CN113283344B - Mining conveyor belt deviation detection method based on semantic segmentation network - Google Patents
Mining conveyor belt deviation detection method based on semantic segmentation network Download PDFInfo
- Publication number
- CN113283344B CN113283344B CN202110584550.2A CN202110584550A CN113283344B CN 113283344 B CN113283344 B CN 113283344B CN 202110584550 A CN202110584550 A CN 202110584550A CN 113283344 B CN113283344 B CN 113283344B
- Authority
- CN
- China
- Prior art keywords
- conveyor belt
- area
- mining conveyor
- mining
- training
- 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
- 238000005065 mining Methods 0.000 title claims abstract description 149
- 238000001514 detection method Methods 0.000 title claims abstract description 58
- 230000011218 segmentation Effects 0.000 title claims abstract description 51
- 238000012549 training Methods 0.000 claims abstract description 72
- 238000012544 monitoring process Methods 0.000 claims abstract description 57
- 238000012795 verification Methods 0.000 claims abstract description 22
- 230000004927 fusion Effects 0.000 claims abstract description 21
- 238000000034 method Methods 0.000 claims abstract description 17
- 239000003245 coal Substances 0.000 claims abstract description 12
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 230000006870 function Effects 0.000 claims description 30
- 238000012360 testing method Methods 0.000 claims description 24
- 238000011176 pooling Methods 0.000 claims description 21
- 238000007726 management method Methods 0.000 claims description 17
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000013135 deep learning Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 8
- 230000005540 biological transmission Effects 0.000 claims description 7
- 238000002372 labelling Methods 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000013523 data management Methods 0.000 claims description 6
- 238000009826 distribution Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 5
- 238000003860 storage Methods 0.000 claims description 5
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 238000004891 communication Methods 0.000 claims description 3
- 238000005520 cutting process Methods 0.000 claims description 3
- 238000009434 installation Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 239000013307 optical fiber Substances 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 239000000725 suspension Substances 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 claims description 3
- 230000008569 process Effects 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 description 10
- 230000006872 improvement Effects 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000005674 electromagnetic induction Effects 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 239000004566 building material Substances 0.000 description 1
- 238000003889 chemical engineering Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000003384 imaging method Methods 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
- 238000005457 optimization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000003466 welding Methods 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
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
- G06V20/42—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items of sport video content
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
- B65G—TRANSPORT OR STORAGE DEVICES, e.g. CONVEYORS FOR LOADING OR TIPPING, SHOP CONVEYOR SYSTEMS OR PNEUMATIC TUBE CONVEYORS
- B65G43/00—Control devices, e.g. for safety, warning or fault-correcting
- B65G43/02—Control devices, e.g. for safety, warning or fault-correcting detecting dangerous physical condition of load carriers, e.g. for interrupting the drive in the event of overheating
-
- 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
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/20016—Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20132—Image cropping
-
- 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
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
A mining conveyor belt deviation detection method based on a semantic segmentation network belongs to the technical field of coal mine monitoring. The method comprises the following steps: real-time recording a mining conveyor belt operation video, and acquiring a mining conveyor belt sample image according to a screenshot of the video; preprocessing an image to obtain a mining conveyor belt deviation detection data set; constructing a semantic segmentation network for detecting the deviation of the mining conveyor belt; training the sample image training set and the verification set to generate a training model; inputting an original mining conveyor belt image to be segmented based on a trained semantic segmentation network model, and segmenting out a planned safety area and a conveyor belt actual area subgraph; evaluating the position relation, judging the safety area of the conveyor belt, and judging that the mining conveyor belt has deviation faults at the moment if the safety area exceeds the safety area and the fusion area is larger than the area of the safety area; after detecting the deviation fault, the system automatically sends out an alarm signal on site. The advantages are that: the real area of the conveyer belt in the running process is rapidly and accurately judged, and the instantaneity and the accuracy are improved.
Description
Technical Field
The invention relates to the technical field of coal mine monitoring, in particular to a mining conveyor belt deviation detection method based on a semantic segmentation network.
Background
The belt conveyor is used as a material conveying device with simple structure, convenient maintenance and extremely strong conveying force, and is widely applied to industries such as building materials, chemical engineering, ports, grains, electric power and the like. The mining conveyor belt is an indispensable part in a coal mine belt conveyor system and plays a vital role in coal production. Along with the continuous increase of development of coal energy, the continuous improvement of production efficiency, the continuous improvement of requirements on various aspects such as load, speed and transmission distance of a mining conveyor belt, whether the mining conveyor belt can work efficiently, safely and reliably, and the continuous improvement of coal production efficiency, improvement of scientific research technology and optimization of economic indexes are related. Conveyor belt deflection is quite common in coal production operation and is one of the faults with great influence on a belt conveyor system. This phenomenon can exacerbate belt wear and even damage the entire belt conveyor system. In addition, if the conveyer belt is deviated in the load operation process, accidents such as material sprinkling or material flying and the like can be caused. This not only reduces the actual transport efficiency, but also causes unnecessary losses, and even causes large-scale accidents such as fire disasters and the like in serious cases. Therefore, the method has important significance in the deviation detection and fault early warning of the mining conveyor belt.
The method is characterized in that the deviation detection of the mining conveyer belt is finished through manual monitoring, the main implementation mode is multi-person on-duty monitoring, the faults of the conveyer belt which occur in the inspection process are repaired or reported, but the long-time manual monitoring can cause serious fatigue of inspection personnel, and the problems of missing report, false report and the like are easy to occur. With development of micro-processing technology and application of sensors, mining conveyor belt deviation detection devices are gradually developed and mainly divided into contact type and non-contact type; in addition, as machine vision technology advances, machine vision inspection methods have emerged.
The contact type device mainly comprises a bar-shaped detector, a linear detector, a leakage detector, a bandwidth detector, a vibration monitor and the like, and most of the devices adopt a mechanical structure, and the contact type device is convenient to install and simple in structure, but low in reliability, low in automation level and poor in self-adaptation capability.
