CN111432179A - Intelligent coal conveying belt inspection system and method based on computer vision - Google Patents
Intelligent coal conveying belt inspection system and method based on computer vision Download PDFInfo
- Publication number
- CN111432179A CN111432179A CN202010336718.3A CN202010336718A CN111432179A CN 111432179 A CN111432179 A CN 111432179A CN 202010336718 A CN202010336718 A CN 202010336718A CN 111432179 A CN111432179 A CN 111432179A
- Authority
- CN
- China
- Prior art keywords
- belt
- coal
- deviation
- detection
- time
- 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
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- H04N7/181—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a plurality of remote sources
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C1/00—Registering, indicating or recording the time of events or elapsed time, e.g. time-recorders for work people
- G07C1/20—Checking timed patrols, e.g. of watchman
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B19/00—Alarms responsive to two or more different undesired or abnormal conditions, e.g. burglary and fire, abnormal temperature and abnormal rate of flow
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a coal conveying belt intelligent inspection system based on computer vision. The invention further provides a computer vision-based intelligent inspection method for the coal conveying belt. The invention can detect the running state of the belt in real time in the running process of the belt, find abnormal conditions in time and give an alarm. The system provided by the invention has universality, is suitable for various belt transportation scenes, helps operation and maintenance personnel to master actual conditions on site more quickly, reduces the operation of manual timing and aging of the existing detection sensor, prevents further expansion of accidents caused by equipment faults, ensures operation safety, and reduces the waste of manpower and time.
Description
Technical Field
The invention relates to a coal conveying belt inspection alarm system and a method, belonging to the field of computer vision and artificial intelligence.
Background
After the power plant is formally put into operation, in order to ensure the safe production operation of the equipment, a large amount of manpower and material resources are required to inspect the equipment in daily life, wherein a coal conveying system is the most typical system. The coal conveying system has great influence on the production of a thermal power plant, the normal operation of the coal conveying system can ensure the safe and normal operation of the power plant, and once the coal conveying system fails or is abnormal, if the failure cannot be eliminated in a short time, great economic loss is caused.
A manual inspection mode is adopted in a traditional power plant to continuously inspect a coal conveying system, the manual inspection mode is restricted by the responsibility and the technical level of personnel, the condition that equipment is missed to be inspected easily occurs, potential safety hazards exist, and a large amount of labor cost is consumed. Therefore, the electric power system urgently needs to provide an intelligent inspection working mode by means of informatization and automation technical means, automatically inspects running equipment in real time, achieves early discovery and early processing, reduces labor cost, improves inspection working efficiency, and improves identification capability and predictability of running personnel on equipment defects.
In recent years, computer vision and artificial intelligence technologies have been developed rapidly, and are widely applied in the fields of video monitoring, automatic driving, virtual reality, intelligent assembly and the like.
Disclosure of Invention
The purpose of the invention is: based on computer vision algorithm and artificial intelligence technique, build a coal conveyor belt intelligence system of patrolling and examining, can effectively replace artifical the patrolling and examining, reduce patrolling and examining personnel's working strength.
