CN117315593A - Recognition method for foreign matter invasion of underground coal mine transportation system - Google Patents
Recognition method for foreign matter invasion of underground coal mine transportation system Download PDFInfo
- Publication number
- CN117315593A CN117315593A CN202311591274.8A CN202311591274A CN117315593A CN 117315593 A CN117315593 A CN 117315593A CN 202311591274 A CN202311591274 A CN 202311591274A CN 117315593 A CN117315593 A CN 117315593A
- Authority
- CN
- China
- Prior art keywords
- foreign matter
- image
- network
- coal mine
- yolov5
- 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.)
- Pending
Links
- 239000003245 coal Substances 0.000 title claims abstract description 66
- 230000009545 invasion Effects 0.000 title claims abstract description 29
- 238000000034 method Methods 0.000 title claims abstract description 28
- 230000007246 mechanism Effects 0.000 claims abstract description 18
- 238000012549 training Methods 0.000 claims description 16
- 238000012360 testing method Methods 0.000 claims description 15
- 230000004927 fusion Effects 0.000 claims description 9
- 230000004913 activation Effects 0.000 claims description 6
- 238000002372 labelling Methods 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 230000002411 adverse Effects 0.000 abstract description 4
- 230000000694 effects Effects 0.000 abstract description 4
- 206010063385 Intellectualisation Diseases 0.000 abstract description 2
- 238000004519 manufacturing process Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 230000006872 improvement Effects 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000013145 classification model Methods 0.000 description 2
- 230000033001 locomotion Effects 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 241000512668 Eunectes Species 0.000 description 1
- 229910000831 Steel Inorganic materials 0.000 description 1
- 238000005299 abrasion Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 235000000396 iron Nutrition 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- 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
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- 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
- G06N3/048—Activation functions
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- 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
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a recognition method for foreign matter invasion of a coal mine underground transportation system, and belongs to the technical field of coal mine intellectualization. Comprising the following steps: acquiring an image to be detected of a coal mine underground transportation system; performing debouncing treatment on the image to be detected to obtain a target image; inputting a target image into a pre-trained CA-YOLOv5 foreign body intrusion recognition model; and determining whether the underground coal mine transportation system is invaded by the foreign matters according to the output result of the CA-YOLOv5 foreign matter invasion identification model, and determining the foreign matter type, the foreign matter position and the foreign matter size if the underground coal mine transportation system is invaded by the foreign matters. After the target image is obtained by debouncing the image to be detected, the target image is identified by the CA-YOLOv5 foreign matter intrusion identification model, so that adverse effects of vibration on the original image can be reduced, identification accuracy is improved, and identification efficiency and accuracy of the identification method are further improved by introducing an attention mechanism network.
Description
Technical Field
The invention relates to the technical field of coal mine intellectualization, in particular to a recognition method for foreign matter invasion of a coal mine underground transportation system.
Background
The underground large-scale transportation systems of the coal mine such as the scraper conveyor, the belt conveyor and the like are key mechanical equipment for coal transportation of a coal face of the coal mine, play an important role in underground coal transportation and are responsible for transporting and transferring coal. Along with the exploitation and utilization of coal resources and the improvement of the production efficiency of coal, the requirements of the coal industry on a transportation system are continuously improved, including the aspects of strength, coal loading weight, normal operation and the like. Ensuring the safe, healthy and efficient operation of the transportation system is important for coal exploitation and underground transportation processes. When the underground coal mine conveying system is invaded by foreign matters, the foreign matters can enter key parts of the conveying system, such as chains, belts, carrier rollers and the like, so that the abrasion of the conveying system is increased or the conveying system is accelerated to be damaged, and the normal operation of equipment is affected; the existence of the foreign matters can cause faults such as blockage, clamping or slipping of a transportation system, so that the transportation of coal is interrupted, and the production efficiency is influenced; the safety accidents such as friction, spark or even fire explosion can be caused, and the life safety of workers and the stability of mine production are endangered; due to faults and downtime caused by foreign matter invasion, production delay and economic loss can be caused, and the operation benefit of mines is affected. Therefore, the method and the device can timely find and remove the foreign matters in the transportation system, strengthen the monitoring and maintenance work of the transportation system and are very important to ensure the safe and efficient operation of underground transportation of the coal mine.