The non-contact detection method comprises an X-ray detection method, an ultrasonic detection method, an electromagnetic induction detection method and the like, wherein the X-ray detection method cannot be used for rapid imaging, the required running belt speed cannot be too high, the anti-interference capability is poor, and the method is harmful to human bodies. The ultrasonic detection method detects the internal defects of the butt welding seam of the roller of the belt conveyor, but more conditions are required to be met, the detection steps are complicated, and the equipment cost is high. The electromagnetic induction type detection method is characterized in that a sensing coil is embedded in a conveyer belt at intervals, a group of signal transmitting and receiving detectors on two sides of the conveyer belt are used for detecting signals to judge whether faults occur or not, but the anti-interference capability is poor under severe and complex environments, and the sensor is easy to fail and damage.
The machine vision detection method mainly comprises the steps of shooting and acquiring a field image through an industrial CCD camera, performing image binarization through a computer, and performing difference contrast on the acquired image and a preset background image or directly extracting edge characteristics of a conveying belt to judge whether deviation faults occur. The method has high requirements on image segmentation processing and surrounding environment of the conveyor belt, and the changes of scattered coal, mud blocks and other impurities adhered to the surface of the conveyor belt or external illumination can directly influence the image segmentation effect, so that detection algorithm parameters need to be adjusted according to different scenes.
In view of the above, there is a need for an efficient and reliable mining conveyor belt deviation detection method.
Disclosure of Invention
The invention aims to: aiming at the problems, the invention provides a mining conveyor belt deviation detection method based on a semantic segmentation network, which aims to improve the real-time performance and accuracy of the mining conveyor belt deviation detection and realize the state monitoring and safety protection of the mining conveyor belt by rapidly and accurately judging the actual area of the conveyor belt in the running process.
The technical scheme is as follows: in order to achieve the purpose of the invention, the technical scheme adopted by the invention is as follows: a mining conveyor belt deviation detection method based on a semantic segmentation network comprises the following steps:
step 1: collecting a mining conveyor belt operation video recorded in real time by a monitoring camera, and obtaining a mining conveyor belt sample image according to a screenshot of the video;
step 2: screening the sample images, and selecting a plurality of mining conveyer belt images in different running states; preprocessing an image, and marking the image of the mining conveyor belt by using a deep learning marking tool to obtain a mining conveyor belt deviation detection data set;
step 3: generating a training set, a verification set and a test set of sample images based on the marked mining conveyor belt deviation detection data set;
step 4: constructing a semantic segmentation network for detecting the deviation of the mining conveyor belt;
step 5: inputting the sample image training set and the verification set into a semantic segmentation network for image data training until the network model reaches a convergence state, and generating a training model;
step 6: testing the segmentation result of the semantic segmentation network, inputting the conveyor belt test set image into the semantic segmentation network to obtain a semantic segmentation image, and calculating the average intersection ratio and the average pixel accuracy according to the marked mining conveyor belt image;
step 7: analyzing the test result of the semantic segmentation network, if the test result does not meet the expected requirement, adjusting the number of samples, the training times, the cross entropy loss function and the learning rate of the network model training, and retraining and testing the network model again until the network model meets the expected requirement;
step 8: inputting an original mining conveyor belt image to be segmented based on a trained semantic segmentation network model, and segmenting out a planned safety area and a conveyor belt actual area subgraph;
step 9: evaluating the position relation between each planned safety area and the conveyer belt area subgraph, judging whether the conveyer belt area exceeds the safety area, calculating the fusion area of the two areas, and judging that the mining conveyer belt has deviation faults at the moment if the conveyer belt exceeds the safety area and the fusion area is larger than the area of the safety area;
step 10: after the deviation fault is detected, the mining conveyer belt state monitoring software platform records the fault information at the moment, and the system automatically sends out an alarm signal on site.
Further, in the step 1, the selected monitoring camera can meet different requirements of monitoring the mining conveyor belt in the working environment of the coal mine, and the installation position of the monitoring camera is required to ensure that the monitoring video can obtain a complete conveyor belt operation area, and the specific steps are as follows:
step 1.1: the working temperature of the selected monitoring camera is-30-60 ℃, the humidity is less than 95%, and the functions of backlight compensation, digital noise reduction and day-night conversion are supported;
step 1.2: the selected monitoring camera supports HTTP, TCP/IP and DNS network protocols, and the transmission rate is 10/100Mbps;
step 1.3: the selected monitoring camera is arranged at a suspension end at a position 1.8m right above the head of the belt conveyor, and the horizontal view angle of a camera lens is 89 degrees;
step 1.4: the bandwidth of the mining conveyor belt is 1.4m, the average running speed is 4m/s, the frame rate of the selected monitoring camera is 25fps, and the shutter speed is 1/3s to 1/100,000s;
step 1.5: and acquiring a mining conveyor belt operation video in real time through a monitoring camera, and intercepting the video into a conveyor belt operation image according to frames, wherein the intercepted image resolution is X.
Further, in the step 2, the sample images are screened, a plurality of mining conveyer belt images with different running states are selected, the planned safe areas and conveyer belt areas in the images are marked by using a deep learning marking tool labelme, and a conveyer belt data set is generated in batches, and the specific steps are as follows:
step 2.1: the screened images are taken from mining conveyor belt sample images in different running states in a plurality of time periods;
step 2.2: preprocessing the images, and cutting the images in batches according to the position of the target area to be detected, wherein the resolution ratio of the obtained images is multiplied by 533;
step 2.3: importing the cut images into a labelme, respectively setting a reading position and a storage position of the images to be processed, respectively marking a conveyor belt area and a planned safety area in each image in sequence according to sequence numbers, and setting names of two labels as a belt and a common respectively;
step 2.4: after the labeling is completed, storing the newly generated json into a target folder, and converting all the json into png images in batches, wherein the bit depth of the images is 24;
step 2.5: and performing bit depth conversion on the generated png images, wherein the bit depth of the images after batch processing is 8, and storing all the images in a target folder.
Further, in the step 3, a training set, a verification set and a test set of sample images are generated based on the marked mining conveyor belt deviation detection data set; obtaining 560 samples of the mining conveyor belt deviation detection data set through labelme labeling, wherein the number of samples of the training set and the number of samples of the verification set are set to be 9:1, the number of samples of the training set is 450, the number of samples of the verification set is 50, and the number of samples of the test set is 60.