In order to achieve the above object, one technical solution of the present invention is to provide a computer vision-based intelligent patrol system for a coal belt, which is characterized by comprising:
a belt deviation detection camera fixed above the coal conveying belt collects real-time belt deviation detection video images at a fixed visual angle;
a coal blockage detection camera fixed above a coal drop port of the transfer station collects real-time coal blockage detection video images at a fixed visual angle;
abnormal personnel intrusion detection cameras fixed in the early warning perimeter area collect real-time abnormal personnel intrusion detection images at a fixed view angle;
belt off tracking detects the camera, the video stream information that stifled coal detection camera and unusual personnel break into the detection camera and all will gather and obtain sends into high performance processing server, high performance processing server utilizes the video image that belt off tracking detection camera uploaded to carry out the belt off tracking and detects, the video image that utilizes stifled coal detection camera to upload carries out the chute blockage and detects, the video image that utilizes unusual personnel to break into the detection camera and uploads carries out unusual personnel and break into the detection, if high performance processing server detects obtains unusual information, then produce by early warning system and report an emergency and ask for help or increased vigilance, wherein:
when the high-performance processing server carries out belt deviation detection, a belt area contained in each frame image in a belt deviation detection video image is selected, then boundary reference lines under the normal condition of a belt are arranged on two sides of the belt area, deviation warning reference lines are arranged on two sides of the outer side of the boundary reference lines in combination with a deviation threshold value, the boundary reference lines and a preset belt width are used as a tracking target identification standard, the belt edge of the currently obtained belt area is detected in real time, a deviation distance is calculated by utilizing the belt edge and the boundary reference lines, if the deviation distance exceeds the deviation warning reference lines, abnormal information is generated, meanwhile, the high-performance processing server generates a deviation distance change curve by utilizing the deviation distance obtained by real-time detection, the deviation distance change curve is combined with the existing actual condition, the boundary reference lines are adjusted and optimized, to strengthen the range of the deviation warning reference line;
when the high-performance processing server carries out the chute blockage and detects, detect the height that the camera real-time detection kept off the deformation condition of leaf piece and band conveyer coal stream form and coal pile through the chute blockage, when the coal of band conveyer blanking mouth department appears piling up, the coal pile height that detects the camera through the chute blockage highly uprises gradually, and the volume also changes, and the while becomes and keeps off the emergence of leaf piece and violently warp. When the coal pile reaches a preset warning height or the deformation of the variable-grade blade exceeds a preset threshold value and cannot disappear after lasting for a preset time, determining that a coal blockage event occurs and generating abnormal information;
when the high-performance processing server detects abnormal personnel intrusion, after a human body recognition algorithm is trained through a large number of real scene samples, the human body recognition algorithm is used for detecting and tracking personnel entering the early warning perimeter area, detection, analysis and recognition of the human body are achieved, when the fact that the personnel enter the early warning perimeter area in unauthorized time is recognized, warning and alarming are carried out, and when the personnel detected in the continuous preset time length still do not leave the early warning perimeter area, abnormal information is generated.
Preferably, the whole coal conveying belt area can be shot by the belt deviation detection camera, and the shot picture is kept horizontal; the coal blockage detection camera is fixed at a preset position of a blanking port of the belt conveyor.
Preferably, the belt deviation detection camera, the coal blockage detection camera and the abnormal personnel intrusion detection camera all adopt high-definition network cameras.
The invention also provides a computer vision-based intelligent inspection method for the coal conveying belt, which is characterized in that the intelligent inspection system for the coal conveying belt comprises belt deviation detection, coal blockage detection and abnormal personnel intrusion detection, wherein the method comprises the following steps:
the belt deviation detection method comprises the following steps:
step 1, a belt deviation detection camera collects real-time belt deviation detection video images at a fixed visual angle;
step 2, selecting a belt area contained in each frame of image in the belt deviation detection video image, then setting boundary reference lines under the normal condition of the belt on two sides of the belt area, and setting deviation warning reference lines on two sides outside the boundary reference lines by combining deviation threshold values;
step 3, taking the boundary reference line and a preset belt width as a tracking target identification standard, detecting the belt edge of a currently obtained belt area in real time, calculating by using the belt edge and the boundary reference line to obtain a deviation distance, if the deviation distance exceeds a deviation warning reference line, generating abnormal information, generating a deviation distance change curve by using the deviation distance obtained by real-time detection by using a high-performance processing server, combining the deviation distance change curve with a ready-made actual condition, and adjusting and optimizing the boundary reference line to strengthen the range of the deviation warning reference line;
the coal blockage detection method comprises the following steps:
detect the camera real-time detection through the chute blockage and keep off the height that page or leaf piece warp the condition and band conveyer coal stream form and coal were piled up, when the coal of band conveyer blanking mouth department appears piling up, the coal pile height that detects the camera through the chute blockage height and discern becomes gradually high, and the volume also changes, and the violent deformation takes place for the variable fender page or leaf piece simultaneously. When the coal pile reaches a preset warning height or the deformation of the variable-grade blade exceeds a preset threshold value and cannot disappear after lasting for a preset time, determining that a coal blockage event occurs and generating abnormal information;
the abnormal personnel intrusion detection method comprises the following steps:
after a human body recognition algorithm is trained through a large number of real scene samples, detecting, tracking and analyzing the human body by utilizing the human body recognition algorithm to detect and track the personnel entering the early warning perimeter area, and generating abnormal information when recognizing that the personnel enter the early warning perimeter area for warning and alarming in unauthorized time and the personnel detected in the continuous preset time length still do not leave the early warning perimeter area;
when the early warning system receives the abnormal information, the early warning system informs workers of the occurrence of an abnormal event, displays images of monitoring points, prints a time stamp, and gives instructions to an alarm or a belt conveyor control system so as to be capable of taking emergency stop or braking.
