on the deep learning is as follows:
Loss1＝error_same(x1，x2)；
Loss2＝error_different(x1，x2)；
Loss＝loss1-loss2+α；
when there is no difference between the historical picture and the current picture, the loss function is required to be minimized, which is equivalent to minimizing loss1 as much as possible, and it can be understood that the ability of the network to identify two pictures as the same picture is as strong as possible, the parameter α is set to avoid the loss function being 0, the similarity is calculated for the extracted features, the threshold is set, and whether there is a difference is compared.
S12, comparing the positioning area of the archival picture shaft temperature probe connector with the positioning area of the snap shot picture shaft temperature probe connector, and judging whether the archival picture shaft temperature probe connector is abnormal or not.
Embodiment two:
An automatic detection algorithm for the falling off of a connecting wire of a locomotive shaft temperature probe comprises the following steps:
s1, detecting a snap-shot image axle box and an axle temperature probe connector;
S2, acquiring a grabbing picture at the bottom of the train and a historical archiving picture;
S3, adopting a rectangular target detection algorithm based on deep learning for the two pictures, and firstly collecting pictures containing the axle box region of the train, and marking out the axle box and the region where the axle temperature probe connector is located by using a rectangular frame;
S4, positioning the rectangular area of the axle box and the probe connector;
S5, sending the rectangular frame coordinate information and the train pictures into a rectangular target detection network based on deep learning for training;
S6, predicting a rectangular frame candidate region of the target picture by using the model obtained through training;
s7, performing NMS non-maximum suppression on the candidate areas, and filtering by using a threshold according to the set confidence level;
S8, outputting final rectangular frame target area information when the confidence coefficient of the predicted rectangular frame target area is greater than a set threshold value theta;
S9, acquiring an archive picture and a current snap shot picture;
s10, aligning the file pictures and the snap shots according to the axle box area;
S11, simultaneously transmitting the file map and the shaft temperature probe connector region of the snap map to a twin network based on deep learning to extract characteristics, wherein a loss function during twin network training based on the deep learning is as follows:
Loss1＝error_same(x1，x2)；
Loss2＝error_different(x1，x2)；
Loss＝loss1-loss2+α；
When the historical picture and the current picture are different, the method is equivalent to making loss2 as large as possible, and can be understood that the capability of the network for distinguishing the picture with the difference is as strong as possible, the parameter alpha is set for avoiding the loss function being 0, the similarity is calculated for the extracted characteristics, the threshold is set, and whether the difference exists or not is compared.
S12, comparing the positioning area of the archival picture shaft temperature probe connector with the positioning area of the snap shot picture shaft temperature probe connector, and judging whether the archival picture shaft temperature probe connector is abnormal or not.
The beneficial effects of the invention are as follows: a rectangular target detection algorithm based on deep learning is adopted, the positioning area of the archival picture shaft temperature probe connector and the positioning area of the snap shot picture shaft temperature probe connector are compared, whether abnormality exists or not is judged, and if abnormality exists, an alarm is given; adopting twin network comparison based on deep learning, simultaneously transmitting the shaft temperature probe connector areas of the archive image and the snap image to the twin network to extract features, calculating similarity of the extracted features, setting a threshold value, comparing whether differences exist, and giving an alarm if the differences exist; the intelligent positioning and comparison algorithm is used for judging whether the shaft temperature probe connector has loose faults or not through deep learning, and compared with a method for manually checking pictures, the intelligent detection algorithm utilizing the deep learning has great advantages in accuracy and instantaneity, saves labor cost and improves train checking efficiency.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (8)
1. An automatic detection algorithm for the falling off of a connecting wire of a locomotive shaft temperature probe is characterized by comprising the following steps:
s1, detecting a snap-shot image axle box and an axle temperature probe connector;
S2, acquiring a grabbing picture at the bottom of the train and a historical archiving picture;
s3, adopting a rectangular target detection algorithm based on deep learning for the two pictures;
S4, positioning the rectangular area of the axle box and the probe connector;
S5, sending the rectangular frame coordinate information and the train pictures into a rectangular target detection network based on deep learning for training;
S6, predicting a rectangular frame candidate region of the target picture by using the model obtained through training;
s7, performing NMS non-maximum suppression on the candidate areas, and filtering by using a threshold according to the set confidence level;
S8, outputting final rectangular frame target area information when the confidence coefficient of the predicted rectangular frame target area is greater than a set threshold value theta;
S9, acquiring an archive picture and a current snap shot picture;
s10, aligning the file pictures and the snap shots according to the axle box area;
S11, simultaneously transmitting the file map and the shaft temperature probe connector region of the snap map to a twin network extraction feature based on deep learning;
S12, comparing the positioning area of the archival picture shaft temperature probe connector with the positioning area of the snap shot picture shaft temperature probe connector, and judging whether the archival picture shaft temperature probe connector is abnormal or not.
2. The automatic detection algorithm for locomotive axle temperature probe connecting line drop-out according to claim 1, wherein in S3, a picture including a train axle box area is collected first based on a rectangular target detection algorithm of deep learning.
3. The automatic detection algorithm for the disconnection of the connecting wire of the locomotive axle temperature probe according to claim 1, wherein in the step S3, the rectangular target detection algorithm based on deep learning is used for marking the areas where the axle box and the axle temperature probe connector are located by using rectangular boxes.
4. The automatic detection algorithm for locomotive axle temperature probe connecting line drop according to claim 1, wherein in 11, a twin network based on deep learning is provided, and a loss function during twin network training is:
Loss1＝error_same(x1，x2)；
Loss2＝error_different(x1，x2)；
Loss＝loss1-loss2+α。
5. The automatic detection algorithm for locomotive axle temperature probe connecting line drop-out according to claim 1, wherein in 11, based on a deep learning twin network, when there is no difference between a history picture and a current picture, the loss function is to be minimized, which is equivalent to minimizing loss1 as small as possible, and it can be understood that the ability of the network to identify two pictures as the same picture is as strong as possible.