Further, in the step 4, a semantic segmentation network for mining conveyor belt deviation detection is constructed; the semantic segmentation network is a multi-scale network PSPNet based on a pyramid pooling model, and the PSPNet is divided into a Resnet convolution module, a pyramid pooling module and an FCN module; the depth of the Resnet convolution module is 50 layers, and a characteristic layer of an input image is extracted; the pyramid pooling module is divided into 4 layers of collection areas; the collection area comprises 4 pooling cores with different sizes respectively; the sizes of the pooling cores are 1 multiplied by 1, 2 multiplied by 2, 3 multiplied by 3 and 6 multiplied by 6 respectively, and pooling feature maps with different sizes are extracted in a mapping way; the output end of the pooling feature map is connected with the input end of a convolution kernel, the size of the convolution kernel is 1 multiplied by 1, and the weight of the global feature is kept; the FCN module outputs a predicted image having the same size as the input image.
Further, in the step 5, selecting hardware equipment and software configuration meeting training requirements, adjusting PSPNet network model parameters, inputting a sample image training set and a verification set into a semantic segmentation network for image data training until the network model reaches a convergence state, and generating a mining conveyor belt semantic segmentation training model; the method comprises the following specific steps:
step 5.1: the selected computer hardware is configured as an i9-K CPU, a memory 32G, an RTX GPU, a video memory 16G, an operating system win10, a programming language Python3.8 and a deep learning frame PyTorch;
step 5.2: respectively adding a manufactured mining conveyor belt deviation detection data set and an original mining conveyor belt sample image into a segmenttaionclass and an Img folder, and executing voc2pspnet.py to realize data reading;
step 5.3: setting the number of target CLASSES to be segmented in pspnet.py and train.py, num_class=3;
step 5.4: setting part of training parameters in train. Py, training round number of epoch hs is 200, training Batch sample size of previous 100 training round number of epoch hs is set to 8, the learning rate is set to 0.01, the training Batch sample size of the last 100 training rounds of epoch hs is set to 4, and the learning rate is set to 0.001;
step 5.5: and training the PSPNet model, wherein the loss function values of the training set and the verification set gradually decrease and gradually become gentle along with the continuous increase of the number of the training rounds of the epoch, and the training model basically converges.
Further, the average cross ratio MIoU in the step 6 has a calculation formula:
wherein k is a category and does not contain a background category; p is p ij For the number of false positive cases, p ii For the number of true examples, p ji Is the number of false negatives.
Further, the average pixel accuracy MPA in the step 6 has a calculation formula as follows:
wherein k is a category and does not contain a background category; p is p ij For the number of false positive cases, p ii Is the true example number.
Further, the cross entropy loss function Cross Entropy loss in the step 7 has a calculation formula as follows:
wherein p (x) i ) For the true probability distribution, q (x i ) To predict probability distribution.
Further, in the step 9, the position relationship between each planned safety area and the conveyer belt area subgraph and the area size of the two areas are evaluated, and the specific steps are as follows:
step 9.1: judging the position relation between the belt area and the common area divided in the drawing for a single mining conveyor belt sub-graph, if the belt area is contained in the common area, considering that the mining conveyor belt runs safely at the moment, and if the belt area is not completely contained in the common area, considering that the mining conveyor belt is likely to have deviation faults at the moment;
step 9.2: calculating the area of the common region and the fused area of the two regions respectively;
step 9.3: judging the size relation between the common area and the fusion area, if the ratio of the common area to the fusion area is larger than 0.90, considering that the mining conveyor belt runs safely at the moment, if the ratio of the common area to the fusion area is 0.75-0.90, considering that the mining conveyor belt has slight deviation faults at the moment, and if the ratio of the common area to the fusion area is smaller than 0.75, considering that the mining conveyor belt has serious deviation faults at the moment;
step 9.4: and respectively evaluating other subgraphs according to the judgment basis.
Further, in the step 10, when a deviation fault is detected at a certain moment, the mining conveyor belt state monitoring software platform records the fault information at the moment, and the system immediately and automatically sends an alarm signal to the site; the mining conveyor belt state monitoring software platform realizes the deviation detection of the mining conveyor belt through a monitoring camera at the front end, a transmission communication of an optical fiber network in the middle and a visual detection algorithm at the background; when the deviation fault is detected at a certain moment, the mining conveyer belt state monitoring software platform records the fault information at the moment, and the system immediately and automatically sends an alarm signal to the site.
The mining conveyor belt state monitoring software platform comprises: the system comprises a system login module, a system management module, a video monitoring module, a data management module and an alarm management module;
a system login module: authenticating the identity of the access user; and judging whether the user is allowed to enter the system according to the user name and the password by entering the system login interface.
And a system management module: parameter configuration and password management of the mining conveyer belt status monitoring software platform are realized, the parameter setting realizes the system stopping or restarting and system alarming functions, and the password modifying function can realize the self-changing of login passwords by users.
And the video monitoring module is used for: the video or image acquisition system comprises a field monitoring function, a camera adding function, a monitoring debugging function and a monitoring equipment inquiring function, acquires acquired video or images, and realizes storage of target video images and inquiry of equipment name, equipment address and port number information.
And a data management module: and automatically storing the data information detected by the system according to the fault type, and allowing an access user to perform real-time query, history browsing and data export operation.
Alarm management module: alarm management is realized based on the detected running state of the conveyer belt; if the running state of the conveyer belt is abnormal and the system detects a fault event, the system automatically alarms, reminds on-site personnel to arrive at the on-site for processing in time, and records alarm information so as to facilitate subsequent shutdown and overhaul.
The beneficial effects are that: compared with the prior art, the technical scheme of the invention has the following beneficial technical effects:
(1) The invention aims to rapidly and accurately judge the actual area of the conveyer belt in the running process, and improve the real-time performance and accuracy of the deviation detection of the mine conveyer belt so as to realize the state monitoring and the safety protection of the mine conveyer belt.
(2) According to the mining conveyor belt deviation detection method, based on the trained mining conveyor belt semantic segmentation network model, the mining conveyor belt deviation fault is detected by calculating and judging the position relation of the planned safety area and the actual area of the conveyor belt and the size of the fusion area of the two areas.