Preferably, in the step 2, the belt regions included in the image are selected by a foreground segmentation method, or a large number of training samples with labels are obtained by an image segmentation method, and are used as training data of the Mask RCNN image segmentation model to train the Mask RCNN image segmentation model based on the convolutional neural network, and then the belt regions included in the image are selected by the Mask RCNN image segmentation model.
Preferably, in the step 3, the deviation distance is obtained by proportionally calculating the pixel distances of the belt edge and the deviation warning reference line.
Preferably, in step 3, the high performance processing server generates the abnormal information by using an identification type tracking method, the identification type tracking method tracks and trains a target detector in the belt transportation process, the target detector includes a boundary reference line preset on the belt in advance, a deviation distance and time, when the deviation distance exceeds the deviation warning reference line and the duration time is greater than a predetermined time threshold, the belt is determined to be deviated, the abnormal information is generated, under the condition that the deviation is not deviated, data statistics and analysis are performed on the deviation distance, the deviation threshold is combined, the boundary reference line is optimized, and then a new boundary reference line is used for updating the training set so as to update the target detector.
Preferably, step 3 is followed by:
and 4, when abnormal information is generated, extracting a 100 pix-300 pix image around the anchor point of the specific scale by the high-performance processing server, transmitting an alarm video frame corresponding to the abnormal information to an image classification algorithm for further discrimination, receiving the alarm video frame by the image classification algorithm, analyzing whether the belt deviation range actually exceeds a threshold value or not, if so, generating an alarm by an early warning system, and otherwise, eliminating false alarm.
Preferably, when carrying out the chute blockage and detecting, carry out real-time intelligent comparison analysis to same belt respectively by two at least chute blockage detection cameras, when detecting through a chute blockage detection camera and producing abnormal information, the rethread chute blockage detects the camera and carries out the secondary and confirms, when confirming the correct back, is reported an emergency and asked for help or increased vigilance by early warning system production.
Preferably, before belt deviation detection, coal blockage detection or abnormal personnel intrusion detection, an image enhancement processing mode based on a deep learning method is used for processing the image, so that the image is clearer and is easy to recognize systematically.
The invention can detect the running state of the belt in real time in the running process of the belt, find abnormal conditions in time and give an alarm. The system provided by the invention has universality, is suitable for various belt transportation scenes, helps operation and maintenance personnel to master actual conditions on site more quickly, reduces the operation of manual timing and aging of the existing detection sensor, prevents further expansion of accidents caused by equipment faults, ensures operation safety, and reduces the waste of manpower and time.
Drawings
FIG. 1 is a logic design diagram of the intelligent inspection system of the present invention;
FIG. 2 is a flow chart of the intelligent inspection system of the present invention;
fig. 3 is an architecture diagram of the intelligent inspection system of the present invention.