At present, foreign matters invaded into a coal mine underground transportation system are usually identified through a ready-made classification model, namely, collected operation images are directly input into the classification model to identify the foreign matters, and the identification accuracy is low in the mode, and false alarm are easy to occur.
Disclosure of Invention
In order to solve the technical problems, the invention provides a recognition method for foreign matter invasion of a coal mine underground transportation system. The technical scheme of the invention is as follows:
an identification method for foreign matter invasion of a coal mine underground transportation system, which comprises the following steps:
s1, acquiring an operation image of a coal mine underground transportation system as an image to be detected;
s2, performing de-jittering treatment on the image to be detected to obtain a target image;
s3, inputting the target image into a pre-trained CA-Yolov5 foreign body intrusion recognition model, wherein the CA-Yolov5 foreign body intrusion recognition model comprises an input network, a main network, an attention mechanism network, a neck network and an output network which are connected in sequence;
s4, identifying the target image through the input network, the main network, the attention mechanism network, the neck network and the output network, determining whether the underground coal mine transportation system is invaded by the foreign matter according to the output result of the CA-YOLOv5 foreign matter invasion identification model, and determining the foreign matter type, the foreign matter position and the foreign matter size if the underground coal mine transportation system is invaded by the foreign matter.
Optionally, the step S3 further includes, before inputting the target image into a pre-trained CA-YOLOv5 foreign object intrusion recognition model:
s31, acquiring a plurality of original images of foreign matter invading a coal mine underground transportation system by using an industrial high-definition camera arranged on a hydraulic support of a coal face or at the top of a transportation roadway;
s32, performing de-jittering treatment and image enhancement treatment on each original image to obtain a plurality of sample images;
s33, labeling the foreign matter type, the foreign matter position and the foreign matter size in each sample image to obtain a training set and a testing set for training a CA-YOLOv5 foreign matter intrusion recognition model;
s34, training the CA-YOLOv5 foreign matter intrusion recognition model through the training set until the loss value of the CA-YOLOv5 foreign matter intrusion recognition model tends to be converged, testing the CA-YOLOv5 foreign matter intrusion recognition model when the loss value is converged through the testing set, and obtaining the trained CA-YOLOv5 foreign matter intrusion recognition model when the test is qualified.
Optionally, the step S4, when identifying the target image through the input network, the backbone network, the attention mechanism network, the neck network, and the output network, includes:
s41, adjusting the target image into an input image meeting the input requirement of the CA-YOLOv5 foreign object invasion recognition model through an input network;
s42, extracting global characteristic information of an input image through a backbone network;
s43, extracting key feature information from the global feature information through the attention mechanism network;
s44, feature fusion and dimension reduction are carried out on the global feature information and the key feature information through a neck network, so that fusion features are obtained;
s45, calculating the fusion characteristics through an output network, and outputting the confidence coefficient of the target category, the target boundary frame and the coordinates of the target boundary frame.
Optionally, the step S43, when extracting key feature information from the global feature information through the attention mechanism network, includes:
s431, global average pooling is carried out on global feature information in the width direction and the height direction respectively, and a width direction feature image and a height direction feature image are obtained respectively;
s432, merging the width direction feature image and the height direction feature image, and then sequentially calculating through convolution operation, standardization operation and activation function to obtain merged features;
s433, dividing the combined feature into a height feature and a width feature again;
s434, calculating an activation function value after adjusting the channel number of the height feature and the width feature to obtain the attention condition on the wide and high dimensions;
and S435, multiplying the attention condition in the wide and high dimensions by global feature information to obtain key feature information.