(3) The method improves the deviation detection precision of the mining conveyor belt, has strong real-time performance, strong reliability and high accuracy, can be effectively applied to a vision-based mining conveyor belt state monitoring system, realizes the deviation fault detection of the mining conveyor belt in the operation process, is beneficial to preventing major coal safety accidents, and has great application value.
Drawings
Fig. 1 is an overall flowchart of a mining conveyor belt deviation detection method based on a semantic segmentation network.
Fig. 2 is an image of a mining conveyor belt deviation detection dataset of the present invention after labeling.
Fig. 3 is a schematic diagram of the PSPNet model structure of the present invention.
Fig. 4 is a graph of the result of the mining conveyor belt deviation detection of the present invention.
Fig. 5 is a schematic functional diagram of a mining conveyor belt status monitoring software platform of the present invention.
Detailed Description
The technical scheme of the invention is further described below with reference to the accompanying drawings and examples.
Example 1: according to the mining conveyor belt deviation detection method based on the semantic segmentation network, the whole flow is shown in fig. 1, and the mining conveyor belt deviation fault detection is realized by calculating and judging the position relation between a planned safety area and an actual area of the conveyor belt and the size of a fusion area of the two areas based on a trained mining conveyor belt semantic segmentation network model.
The invention discloses a mining conveyor belt deviation detection method based on a semantic segmentation network, which comprises the following steps of:
step 1: collecting a mining conveyor belt operation video recorded in real time by a monitoring camera, and obtaining a mining conveyor belt sample image according to a screenshot of the video;
step 2: screening sample images, selecting a plurality of mining conveyor belt images in different running states, preprocessing the images, and marking the mining conveyor belt images by using a deep learning marking tool to obtain a mining conveyor belt deviation detection data set;
step 3: generating a training set, a verification set and a test set of sample images based on the marked mining conveyor belt deviation detection data set;
step 4: constructing a semantic segmentation network for detecting the deviation of the mining conveyor belt;
step 5: inputting the sample image training set and the verification set into a semantic segmentation network for image data training until the network model reaches a convergence state, and generating a training model;
step 6: testing the segmentation result of the semantic segmentation network, inputting the conveyor belt test set image into the semantic segmentation network to obtain a semantic segmentation image, and calculating the average intersection ratio and the average pixel accuracy according to the marked mining conveyor belt image;
step 7: analyzing the test result of the semantic segmentation network, if the test result does not meet the expected requirement, adjusting the number of samples, the training times, the cross entropy loss function and the learning rate of the network model training, and retraining and testing the network model again until the network model meets the expected requirement;
step 8: inputting an original mining conveyor belt image to be segmented based on a trained semantic segmentation network model, and segmenting out a planned safety area and a conveyor belt actual area subgraph;
step 9: evaluating the position relation between each planned safety area and the conveyer belt area subgraph, judging whether the conveyer belt area exceeds the safety area, calculating the fusion area of the two areas, and judging that the mining conveyer belt has deviation faults at the moment if the conveyer belt exceeds the safety area and the fusion area is larger than the area of the safety area;
step 10: after the deviation fault is detected, the mining conveyer belt state monitoring software platform records the fault information at the moment, and the system automatically sends out an alarm signal on site.
In step 1, the selected monitoring camera can meet different requirements on mining conveyor belt state monitoring in a coal mine working environment, and the installation position of the monitoring camera is required to ensure that the monitoring video can acquire a complete conveyor belt operation area, and the specific steps are as follows:
step 1.1: the working temperature of the selected monitoring camera is-30-60 ℃, the humidity is less than 95%, and the functions of backlight compensation, digital noise reduction and day-night conversion are supported;
step 1.2: the selected monitoring camera supports HTTP, TCP/IP and DNS network protocols, and the transmission rate is 10/100Mbps;
step 1.3: the selected monitoring camera is arranged at a suspension end at a position 1.8m right above the head of the belt conveyor, and the horizontal view angle of a camera lens is 89 degrees;
step 1.4: the bandwidth of the mining conveyor belt is 1.4m, the average running speed is 4m/s, the frame rate of the selected monitoring camera is 25fps, and the shutter speed is 1/3s to 1/100,000s;
step 1.5: and acquiring a mining conveyor belt operation video in real time through a monitoring camera, and intercepting the video into a conveyor belt operation image according to frames, wherein the intercepted image resolution is X.
In step 2, screening sample images, selecting a plurality of mining conveyer belt images in different running states, marking planned safety areas and conveyer belt areas in the images by using a deep learning marking tool labelme, and generating conveyer belt data sets in batches, wherein the specific steps are as follows:
step 2.1: the screened images are taken from mining conveyor belt sample images in different running states in a plurality of time periods;
step 2.2: preprocessing the images, and cutting the images in batches according to the position of the target area to be detected, wherein the resolution ratio of the obtained images is multiplied by 533;
step 2.3: importing the cut images into a labelme, respectively setting a reading position and a storage position of the images to be processed, respectively marking a conveyor belt area and a planned safety area in each image in sequence according to sequence numbers, and setting names of two labels as a belt and a common respectively;
step 2.4: after the labeling is completed, storing the newly generated json into a target folder, and converting all the json into png images in batches, wherein the bit depth of the images is 24;
step 2.5: and performing bit depth conversion on the generated png images, wherein the bit depth of the images after batch processing is 8, and storing all the images in a target folder.
In step 3, generating a training set, a verification set and a test set of sample images based on the marked mining conveyor belt deviation detection data set; obtaining 560 samples of the mining conveyor belt deviation detection data set through labelme labeling, wherein the number of samples of the training set and the number of samples of the verification set are set to be 9:1, the number of samples of the training set is 450, the number of samples of the verification set is 50, and the number of samples of the test set is 60. Fig. 2 is a diagram of a comparison of an original mining conveyor belt sample image of the present invention with an annotated mining conveyor belt image. Wherein, fig. 2 (a) is a sample diagram of an original mining conveyor belt when a deviation fault occurs; FIG. 2 (b) is a sample view of the mining conveyor belt marked when a deviation fault occurs; FIG. 2 (c) is a sample view of the original mining conveyor belt during normal operation; fig. 2 (d) is a sample diagram of the mining conveyor belt marked during normal operation.