Detailed Description
The invention will be further illustrated with reference to the following specific examples. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the present invention. Further, it should be understood that various changes or modifications of the present invention may be made by those skilled in the art after reading the teaching of the present invention, and such equivalents may fall within the scope of the present invention as defined in the appended claims.
As shown in fig. 1, the intelligent inspection system provided by the invention comprises a camera front-end acquisition device, a processing server, a computer vision algorithm analysis engine, a cloud motion storage system and a disposal and emergency system. The method comprises the steps that video images of deviation accidents, coal blockage accidents and personnel intrusion events are collected in real time by a camera front-end collection device and then sent to a processing server, the processing server calls a computer vision algorithm analysis engine to analyze and process video streams, if special conditions occur, information is notified to a disposal emergency system, related video original data of the events are sent to a storage system to be stored, and the disposal emergency system can carry out different emergency disposal schemes according to the intensity of the conditions, such as ring alarm, intervention information sending and the like.
The computer vision algorithm analysis engine specifically comprises three subtasks:
(1) belt deviation detection: the deviation condition of the coal conveying belt is identified in time, the belt deviation accident identification efficiency is improved, and the occurrence of an event is judged more accurately by assisting the original deviation signal.
(2) And (3) coal blockage detection: the coal blockage situation of the coal drop port is identified in time, the coal blockage accident identification efficiency of the coal conveying belt is improved, and the occurrence of an event is judged more accurately by an original coal blockage signal.
(3) Abnormal personnel intrusion detection: and identifying the intrusion of personnel in the monitoring range, prompting operation and maintenance personnel and strengthening safety precaution.
The camera front-end acquisition equipment comprises a high-definition network camera fixed above a coal conveying belt (the whole belt area can be shot and the image is as horizontal as possible) and above a transfer station coal breakage port (the appropriate position of a material breakage port of a belt conveyor) so as to acquire real-time video images at a fixed visual angle. The high-definition network camera sends the acquired video stream information to the processing server, and the processing server calls a computer vision algorithm analysis engine therein to analyze the video stream.
The system features mainly include two aspects:
(1) data storage: when the system monitors that deviation, coal blockage and personnel intrusion events exist, the image at the stage is automatically stored; if a large scale specific event is found, the video data within 10 minutes before the occurrence of the situation is automatically saved. (purpose is to accumulate data, identify accumulated data for accident precursor)
(2) Early warning and forecasting: when the occurrence of deviation, coal blockage and personnel intrusion events is found, early warning information (sound, images and the like) is sent out in time, and data is stored according to the two principles in the step (1) for subsequent analysis and research.
As shown in fig. 2, which is a working flow chart of the present invention, the current working states of the belt, the coal distributor, etc. are read from the SIS system to obtain an input interface information list, and a camera list capable of performing intelligent identification is listed according to the corresponding information. And according to the working strategy, the camera adjusts the initial state to a proper preset position, shoots corresponding video information and carries out corresponding intelligent identification.
And obtaining the related information of the existing deviation and coal blockage sensor from the SIS system. The video information is analyzed by using a computer vision technology, and the occurrence of things can be better judged according to deviation, coal blockage and different working conditions and scenes of people intruding. And reminding the operation and maintenance personnel in an alarm mode when the event occurs. And displaying the time-related pictures and video information in a popup window mode.
As shown in fig. 3, the smart identification system is composed of a server and a client. The server is the core of the whole identification system and is used for video real-time identification and event generation. The client is used for interacting with the client, and is used for configuring the server, acquiring the event and the real-time video stream and the display picture corresponding to the event.
The intelligent identification system obtains the working state of the existing coal conveying system through the SIS system. And obtaining an input interface information list through working states of a real-time belt, a coal distributor and the like, and generating a camera list capable of performing intelligent identification. According to the working strategy, the camera adjusts the initial state to a proper preset position, corresponding video information is shot, the intelligent recognition system calls video streaming media information corresponding to the video monitoring platform, a computer vision algorithm analysis engine is used for analyzing and processing the video stream, and corresponding intelligent recognition is carried out. The event information generated by the recognition engine is managed by an event management system. When a specific event occurs, the operation and maintenance personnel are reminded in an alarm mode. And displaying the time-related pictures and video information in a popup window mode.