Optionally, the step S4 further includes, after determining that the underground coal mine transportation system is invaded by the foreign matter according to the output result of the CA-YOLOv5 foreign matter invasion recognition model:
s5, outputting foreign matter intrusion alarm information.
Optionally, the step S2 is implemented by an image feature matching technology when the image to be detected is subjected to a de-jittering process.
All the above optional technical solutions can be arbitrarily combined, and the detailed description of the structures after one-to-one combination is omitted.
By means of the scheme, the beneficial effects of the invention are as follows:
after the target image is obtained by debouncing the image to be detected, the target image is identified by the CA-YOLOv5 foreign matter intrusion identification model, so that adverse effects on the original image caused by vibration generated by production behaviors such as movement of a hydraulic support and coal mining can be reduced, identification accuracy is improved, key characteristic information in the target image is extracted more easily by introducing an attention mechanism network, identification efficiency and accuracy of an identification method are further improved, and accordingly occurrence of coal mine underground transportation system accidents can be reduced, and the overall safety degree and coal yield of a coal mine are improved.
The foregoing description is only an overview of the present invention, and is intended to provide a better understanding of the present invention, as it is embodied in the following description, with reference to the preferred embodiments of the present invention and the accompanying drawings.
Drawings
Fig. 1 is a flow chart of the present invention.
FIG. 2 is a schematic diagram showing the constitution of a model for identifying invasion of a CA-YOLOv5 foreign matter in the present invention.
Detailed Description
The following describes in further detail the embodiments of the present invention with reference to the drawings and examples. The following examples are illustrative of the invention and are not intended to limit the scope of the invention.
The recognition method for the invasion of the foreign matters in the underground coal mine transportation system can be realized through equipment with a calculation function, such as a server, a computer or a mobile terminal. As shown in FIG. 1, the method for identifying the invasion of the foreign matters in the underground coal mine transportation system provided by the invention comprises the following steps of.
S1, acquiring an operation image of a coal mine underground transportation system as an image to be detected.
The transport system may be a scraper conveyor, a belt conveyor, or the like. The running image is an image acquired in the running process of the transportation system. In the specific implementation, the method can be realized by acquiring operation images through an industrial high-definition camera arranged on a hydraulic support of a coal face or at the top of a transportation roadway and then transmitting the operation images to equipment with a calculation function.
S2, performing de-jittering treatment on the image to be detected to obtain a target image.
Specifically, the step S2 is implemented by an image feature matching technology when the image to be detected is subjected to the de-jittering processing.
Through the shake-free processing, adverse effects on the quality of the image to be detected caused by vibration of actions such as frame moving, coal cutting and the like in the coal mine production process can be reduced, and the accuracy of foreign matter intrusion recognition is improved.
S3, inputting the target image into a pre-trained CA-Yolov5 foreign body intrusion recognition model, wherein the CA-Yolov5 foreign body intrusion recognition model comprises an input network, a main network, an attention mechanism network, a neck network and an output network which are sequentially connected.
The CA-YOLOv5 foreign matter intrusion recognition model provided by the embodiment of the invention improves an original YOLOv5 network, and adds an attention mechanism network after a main network of the original YOLOv5 network, as shown in figure 2, improves an original YOLOv5 algorithm, and obtains the CA-YOLOv5 foreign matter intrusion recognition model. By introducing the attention mechanism network, the information which is more critical to the current task can be focused in a plurality of input information, the attention degree of other information is reduced, and even irrelevant information is filtered out, so that the information overload problem is solved, and the task processing efficiency and accuracy are improved.
The step S3 is to train the CA-Yolov5 foreign matter invasion recognition model before inputting the target image into the pre-trained CA-Yolov5 foreign matter invasion recognition model.
The specific training of the CA-YOLOv5 foreign object invasion recognition model can be achieved by the following steps S31 to S34.
S31, acquiring a plurality of original images of foreign matter invading the underground coal mine transportation system by using an industrial high-definition camera arranged on a hydraulic support of a coal face or at the top of a transportation roadway.