In step 4, constructing a semantic segmentation network for detecting the deviation of the mining conveyor belt; the semantic segmentation network is a multi-scale network PSPNet based on pyramid pooling model, and FIG. 3 is a schematic diagram of a PSPNet model structure of the invention. The PSPNet is divided into a Resnet convolution module, a pyramid pooling module and an FCN module; the depth of the Resnet convolution module is 50 layers, and a characteristic layer of an input image is extracted; the pyramid pooling module is divided into 4 layers of collection areas; the collection area comprises 4 pooling cores with different sizes respectively; the sizes of the pooling cores are 1 multiplied by 1, 2 multiplied by 2, 3 multiplied by 3 and 6 multiplied by 6 respectively, and pooling feature maps with different sizes are extracted in a mapping way; the output end of the pooling feature map is connected with the input end of a convolution kernel, the size of the convolution kernel is 1 multiplied by 1, and the weight of the global feature is kept; the FCN module outputs a predicted image having the same size as the input image.
In step 5, selecting hardware equipment and software configuration meeting training requirements, adjusting PSPNet network model parameters, inputting a sample image training set and a verification set into a semantic segmentation network for image data training until the network model reaches a convergence state, and generating a mining conveyor belt semantic segmentation training model, wherein the specific steps are as follows:
step 5.1: the selected computer hardware is configured as an i9-K CPU, a memory 32G, an RTX GPU, a video memory 16G, an operating system win10, a programming language Python3.8 and a deep learning frame PyTorch;
step 5.2: respectively adding a manufactured mining conveyor belt deviation detection data set and an original mining conveyor belt sample image into a segmenttaionclass and an Img folder, and executing voc2pspnet.py to realize data reading;
step 5.3: setting the number of target CLASSES to be segmented in pspnet.py and train.py, num_class=3;
step 5.4: setting partial training parameters in train. Py, wherein the training round number of the epoch hs is 200, the training Batch sample size (batch_size) of the first 100 training round numbers of the epoch hs is set to 8, the learning rate is set to 0.01, the training Batch sample size (batch_size) of the last 100 training round numbers of the epoch hs is set to 4, and the learning rate is set to 0.001;
step 5.5: and training the PSPNet model, wherein the loss function values of the training set and the verification set gradually decrease and gradually become gentle along with the continuous increase of the number of the training rounds of the epoch, and the training model basically converges.
In step 6, the average cross over ratio (MIoU) calculation formula is:
where k is a category (excluding background category), p ij For the number of false positive cases, p ii For the number of true examples, p ji Is the number of false negatives.
In step 6, the average pixel accuracy (MPA) calculation formula is:
where k is a category (excluding background category), p ij For the number of false positive cases, p ii Is the true example number.
In step 7, the cross entropy loss function (Cross Entropy loss) has the following calculation formula:
wherein p (x) i ) For the true probability distribution, q (x i ) To predict probability distribution.
In step 9, the position relationship between each planned safety area and the conveyer belt area subgraph and the area size of the two areas are evaluated, and the specific steps are as follows:
step 9.1: judging the position relation between the belt area and the common area divided in the drawing for a single mining conveyor belt sub-graph, if the belt area is contained in the common area, considering that the mining conveyor belt runs safely at the moment, and if the belt area is not completely contained in the common area, considering that the mining conveyor belt is likely to have deviation faults at the moment;
step 9.2: calculating the area of the common region and the fused area of the two regions respectively;
step 9.3: judging the size relation between the common area and the fusion area, if the ratio of the common area to the fusion area is larger than 0.90, considering that the mining conveyor belt runs safely at the moment, if the ratio of the common area to the fusion area is 0.75-0.90, considering that the mining conveyor belt has slight deviation faults at the moment, and if the ratio of the common area to the fusion area is smaller than 0.75, considering that the mining conveyor belt has serious deviation faults at the moment;
step 9.4: and respectively evaluating other subgraphs according to the judgment basis. Fig. 4 is a diagram of a result of detecting a deviation of a mining conveyor belt according to the present invention, wherein fig. 4 (a) is a diagram of a result of detecting a deviation fault of the mining conveyor belt; fig. 4 (b) is a diagram of the detection result when the mining conveyor belt is operating normally.
In step 10, when a deviation fault is detected at a certain moment, the mining conveyor belt state monitoring software platform records the fault information at the moment, and the system immediately and automatically sends an alarm signal to the site; the mining conveyor belt state monitoring software platform realizes the deviation detection of the mining conveyor belt through a monitoring camera at the front end, a middle optical fiber network transmission communication and a visual detection algorithm at the background. When the deviation fault is detected at a certain moment, the mining conveyer belt state monitoring software platform records the fault information at the moment, and the system immediately and automatically sends an alarm signal to the site. Fig. 5 is a schematic functional diagram of a mining conveyor belt status monitoring software platform of the present invention.
The mining conveyor belt state monitoring software platform comprises: the system comprises a system login module, a system management module, a video monitoring module, a data management module and an alarm management module.
A system login module: the module is mainly responsible for identity authentication of the access user. And judging whether the user is allowed to enter the system according to the user name and the password by entering the system login interface.
And a system management module: the module is mainly responsible for parameter configuration and password management of the monitoring system, the parameter configuration can realize the functions of system stop or restart and system alarm, and the password modification function can realize the self-change of login passwords by users.
And the video monitoring module is used for: the module mainly comprises a field monitoring function, a camera adding function, a monitoring debugging function and a monitoring equipment inquiring function, and mainly aims to acquire acquired videos or images, store target video images and inquire equipment names, equipment addresses and port number information.
And a data management module: the module is mainly responsible for automatically storing the data information detected by the system according to the fault type, and allowing an access user to perform real-time query, history browsing and data export operations.
Alarm management module: the module is realized based on the detected running state of the conveyer belt, if the running state of the conveyer belt is abnormal and the system detects a fault event, the system automatically alarms, reminds field personnel to arrive at the field for processing in time, and records alarm information so as to facilitate subsequent shutdown and maintenance.