The server comprises three components of an identification engine, a video stream control system and an event management system. The recognition engine is a machine vision core and is used for recognizing the deviation, coal piling and personnel intrusion events through a convolutional neural network algorithm, and the video stream control module is mainly used for introducing real-time video from a video monitoring platform and accessing the recognition engine, so that corresponding video stream media information can be better used and managed. The event management system is used for managing the event information generated by the recognition engine.
Data flow and logic determination:
(1) equipment state acquisition and linkage judgment logic, which is used for reading in the running state of the equipment from the SIS system, controlling the direction of the cameras in a linkage manner (some equipment share the same camera), and starting or stopping the real-time identification of the equipment;
(2) configuration and decision logic-for configuring the recognition engine and video stream access;
(3) communication logic-for communicating with clients, exchanging information.
And the client is used for system display and management. And managing the system through the windows-based interface for interacting and using with the client. And through video stream control, acquiring the concerned video stream and image in real time according to user input, and displaying the concerned video stream and image on an interface system.
Aiming at belt deviation detection, the belt region in the current video frame is selected by a traditional foreground segmentation method (a Mask RCNN image segmentation model based on a convolutional neural network can also be adopted, and a large number of training samples with labels are obtained by the traditional image segmentation method and are used as training data of the Mask RCNN). And then, setting a boundary reference line and a belt width under the normal condition of the belt at two sides of the belt area as a tracking target identification standard, detecting the edge of the belt in real time by using an image, and calculating the deviation distance according to the equal proportion of the pixel distance of the edge of the belt and the boundary reference line. And prompting and alarming when the deviation distance exceeds a constraint red line range determined by a deviation threshold and a boundary reference line. The deviation threshold value is adjustable, and the set adjustment range is 5-30 cm. And (3) analyzing the deviation distance in real time by using a computer vision algorithm analysis engine, generating a deviation distance change curve, adjusting and optimizing the optimal value of the boundary reference line by combining the existing actual condition, and strengthening the restraint red line range.
The computer vision algorithm analysis engine adopts a discrimination type tracking method, the discrimination type tracking method is to track and train a target detector in the belt transportation process, the target detector comprises a boundary reference line preset on a belt in advance and a tracking target deviation distance and time, when the deviation distance exceeds a deviation threshold value and the duration is greater than a certain range (such as 1s), the deviation of the tracking target is determined, under the condition of no deviation, data statistics and analysis are carried out on the deviation value, the boundary reference line is optimized by combining the deviation threshold value, and then a new boundary reference line is used for updating a training set so as to update the target detector;
when the belt deviation early warning occurs, the server extracts images with the radius of 100 pix-300 pix around the anchor point of the specific scale, transmits alarm pictures to an image classification algorithm for further discrimination, receives alarm video frames by the image classification algorithm, analyzes that the deviation range of the belt actually exceeds a threshold value, alarms if yes, and eliminates false alarms if not. The image classification algorithm may employ KNN, SVM, CNN, or the like.
For coal blockage event detection, the camera is laterally opposite to the conveying belt, and judgment is carried out by detecting the height of the coal pile. The detection precision is adjustable, and the precision of the height profile shape of the coal pile is set to be 3 cm. And a camera is arranged at a proper position of a blanking port of the belt conveyor, and the deformation condition of the leaf blocking sheet, the coal flow form of the belt conveyor and the height of a coal pile are monitored and detected in real time. When coal at the discharging opening of the belt conveyor is piled up, the height of the coal pile recognized by the camera gradually becomes higher, and the volume also changes. Under normal conditions, the blocking sheet is in a vertical state, and when coal is accumulated, the deformation of the blocking sheet is severe. And when the coal pile reaches the preset warning height or the variable baffle plate has large quantity and cannot disappear after lasting for the preset time, determining that the coal blockage event occurs.