Wherein the size of the original image is not fixed, and in order to ensure that foreign matters are visible in the field of view of the original image, the size of the original image may be set to be not less than 1920×1020. The original image may be stored in a PNG, JPG, JPEG, BMP, TIF, GIF, PCX or FPX format. In order to ensure the recognition accuracy of the trained CA-YOLOv5 foreign object invasion recognition model, the number of original images acquired for a single foreign object is not less than 500.
Further, after the industrial high-definition camera collects the image, irrelevant areas in the image can be removed in a computer cutting mode, so that detail information of foreign matters in the image is improved, background interference is eliminated, and the cut image is used as an original image.
S32, performing debouncing processing and image enhancement processing on each original image to obtain a plurality of sample images.
The de-dithering process may be implemented by using an image feature matching technique, and the image enhancement process may be implemented by using an image enhancement technique, which will not be described in detail in the embodiments of the present invention. The image enhancement processing can be performed by rotating, changing the brightness and color saturation of the image, and adjusting the size.
The adverse effect on the original image caused by vibration generated by production behaviors such as hydraulic support movement and coal mining is reduced through the shake removing treatment, and the original image is processed through the image enhancement treatment, so that the purposes of enriching the original image information and sharpening the image can be achieved.
S33, labeling the foreign matter type, the foreign matter position and the foreign matter size in each sample image to obtain a training set and a testing set for training the CA-YOLOv5 foreign matter intrusion recognition model.
Wherein, the foreign matter category includes helmets, angle irons, steel plates, anchor rods, spades and the like. The foreign object location may be characterized by the coordinates of the center point of the bounding box where the foreign object is located. The foreign object size may be characterized by the width and height of the bounding box in which the foreign object is located.
Specifically, during labeling, foreign matters in the sample image can be labeled piece by piece in labeling software such as labelimg, labelme in an Anaconda virtual environment.
And after the labeling of each sample image is completed, converting the sample image and corresponding labeling information into a txt format file required by a CA-YOLOv5 foreign object intrusion recognition model to obtain a training data set, and dividing the training data set into a training set and a testing set according to 8:2.
S34, training the CA-YOLOv5 foreign matter intrusion recognition model through the training set until the loss value of the CA-YOLOv5 foreign matter intrusion recognition model tends to be converged, testing the CA-YOLOv5 foreign matter intrusion recognition model when the loss value is converged through the testing set, and obtaining the trained CA-YOLOv5 foreign matter intrusion recognition model when the test is qualified.
Specifically, when the CA-YOLOv5 foreign matter invaded identification model is tested by the test set when the loss value is converged, when the identification success rate reaches a preset threshold value, the test is confirmed to be qualified. The preset threshold may be set as desired. However, in order to ensure that the trained CA-YOLOv5 foreign matter intrusion recognition model can accurately recognize foreign matter, the higher the preset threshold value is set, the better, for example, the preset threshold value is set to 90%, 95%, or the like.
The size of the foreign matters is not limited in the sample image, so that the CA-YOLOv5 foreign matter intrusion recognition model trained by the embodiment of the invention can recognize foreign matters with various sizes at the same time.
S4, identifying the target image through the input network, the main network, the attention mechanism network, the neck network and the output network, determining whether the underground coal mine transportation system is invaded by the foreign matter according to the output result of the CA-YOLOv5 foreign matter invasion identification model, and determining the foreign matter type, the foreign matter position and the foreign matter size if the underground coal mine transportation system is invaded by the foreign matter.
On the basis of the above, the step S4 may be implemented by the following steps S41 to S45 when the target image is recognized by the CA-YOLOv5 foreign matter intrusion recognition model.
S41, adjusting the target image into an input image meeting the input requirement of the CA-YOLOv5 foreign matter intrusion recognition model through an input network.
In the specific implementation, the brightness, the size, the contrast, the color and other contents of the target image are adjusted to ensure that the input image meets the requirements of the CA-YOLOv5 foreign matter intrusion recognition model, and a better input image is provided, so that the accuracy and the stability of recognition are improved.