Claims (2)
1. A mining conveyor belt deviation detection method based on a semantic segmentation network is characterized by comprising the following steps: comprises the following steps:
step 1: collecting a mining conveyor belt operation video recorded in real time by a monitoring camera, and obtaining a mining conveyor belt sample image according to a screenshot of the video;
step 2: screening the sample images, and selecting a plurality of mining conveyer belt images in different running states; preprocessing an image, and marking the image of the mining conveyor belt by using a deep learning marking tool to obtain a mining conveyor belt deviation detection data set;
step 3: generating a training set, a verification set and a test set of sample images based on the marked mining conveyor belt deviation detection data set;
step 4: constructing a semantic segmentation network for detecting the deviation of the mining conveyor belt;
step 5: inputting the sample image training set and the verification set into a semantic segmentation network for image data training until the network model reaches a convergence state, and generating a training model;
step 6: testing the segmentation result of the semantic segmentation network, inputting the conveyor belt test set image into the semantic segmentation network to obtain a semantic segmentation image, and calculating the average intersection ratio and the average pixel accuracy according to the marked mining conveyor belt image;
step 7: analyzing the test result of the semantic segmentation network, if the test result does not meet the expected requirement, adjusting the number of samples, the training times, the cross entropy loss function and the learning rate of the network model training, and retraining and testing the network model again until the network model meets the expected requirement;
step 8: inputting an original mining conveyor belt image to be segmented based on a trained semantic segmentation network model, and segmenting out a planned safety area and a conveyor belt actual area subgraph;
step 9: evaluating the position relation between each planned safety area and the conveyer belt area subgraph, judging whether the conveyer belt area exceeds the safety area, calculating the fusion area of the two areas, and judging that the mining conveyer belt has deviation faults at the moment if the conveyer belt exceeds the safety area and the fusion area is larger than the area of the safety area;
step 10: after detecting the deviation fault, the mining conveyer belt state monitoring software platform records the fault information at the moment, and the system automatically sends out an alarm signal on site;
in the step 1, the selected monitoring camera can meet different requirements of monitoring the mining conveyor belt in the working environment of the coal mine, the installation position of the monitoring camera is required to ensure that the monitoring video can acquire the complete conveyor belt operation area, and the specific steps are as follows: step 1.1: the working temperature of the selected monitoring camera is-30-60 ℃, the humidity is less than 95%, and the functions of backlight compensation, digital noise reduction and day-night conversion are supported;
step 1.2: the selected monitoring camera supports HTTP, TCP/IP and DNS network protocols, and the transmission rate is 10/100Mbps; step 1.3: the selected monitoring camera is arranged at a suspension end at a position 1.8m right above the head of the belt conveyor, and the horizontal view angle of a camera lens is 89 degrees;
step 1.4: the bandwidth of the mining conveyor belt is 1.4m, the average running speed is 4m/s, the frame rate of the selected monitoring camera is 25fps, and the shutter speed is 1/3s to 1/100,000s;
step 1.5: the method comprises the steps of collecting a mining conveyor belt operation video in real time through a monitoring camera, and intercepting the video into a conveyor belt operation image according to frames, wherein the intercepted image resolution is 1920 multiplied by 1080;
in the step 2, screening sample images, selecting a plurality of mining conveyer belt images with different running states, marking planned safety areas and conveyer belt areas in the images by using a deep learning marking tool labelme, and generating conveyer belt data sets in batches, wherein the specific steps are as follows:
step 2.1: the screened images are taken from mining conveyor belt sample images in different running states in a plurality of time periods;
step 2.2: preprocessing images, and cutting the images in batches according to the position of a target area to be detected, wherein the resolution ratio of the obtained images is 1084 multiplied by 533;
step 2.3: importing the cut images into a labelme, respectively setting a reading position and a storage position of the images to be processed, respectively marking a conveyor belt area and a planned safety area in each image in sequence according to sequence numbers, and setting names of two labels as a belt and a common respectively;
step 2.4: after the labeling is completed, storing the newly generated json into a target folder, and converting all the json into png images in batches, wherein the bit depth of the images is 24;
step 2.5: performing bit depth conversion on the generated png image, wherein the bit depth of the image after batch processing is 8, and storing the image to a target folder;
in the step 3, a training set, a verification set and a test set of sample images are generated based on the marked mining conveyor belt deviation detection data set; obtaining 560 samples of a mining conveyor belt deviation detection data set through labelme labeling, wherein the number of samples of a training set and the number of samples of a verification set are set to be 9:1, the number of samples of the training set is 450, the number of samples of the verification set is 50, and the number of samples of a test set is 60;
in the step 4, a semantic segmentation network for detecting the deviation of the mining conveyor belt is constructed; the semantic segmentation network is a multi-scale network PSPNet based on a pyramid pooling model, and the PSPNet is divided into a Resnet convolution module, a pyramid pooling module and an FCN module; the depth of the Resnet convolution module is 50 layers, and a characteristic layer of an input image is extracted; the pyramid pooling module is divided into 4 layers of collection areas; the collection area comprises 4 pooling cores with different sizes respectively; the sizes of the pooling cores are 1 multiplied by 1, 2 multiplied by 2, 3 multiplied by 3 and 6 multiplied by 6 respectively, and pooling feature maps with different sizes are extracted in a mapping way; the output end of the pooling feature map is connected with the input end of a convolution kernel, the size of the convolution kernel is 1 multiplied by 1, and the weight of the global feature is kept; the FCN module outputs a predicted image with the same size as the input image;
in the step 5, selecting hardware equipment and software configuration meeting training requirements, adjusting PSPNet network model parameters, inputting a sample image training set and a verification set into a semantic segmentation network for image data training until a network model reaches a convergence state, and generating a mining conveyor belt semantic segmentation training model; the method comprises the following specific steps:
step 5.1: the selected computer hardware is configured as i9-9900K CPU, memory 32G, RTX2080 GPU, video memory 16G, operating system win10, programming language Python3.8, deep learning frame PyTorch;
step 5.2: respectively adding a manufactured mining conveyor belt deviation detection data set and an original mining conveyor belt sample image into a segmenttaionclass and an Img folder, and executing voc2pspnet.py to realize data reading;
step 5.3: setting the number of target CLASSES to be segmented in pspnet.py and train.py, num_class=3;
step 5.4: setting part of training parameters in train. Py, training round number of epoch hs is 200, training Batch sample size of previous 100 training round number of epoch hs is set to 8, the learning rate is set to 0.01, the training Batch sample size of the last 100 training rounds of epoch hs is set to 4, and the learning rate is set to 0.001;
step 5.5: training the PSPNet model, gradually reducing and gradually flattening the loss function values of the training set and the verification set along with the continuous increase of the number of the training rounds of epoch, and basically converging the training model;
the average cross ratio MIoU in the step 6 has the following calculation formula:
wherein k is a category and does not contain a background category; p is p ij For the number of false positive cases, p ii For the number of true examples, p ji Is the number of false negative cases;
the average pixel accuracy MPA in the step 6 has a calculation formula:
wherein k is a category and does not contain a background category; p is p ij For the number of false positive cases, p ii Is the number of real examples;
the calculation formula of the cross entropy loss function Cross Entropy loss in the step 7 is as follows:
wherein p (x) i ) For the true probability distribution, q (x i ) For predicting probability distribution;
in the step 9, the position relationship between each planned safety area and the conveyer belt area subgraph and the area sizes of the two areas are evaluated, and the specific steps are as follows:
step 9.1: judging the position relation between the belt area and the common area divided in the drawing for a single mining conveyor belt sub-graph, if the belt area is contained in the common area, considering that the mining conveyor belt runs safely at the moment, and if the belt area is not completely contained in the common area, considering that the mining conveyor belt is likely to have deviation faults at the moment;
step 9.2: calculating the area of the common region and the fused area of the two regions respectively;
step 9.3: judging the size relation between the common area and the fusion area, if the ratio of the common area to the fusion area is larger than 0.90, considering that the mining conveyor belt runs safely at the moment, if the ratio of the common area to the fusion area is 0.75-0.90, considering that the mining conveyor belt has slight deviation faults at the moment, and if the ratio of the common area to the fusion area is smaller than 0.75, considering that the mining conveyor belt has serious deviation faults at the moment;
step 9.4: and respectively evaluating other subgraphs according to the judgment basis.
2. The mining conveyor belt deviation detection method based on the semantic segmentation network according to claim 1, wherein the mining conveyor belt deviation detection method is characterized in that: in the step 10, when a deviation fault is detected at a certain moment, the mining conveyor belt state monitoring software platform records the fault information at the moment, and the system immediately and automatically sends an alarm signal to the site; the mining conveyor belt state monitoring software platform realizes the deviation detection of the mining conveyor belt through a monitoring camera at the front end, a transmission communication of an optical fiber network in the middle and a visual detection algorithm at the background; when the deviation fault is detected at a certain moment, the mining conveyer belt state monitoring software platform records the fault information at the moment, and the system immediately and automatically sends an alarm signal to the site;
the mining conveyor belt state monitoring software platform comprises: the system comprises a system login module, a system management module, a video monitoring module, a data management module and an alarm management module;
a system login module: authenticating the identity of the access user; judging whether the user is allowed to enter the system according to the user name and the password by entering a system login interface;
and a system management module: parameter configuration and password management of a mining conveyer belt status monitoring software platform are realized, the parameter setting realizes the system stopping or restarting and system alarming functions, and the password modifying function can realize the self-changing of login passwords by a user;
and the video monitoring module is used for: the method comprises the steps of acquiring acquired videos or images by a site monitoring function, a camera adding function, a monitoring debugging function and a monitoring equipment inquiring function, and realizing storage of target video images and inquiry of equipment names, equipment addresses and port number information;
and a data management module: automatically storing the data information detected by the system according to the fault type, and allowing an access user to perform real-time query, history browsing and data export operation;
alarm management module: alarm management is realized based on the detected running state of the conveyer belt; if the running state of the conveyer belt is abnormal and the system detects a fault event, the system automatically alarms, reminds on-site personnel to arrive at the on-site for processing in time, and records alarm information so as to facilitate subsequent shutdown and overhaul.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110584550.2A CN113283344B (en) | 2021-05-27 | 2021-05-27 | Mining conveyor belt deviation detection method based on semantic segmentation network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110584550.2A CN113283344B (en) | 2021-05-27 | 2021-05-27 | Mining conveyor belt deviation detection method based on semantic segmentation network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113283344A CN113283344A (en) | 2021-08-20 |
| CN113283344B true CN113283344B (en) | 2024-03-12 |
Family
ID=77282016
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110584550.2A Active CN113283344B (en) | 2021-05-27 | 2021-05-27 | Mining conveyor belt deviation detection method based on semantic segmentation network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113283344B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113762283B (en) * | 2021-08-30 | 2024-04-09 | 中铁工程装备集团有限公司 | Conveyor belt deviation monitoring method and device |
| CN113763375A (en) * | 2021-09-17 | 2021-12-07 | 深圳市赛为智能股份有限公司 | Conveyor belt deviation degree detection method and device, computer equipment and storage medium |