In order to reduce the coal blockage false detection situation, a plurality of cameras are used for data verification. The two cameras respectively carry out real-time intelligent comparison analysis on the same belt. When one camera detects abnormality, the other camera performs secondary confirmation, and performs early warning after no error is confirmed.
Due to the fact that the field production operation environment is severe, the camera can seriously shake when collecting video information, and normal identification work cannot be carried out. Before the belt deviation early warning and the coal blockage event detection are carried out, the image is clearer and is easy to identify by a system based on a deep learning method, such as image deblurring, image debouncing and other image enhancement processing methods.
Aiming at the detection of the personnel intrusion event, the personnel are accurately detected and tracked through a human body recognition algorithm, so that the detection, analysis and recognition of the human body are realized, and the personnel intrusion event in the boundary area is early warned in real time. When suspicious personnel enter the monitoring range, the suspicious personnel can be automatically identified, namely, the suspicious personnel are captured and the images at that time are transmitted to a management center, and an alarm signal is output from the management center.
The factory specific area intrusion alarm system adopts an AI algorithm, and can timely and accurately send alarm information to personnel intrusion behaviors occurring in scenes in various application scenes after a large number of real scene samples are trained. By intelligently analyzing and identifying the real-time video image, the functions of full-screen perimeter protection, area-demarcating perimeter protection, tripwire protection (single tripwire: one-way tripwire, two-way tripwire: one-way tripwire, two-way tripwire) and the like of the image can be realized.
And monitoring and identifying whether people exist in the unauthorized area or not in real time, alarming when the camera identifies that people exist in the unauthorized area, and sending alarm information to the monitoring host when the camera does not leave for more than 5s for a preset time, wherein the monitoring host gives an instruction to the alarm or can take emergency response.
When the system provided by the invention detects an abnormal condition, the system sends information to the monitoring host and the alarm to inform workers of the occurrence of an abnormal event, displays images of monitoring points and prints timestamps. The monitoring host may give instructions to an alarm or belt conveyor control system to be able to take an emergency stop or brake.
Claims (10)
1. The utility model provides a coal conveying belt intelligence system of patrolling and examining based on computer vision which characterized in that includes:
a belt deviation detection camera fixed above the coal conveying belt collects real-time belt deviation detection video images at a fixed visual angle;
a coal blockage detection camera fixed above a coal drop port of the transfer station collects real-time coal blockage detection video images at a fixed visual angle;
abnormal personnel intrusion detection cameras fixed in the early warning perimeter area collect real-time abnormal personnel intrusion detection images at a fixed view angle;
belt off tracking detects the camera, the video stream information that stifled coal detection camera and unusual personnel break into the detection camera and all will gather and obtain sends into high performance processing server, high performance processing server utilizes the video image that belt off tracking detection camera uploaded to carry out the belt off tracking and detects, the video image that utilizes stifled coal detection camera to upload carries out the chute blockage and detects, the video image that utilizes unusual personnel to break into the detection camera and uploads carries out unusual personnel and break into the detection, if high performance processing server detects obtains unusual information, then produce by early warning system and report an emergency and ask for help or increased vigilance, wherein:
when the high-performance processing server carries out belt deviation detection, a belt area contained in each frame image in a belt deviation detection video image is selected, then boundary reference lines under the normal condition of a belt are arranged on two sides of the belt area, deviation warning reference lines are arranged on two sides of the outer side of the boundary reference lines in combination with a deviation threshold value, the boundary reference lines and a preset belt width are used as a tracking target identification standard, the belt edge of the currently obtained belt area is detected in real time, a deviation distance is calculated by utilizing the belt edge and the boundary reference lines, if the deviation distance exceeds the deviation warning reference lines, abnormal information is generated, meanwhile, the high-performance processing server generates a deviation distance change curve by utilizing the deviation distance obtained by real-time detection, the deviation distance change curve is combined with the existing actual condition, the boundary reference lines are adjusted and optimized, to strengthen the range of the deviation warning reference line;