S42, extracting global characteristic information of the input image through the backbone network.
The backbone network adopts a convolutional neural network architecture to extract global characteristic information of an input image.
S43, extracting key feature information from the global feature information through the attention mechanism network.
Specifically, this step, when embodied, can be realized by the following steps S431 to S435.
S431, global average pooling is carried out on the global feature information in the width direction and the height direction respectively, and a width direction feature map and a height direction feature map are obtained respectively.
S432, merging the width direction feature image and the height direction feature image, and then sequentially calculating through convolution operation, standardization operation and activation function to obtain merging features.
S433, the merged feature is separated into a height feature and a width feature again.
S434, the activation function value is calculated after the channel number of the height feature and the width feature is adjusted, and the attention condition in the wide and high dimensions is obtained.
And S435, multiplying the attention condition in the wide and high dimensions by global feature information to obtain key feature information.
The attention condition in the wide and high dimensions is multiplied by the global characteristic information to serve as key characteristic information, so that the boundary box and the foreign object size of the foreign object obtained through recognition in the embodiment of the invention are more accurate.
S44, feature fusion and dimension reduction are carried out on the global feature information and the key feature information through the neck network, and fusion features are obtained.
This step, when embodied, may be implemented by a series of convolution layers and pooling layers.
S45, calculating the fusion characteristics through an output network, and outputting the confidence coefficient of the target category, the target boundary frame and the coordinates of the target boundary frame.
The output network outputs the confidence coefficient of each category, determines the category with the highest confidence coefficient as the target category, circles the position of the foreign object through the target boundary box, and outputs the coordinates of the center point of the target boundary box to mark the position of the foreign object.
Optionally, after determining that the underground coal mine transportation system is invaded by the foreign matter according to the output result of the CA-YOLOv5 foreign matter invasion recognition model, the step S4 may further include: s5, outputting foreign matter intrusion alarm information. The foreign object intrusion alarm information may include a foreign object type, a foreign object position, and the like. The foreign body intrusion alarm information can be played through voice or text. Through the output foreign matter invasion alarm information, the staff of being convenient for in time handles the foreign matter, avoids the foreign matter to cause the damage or influence the normal operating of equipment to colliery underground transportation system.
In summary, the embodiment of the invention provides a recognition method for foreign matter invasion of a coal mine underground transportation system based on a CA-YOLOv5 algorithm, which can intelligently recognize foreign matter accidentally entering the coal mine underground transportation system, and can effectively improve recognition accuracy of foreign matter invasion of the coal mine underground transportation system under the condition of an actual coal face of the coal mine, and particularly can detect foreign matter of various scales simultaneously, so that possibility of false alarm and missing alarm is reduced. The device can reliably identify under different environmental conditions, replace manual work to identify foreign matters, improve efficiency, reduce personnel risk, improve accuracy and realize real-time monitoring and early warning. The limitation of the traditional manual identification method is improved, and the safety and reliability of the underground coal mine transportation system are improved. On the basis of the existing deep learning framework, key feature information in the target image is more easily extracted by adding the attention mechanism network, so that the recognition efficiency and accuracy of the recognition method are further improved, and ideas are provided for subsequent unknown class autonomous learning research and engineering practical application.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, and it should be noted that it is possible for those skilled in the art to make several improvements and modifications without departing from the technical principle of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention.
Claims (6)
1. The method for identifying the invasion of the foreign matters in the underground coal mine transportation system is characterized by comprising the following steps of:
s1, acquiring an operation image of a coal mine underground transportation system as an image to be detected;
s2, performing de-jittering treatment on the image to be detected to obtain a target image;
s3, inputting the target image into a pre-trained CA-Yolov5 foreign body intrusion recognition model, wherein the CA-Yolov5 foreign body intrusion recognition model comprises an input network, a main network, an attention mechanism network, a neck network and an output network which are connected in sequence;
s4, identifying the target image through the input network, the main network, the attention mechanism network, the neck network and the output network, determining whether the underground coal mine transportation system is invaded by the foreign matter according to the output result of the CA-YOLOv5 foreign matter invasion identification model, and determining the foreign matter type, the foreign matter position and the foreign matter size if the underground coal mine transportation system is invaded by the foreign matter.