| CN113763376B (en) * | 2021-09-17 | 2024-03-01 | 深圳市赛为智能股份有限公司 | Conveyor belt offset detection method, conveyor belt offset detection device, computer equipment and storage medium |
| CN114044325A (en) * | 2021-11-15 | 2022-02-15 | 华能汕头海门发电有限责任公司 | Coal conveying belt fault detection system |
| CN114313883B (en) * | 2022-02-08 | 2024-03-12 | 深圳市铁越电气有限公司 | Automatic detection method and system for belt deviation based on image processing technology |
| CN114155494B (en) * | 2022-02-10 | 2022-05-17 | 力博重工科技股份有限公司 | Belt conveyor deviation monitoring method based on deep learning |
| CN114655655A (en) * | 2022-03-09 | 2022-06-24 | 南京北路软件技术有限公司 | Conveyor belt deviation detection method based on UNet network |
| CN114772208B (en) * | 2022-03-31 | 2023-06-23 | 东北大学 | Non-contact belt tearing detection system and method based on image segmentation |
| CN114937028B (en) * | 2022-06-21 | 2023-12-08 | 苏州上舜精密工业科技有限公司 | Intelligent identification and recognition linear sliding table module quality detection method and system |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002004796A (en) * | 2000-06-23 | 2002-01-09 | Komatsu Ltd | Carrying system in tunnel and load carry out and carry in method |
| CN103359572A (en) * | 2013-05-23 | 2013-10-23 | 中国矿业大学 | Elevator fault diagnosis method and device |
| WO2019126664A1 (en) * | 2017-12-22 | 2019-06-27 | Flexible Steel Lacing Company | Apparatus and method for monitoring conveyor systems |
| CN110163852A (en) * | 2019-05-13 | 2019-08-23 | 北京科技大学 | The real-time sideslip detection method of conveyer belt based on lightweight convolutional neural networks |
| CN110514669A (en) * | 2019-07-31 | 2019-11-29 | 辽宁科技大学 | A kind of belt conveyor multizone spills material detection system and method |
| CN110953984A (en) * | 2019-11-22 | 2020-04-03 | 浙江蓝卓工业互联网信息技术有限公司 | Method and device for judging deviation degree of conveyor belt |
| CN111275711A (en) * | 2020-01-08 | 2020-06-12 | 西安电子科技大学 | Real-time image semantic segmentation method based on lightweight convolutional neural network model |
| CN111369538A (en) * | 2020-03-05 | 2020-07-03 | 北京科技大学 | Conveyor belt deviation detection method based on deep convolutional neural network |
| CN112700419A (en) * | 2020-12-31 | 2021-04-23 | 精英数智科技股份有限公司 | Yield measuring method, device and system based on image segmentation |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8293072B2 (en) * | 2009-01-28 | 2012-10-23 | Georgia-Pacific Consumer Products Lp | Belt-creped, variable local basis weight absorbent sheet prepared with perforated polymeric belt |
-
2021
- 2021-05-27 CN CN202110584550.2A patent/CN113283344B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002004796A (en) * | 2000-06-23 | 2002-01-09 | Komatsu Ltd | Carrying system in tunnel and load carry out and carry in method |
| CN103359572A (en) * | 2013-05-23 | 2013-10-23 | 中国矿业大学 | Elevator fault diagnosis method and device |
| WO2019126664A1 (en) * | 2017-12-22 | 2019-06-27 | Flexible Steel Lacing Company | Apparatus and method for monitoring conveyor systems |
| CN110163852A (en) * | 2019-05-13 | 2019-08-23 | 北京科技大学 | The real-time sideslip detection method of conveyer belt based on lightweight convolutional neural networks |
| CN110514669A (en) * | 2019-07-31 | 2019-11-29 | 辽宁科技大学 | A kind of belt conveyor multizone spills material detection system and method |
| CN110953984A (en) * | 2019-11-22 | 2020-04-03 | 浙江蓝卓工业互联网信息技术有限公司 | Method and device for judging deviation degree of conveyor belt |
| CN111275711A (en) * | 2020-01-08 | 2020-06-12 | 西安电子科技大学 | Real-time image semantic segmentation method based on lightweight convolutional neural network model |
| CN111369538A (en) * | 2020-03-05 | 2020-07-03 | 北京科技大学 | Conveyor belt deviation detection method based on deep convolutional neural network |
| CN112700419A (en) * | 2020-12-31 | 2021-04-23 | 精英数智科技股份有限公司 | Yield measuring method, device and system based on image segmentation |
Non-Patent Citations (4)
| Title |
|---|
| A Computer Vision Based Conveyor Deviation Detection System;Mengchao Zhang 等;《Applied Sciences》;第10卷(第7期);第1-10页 * |
| 基于视觉的矿用输送带状态监测系统研究;胡江迪;《中国优秀硕士学位论文全文数据库 (工程科技Ⅰ辑)》(第03期);第B021-789页 * |
| 基于计算机视觉的矿用输送带异常情况检测方法研究;林俊;《中国优秀硕士学位论文全文数据库 (工程科技Ⅰ辑)》;20210515(第2021年第05期);第B021-170页 * |
| 智慧矿山皮带运输机堵煤视频智能识别算法研究综述;赵学军;《江苏通信》;20210228;第37卷(第01期);第113-115页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113283344A (en) | 2021-08-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113283344B (en) | Mining conveyor belt deviation detection method based on semantic segmentation network | |
| WO2022077605A1 (en) | Wind turbine blade image-based damage detection and localization method | |
| CN108037133B (en) | Intelligent electric power equipment defect identification method and system based on unmanned aerial vehicle inspection image | |
| CN110084165B (en) | Intelligent identification and early warning method for abnormal events in open scene of power field based on edge calculation | |
| Zhang et al. | SOD‐YOLO: a small target defect detection algorithm for wind turbine blades based on improved YOLOv5 | |
| CN109815904B (en) | Fire identification method based on convolutional neural network | |
| CN110516522B (en) | Inspection method and system | |
| CN104535356A (en) | Method and system for monitoring rope arrangement faults of drum steel wire rope on basis of machine vision | |
| CN102673979B (en) | Method and device for judging deviation of conveying belt | |
| CN113343779B (en) | Environment abnormality detection method, device, computer equipment and storage medium | |
| CN116224925B (en) | Intelligent processing management system | |
| CN108664875A (en) | Underground belt-conveying monitoring method based on image recognition | |
| CN113361420A (en) | Mine fire monitoring method, device and equipment based on robot and storage medium | |
| CN110597165A (en) | Steel piling monitoring system and steel piling monitoring method | |
| CN116168019B (en) | Power grid fault detection method and system based on machine vision technology | |
| CN112702570A (en) | Security protection management system based on multi-dimensional behavior recognition | |
| CN115187880A (en) | Communication optical cable defect detection method and system based on image recognition and storage medium | |
| Wang et al. | Offset Detection of Grate Trolley’s Side Plate Based on YOLOv4 | |
| CN117114420B (en) | Image recognition-based industrial and trade safety accident risk management and control system and method | |
| CN104168473B (en) | A kind of detection method of shear wave interference of video image and device | |
| Zhao et al. | Research status and development trend of safety helmet target detection technology | |
| CN106131490A (en) | Intelligent finance system based on intelligent video analysis technology | |
| Gang et al. | Research on key technology of infrared detection of power equipment | |
| CN117893789A (en) | High-precision method for detecting overhead transmission line insulator based on improved Faster R-CNN algorithm | |
| CN112819183A (en) | Algorithm and system for intelligently distinguishing heating defects of power transmission and distribution line |
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 |