when the high-performance processing server carries out the chute blockage and detects, detect the height that the camera real-time detection kept off the deformation condition of leaf piece and band conveyer coal stream form and coal pile through the chute blockage, when the coal of band conveyer blanking mouth department appears piling up, the coal pile height that detects the camera through the chute blockage highly uprises gradually, and the volume also changes, and the while becomes and keeps off the emergence of leaf piece and violently warp. When the coal pile reaches a preset warning height or the deformation of the variable-grade blade exceeds a preset threshold value and cannot disappear after lasting for a preset time, determining that a coal blockage event occurs and generating abnormal information;
when the high-performance processing server detects abnormal personnel intrusion, after a human body recognition algorithm is trained through a large number of real scene samples, the human body recognition algorithm is used for detecting and tracking personnel entering the early warning perimeter area, detection, analysis and recognition of the human body are achieved, when the fact that the personnel enter the early warning perimeter area in unauthorized time is recognized, warning and alarming are carried out, and when the personnel detected in the continuous preset time length still do not leave the early warning perimeter area, abnormal information is generated.
2. The intelligent patrol inspection system for the coal conveying belt based on the computer vision of claim 1, wherein the belt deviation detection camera can shoot the whole area of the coal conveying belt and the shot picture is kept horizontal; the coal blockage detection camera is fixed at a preset position of a blanking port of the belt conveyor.
3. The intelligent patrol inspection system for the coal conveying belt based on the computer vision of claim 1 or 2, wherein the belt deviation detection camera, the coal blockage detection camera and the abnormal personnel intrusion detection camera all adopt high-definition network cameras.
4. A coal-conveying belt intelligent inspection method based on computer vision is characterized in that the coal-conveying belt intelligent inspection system of claim 1 is adopted, and comprises belt deviation detection, coal blockage detection and abnormal personnel intrusion detection, wherein:
the belt deviation detection method comprises the following steps:
step 1, a belt deviation detection camera collects real-time belt deviation detection video images at a fixed visual angle;
step 2, selecting a belt area contained in each frame of image in the belt deviation detection video image, then setting boundary reference lines under the normal condition of the belt on two sides of the belt area, and setting deviation warning reference lines on two sides outside the boundary reference lines by combining deviation threshold values;
step 3, taking the boundary reference line and a preset belt width as a tracking target identification standard, detecting the belt edge of a currently obtained belt area in real time, calculating by using the belt edge and the boundary reference line to obtain a deviation distance, if the deviation distance exceeds a deviation warning reference line, generating abnormal information, generating a deviation distance change curve by using the deviation distance obtained by real-time detection by using a high-performance processing server, combining the deviation distance change curve with a ready-made actual condition, and adjusting and optimizing the boundary reference line to strengthen the range of the deviation warning reference line;
the coal blockage detection method comprises the following steps:
detect the camera real-time detection through the chute blockage and keep off the height that page or leaf piece warp the condition and band conveyer coal stream form and coal were piled up, when the coal of band conveyer blanking mouth department appears piling up, the coal pile height that detects the camera through the chute blockage height and discern becomes gradually high, and the volume also changes, and the violent deformation takes place for the variable fender page or leaf piece simultaneously. When the coal pile reaches a preset warning height or the deformation of the variable-grade blade exceeds a preset threshold value and cannot disappear after lasting for a preset time, determining that a coal blockage event occurs and generating abnormal information;
the abnormal personnel intrusion detection method comprises the following steps:
after a human body recognition algorithm is trained through a large number of real scene samples, detecting, tracking and analyzing the human body by utilizing the human body recognition algorithm to detect and track the personnel entering the early warning perimeter area, and generating abnormal information when recognizing that the personnel enter the early warning perimeter area for warning and alarming in unauthorized time and the personnel detected in the continuous preset time length still do not leave the early warning perimeter area;
when the early warning system receives the abnormal information, the early warning system informs workers of the occurrence of an abnormal event, displays images of monitoring points, prints a time stamp, and gives instructions to an alarm or a belt conveyor control system so as to be capable of taking emergency stop or braking.