2. The method for recognizing foreign matter invasion of a coal mine underground transportation system according to claim 1, wherein S3 further comprises, before inputting the target image into a pre-trained CA-YOLOv5 foreign matter invasion recognition model:
s31, acquiring a plurality of original images of foreign matter invading a coal mine underground transportation system by using an industrial high-definition camera arranged on a hydraulic support of a coal face or at the top of a transportation roadway;
s32, performing de-jittering treatment and image enhancement treatment on each original image to obtain a plurality of sample images;
s33, labeling the foreign matter type, the foreign matter position and the foreign matter size in each sample image to obtain a training set and a testing set for training a CA-YOLOv5 foreign matter intrusion recognition model;
s34, training the CA-YOLOv5 foreign matter intrusion recognition model through the training set until the loss value of the CA-YOLOv5 foreign matter intrusion recognition model tends to be converged, testing the CA-YOLOv5 foreign matter intrusion recognition model when the loss value is converged through the testing set, and obtaining the trained CA-YOLOv5 foreign matter intrusion recognition model when the test is qualified.
3. The method for identifying foreign matter intrusion into a coal mine underground transportation system according to claim 1, wherein S4 when identifying the target image through the input network, the backbone network, the attention mechanism network, the neck network, and the output network comprises:
s41, adjusting the target image into an input image meeting the input requirement of the CA-YOLOv5 foreign object invasion recognition model through an input network;
s42, extracting global characteristic information of an input image through a backbone network;
s43, extracting key feature information from the global feature information through the attention mechanism network;
s44, feature fusion and dimension reduction are carried out on the global feature information and the key feature information through a neck network, so that fusion features are obtained;
s45, calculating the fusion characteristics through an output network, and outputting the confidence coefficient of the target category, the target boundary frame and the coordinates of the target boundary frame.
4. A method for identifying foreign object invasion of underground coal mine transportation system according to claim 3, wherein S43 when extracting key feature information from global feature information through an attention mechanism network comprises:
s431, global average pooling is carried out on global feature information in the width direction and the height direction respectively, and a width direction feature image and a height direction feature image are obtained respectively;
s432, merging the width direction feature image and the height direction feature image, and then sequentially calculating through convolution operation, standardization operation and activation function to obtain merged features;
s433, dividing the combined feature into a height feature and a width feature again;
s434, calculating an activation function value after adjusting the channel number of the height feature and the width feature to obtain the attention condition on the wide and high dimensions;
and S435, multiplying the attention condition in the wide and high dimensions by global feature information to obtain key feature information.
5. The method for recognizing the intrusion of the foreign matter into the underground coal mine transportation system according to claim 1, wherein the step S4 further comprises, after determining that the underground coal mine transportation system is intruded by the foreign matter according to the output result of the CA-YOLOv5 foreign matter intrusion recognition model:
s5, outputting foreign matter intrusion alarm information.