6. The automatic detection algorithm for locomotive axle temperature probe connecting line drop according to claim 1, wherein in 11, based on the deep learning twin network, when the historical picture is different from the current picture, the ability of the network to distinguish the picture as much as possible is understood as the strong as possible.
7. The automatic detection algorithm for locomotive axle temperature probe connection line drop-out according to claim 4, wherein in 11, the parameter α is set to avoid a loss function of 0 based on a deep learning twin network.
8. The automatic detection algorithm for locomotive axle temperature probe connecting line drop-out according to claim 1, wherein in 11, based on a deep learning twin network, similarity is calculated for the extracted features, a threshold is set, and whether there is a difference is compared.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011441929.XA CN114608801B (en) | 2020-12-08 | 2020-12-08 | Automatic detection algorithm for falling off of connecting wire of locomotive shaft temperature probe |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011441929.XA CN114608801B (en) | 2020-12-08 | 2020-12-08 | Automatic detection algorithm for falling off of connecting wire of locomotive shaft temperature probe |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114608801A CN114608801A (en) | 2022-06-10 |
| CN114608801B true CN114608801B (en) | 2024-04-19 |
Family
ID=81856723
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011441929.XA Active CN114608801B (en) | 2020-12-08 | 2020-12-08 | Automatic detection algorithm for falling off of connecting wire of locomotive shaft temperature probe |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114608801B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0848247A (en) * | 1994-08-09 | 1996-02-20 | Central Japan Railway Co | Train approach warning device |
| CN111160407A (en) * | 2019-12-10 | 2020-05-15 | 重庆特斯联智慧科技股份有限公司 | Deep learning target detection method and system |
| CN111191546A (en) * | 2019-12-20 | 2020-05-22 | 广西柳州联耕科技有限公司 | Intelligent product assembling method based on machine vision recognition |
| CN111402211A (en) * | 2020-03-04 | 2020-07-10 | 广西大学 | High-speed train bottom foreign matter identification method based on deep learning |
| CN111680689A (en) * | 2020-08-11 | 2020-09-18 | 武汉精立电子技术有限公司 | Target detection method, system and storage medium based on deep learning |
-
2020
- 2020-12-08 CN CN202011441929.XA patent/CN114608801B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0848247A (en) * | 1994-08-09 | 1996-02-20 | Central Japan Railway Co | Train approach warning device |
| CN111160407A (en) * | 2019-12-10 | 2020-05-15 | 重庆特斯联智慧科技股份有限公司 | Deep learning target detection method and system |
| CN111191546A (en) * | 2019-12-20 | 2020-05-22 | 广西柳州联耕科技有限公司 | Intelligent product assembling method based on machine vision recognition |
| CN111402211A (en) * | 2020-03-04 | 2020-07-10 | 广西大学 | High-speed train bottom foreign matter identification method based on deep learning |
| CN111680689A (en) * | 2020-08-11 | 2020-09-18 | 武汉精立电子技术有限公司 | Target detection method, system and storage medium based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 基于深度卷积神经网络的遥感图像目标检测方法;刘思远;程良伦;;工业控制计算机(05);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114608801A (en) | 2022-06-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104899936B (en) | A kind of photovoltaic module fault cues method and system based on image recognition | |
| CN111024416B (en) | Fault diagnosis method and system for train traction system | |
| CN116394794B (en) | High security intelligence fills electric pile | |
| CN116435634A (en) | Storage battery temperature state monitoring and management system | |
| CN116227788A (en) | Fault diagnosis and evaluation method and system for substation power equipment | |
| CN113418730A (en) | Online monitoring method for operating state of cigarette making machine | |
| CN115372816B (en) | Power distribution switchgear operation fault prediction system and method based on data analysis | |
| CN117250576B (en) | Current sensor real-time abnormality detection method based on multidimensional sensing data | |
| CN114608801B (en) | Automatic detection algorithm for falling off of connecting wire of locomotive shaft temperature probe | |
| CN111371647A (en) | Data center monitoring data preprocessing method and device | |
| CN104749467A (en) | Distribution terminal self-diagnostic method and system | |
| CN113899973A (en) | Power generation equipment fault diagnosis method based on infrared thermal imaging and visible light image | |
| CN111076750B (en) | Device and method for detecting looseness of catenary dropper wire clamp | |
| CN112285631A (en) | Judgment method and system for active early warning of current transformer | |
| CN117041088A (en) | Network equipment temperature monitoring alarm method, system and storage medium | |
| CN112343657A (en) | Mine underground air door identification and abnormal state monitoring method | |
| CN113359639B (en) | Factory equipment monitoring method and system based on safety detection robot | |
| CN104504713A (en) | Automatic trouble identification method based on TEDS (trouble of moving EMU detection system) | |
| CN113992151A (en) | Method, device and equipment for determining working state of photovoltaic array and storage medium | |
| CN110599030B (en) | Electric power risk early warning method based on positive and negative feature library weighted matching | |
| CN114663332A (en) | Automatic detection method for oil leakage fault of locomotive oil pressure shock absorber | |
| CN217238912U (en) | Filter plate breakage monitoring and alarming system | |
| CN117930785B (en) | Workshop production safety monitoring system based on big data | |
| CN116560258A (en) | Network safety control system with automatic power-off function | |
| CN117789333A (en) | Data distribution-based method for detecting abnormal running state of automatic driving vehicle |
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 |