6. The method for identifying the intrusion of foreign matters into the underground coal mine transportation system according to claim 1, wherein the step S2 is realized by an image feature matching technology when the image to be detected is subjected to de-jittering.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311591274.8A CN117315593A (en) | 2023-11-27 | 2023-11-27 | Recognition method for foreign matter invasion of underground coal mine transportation system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311591274.8A CN117315593A (en) | 2023-11-27 | 2023-11-27 | Recognition method for foreign matter invasion of underground coal mine transportation system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117315593A true CN117315593A (en) | 2023-12-29 |
Family
ID=89255566
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311591274.8A Pending CN117315593A (en) | 2023-11-27 | 2023-11-27 | Recognition method for foreign matter invasion of underground coal mine transportation system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117315593A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117557967A (en) * | 2024-01-11 | 2024-02-13 | 陕西开来机电设备制造有限公司 | Intelligent management system for safety protection of belt conveyor |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109697424A (en) * | 2018-12-19 | 2019-04-30 | 浙江大学 | A kind of high-speed railway impurity intrusion detection device and method based on FPGA and deep learning |
| CN115760640A (en) * | 2022-12-06 | 2023-03-07 | 太原理工大学 | Coal mine low-illumination image enhancement method based on noise-containing Retinex model |
| CN116665011A (en) * | 2023-05-31 | 2023-08-29 | 太原理工大学 | Coal flow foreign matter identification method for coal mine belt conveyor based on machine vision |
-
2023
- 2023-11-27 CN CN202311591274.8A patent/CN117315593A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109697424A (en) * | 2018-12-19 | 2019-04-30 | 浙江大学 | A kind of high-speed railway impurity intrusion detection device and method based on FPGA and deep learning |
| CN115760640A (en) * | 2022-12-06 | 2023-03-07 | 太原理工大学 | Coal mine low-illumination image enhancement method based on noise-containing Retinex model |
| CN116665011A (en) * | 2023-05-31 | 2023-08-29 | 太原理工大学 | Coal flow foreign matter identification method for coal mine belt conveyor based on machine vision |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117557967A (en) * | 2024-01-11 | 2024-02-13 | 陕西开来机电设备制造有限公司 | Intelligent management system for safety protection of belt conveyor |
| CN117557967B (en) * | 2024-01-11 | 2024-03-29 | 陕西开来机电设备制造有限公司 | Intelligent management system for safety protection of belt conveyor |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110390691B (en) | Ore dimension measuring method based on deep learning and application system | |
| CN104535356B (en) | A kind of roller steel wire rope rope fault monitoring method and system based on machine vision | |
| CN117315593A (en) | Recognition method for foreign matter invasion of underground coal mine transportation system | |
| CN112001878A (en) | Deep learning ore scale measuring method based on binarization neural network and application system | |
| CN107176432A (en) | A kind of anchor pole foreign matter and belt tearing detecting system based on machine vision | |
| CN110569843B (en) | Intelligent detection and identification method for mine target | |
| CN111260629A (en) | Pantograph structure abnormity detection algorithm based on image processing | |
| CN111161292A (en) | Ore size measurement method and application system | |
| CN111626169B (en) | Image-based railway dangerous falling rock size judgment method | |
| CN115147770B (en) | Belt foreign matter vision recognition system based on image processing | |
| CN113139572B (en) | Image-based train air spring fault detection method | |
| CN110015553A (en) | Travelling belt system foreign bodies detection guard method based on video analysis | |
| CN114689600A (en) | Method and system for detecting surface crack of bridge concrete structure | |
| CN111369538A (en) | Conveyor belt deviation detection method based on deep convolutional neural network | |
| CN116177146A (en) | Conveying belt detection system and method based on AI and laser monitoring technology | |
| CN111582135A (en) | Excavator hand operation proficiency evaluation method and device based on working stage recognition | |
| CN115082850A (en) | Template support safety risk identification method based on computer vision | |
| CN113128555B (en) | Method for detecting abnormality of train brake pad part | |
| CN114445398A (en) | Method and device for monitoring state of side protection plate of hydraulic support of coal mining machine | |
| CN112232235B (en) | Intelligent factory remote monitoring method and system based on 5G sum algorithm | |
| CN105678741A (en) | Mining conveyer belt longitudinal rip image detection method | |
| CN116994066A (en) | Tail rope detection system based on improved target detection model | |
| CN115504187B (en) | Intelligent speed regulation and protection system and control method for mining belt | |
| CN114359779B (en) | Belt tearing detection method based on deep learning | |
| CN114120175A (en) | Method for identifying foreign matters on coal conveying belt based on computer vision |
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 |