5. The computer vision-based intelligent inspection system for the coal-conveying belt as claimed in claim 4, wherein in the step 2, belt regions included in the image are selected by a foreground segmentation method, or a large number of labeled training samples are obtained by an image segmentation method, the labeled training samples are used as training data of the Mask RCNN image segmentation model to train the Mask RCNN image segmentation model based on the convolutional neural network, and then the belt regions included in the image are selected by the Mask RCNN image segmentation model.
6. The intelligent patrol inspection system for the coal conveying belt based on the computer vision as claimed in claim 4, wherein in the step 3, the deviation distance is calculated by the pixel distance of the belt edge and the deviation warning reference line in equal proportion.
7. The computer vision-based intelligent patrol inspection system for the coal conveying belt as claimed in claim 4, wherein in the step 3, the high performance processing server generates the abnormal information by adopting an identification tracking method, the identification tracking method tracks and trains a target detector in the belt transportation process, the target detector comprises a boundary reference line preset on the belt in advance, a running deviation distance and time, when the running deviation distance exceeds the running deviation warning reference line and the duration is greater than a predetermined time threshold, the belt is determined to run off, the abnormal information is generated, data statistics and analysis are carried out on the running deviation distance under the condition of no running deviation, the boundary reference line is optimized by combining the running deviation threshold, and then a new boundary reference line is used for updating the training set so as to update the target detector.
8. The computer vision-based intelligent inspection system for coal conveyor belts according to claim 4, wherein the step 3 is further followed by:
and 4, when abnormal information is generated, extracting a 100 pix-300 pix image around the anchor point of the specific scale by the high-performance processing server, transmitting an alarm video frame corresponding to the abnormal information to an image classification algorithm for further discrimination, receiving the alarm video frame by the image classification algorithm, analyzing whether the belt deviation range actually exceeds a threshold value or not, if so, generating an alarm by an early warning system, and otherwise, eliminating false alarm.
9. The intelligent patrol inspection system for the coal conveying belt based on the computer vision as claimed in claim 4, wherein when the coal blockage detection is performed, at least two coal blockage detection cameras respectively perform real-time intelligent comparison analysis on the same belt, when abnormal information is generated by one coal blockage detection camera, the other coal blockage detection camera performs secondary confirmation, and when no error is confirmed, the early warning system generates an alarm.
10. The intelligent patrol inspection system for the coal conveying belt based on the computer vision as claimed in claim 4, wherein before the belt deviation detection, the coal blockage detection or the abnormal personnel intrusion detection, the image is processed by using an image enhancement processing mode based on a deep learning method, so that the image is clearer and is easy for system identification.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010336718.3A CN111432179A (en) | 2020-04-26 | 2020-04-26 | Intelligent coal conveying belt inspection system and method based on computer vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010336718.3A CN111432179A (en) | 2020-04-26 | 2020-04-26 | Intelligent coal conveying belt inspection system and method based on computer vision |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111432179A true CN111432179A (en) | 2020-07-17 |
Family
ID=71556797
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010336718.3A Pending CN111432179A (en) | 2020-04-26 | 2020-04-26 | Intelligent coal conveying belt inspection system and method based on computer vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111432179A (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112381795A (en) * | 2020-11-16 | 2021-02-19 | 山西三友和智慧信息技术股份有限公司 | Tree ring automatic detection method based on deep learning |
| CN112987035A (en) * | 2021-02-07 | 2021-06-18 | 北京中交创新投资发展有限公司 | Beidou edge computing equipment and method for acquiring inspection facilities based on equipment |
| CN113023293A (en) * | 2021-02-08 | 2021-06-25 | 精锐视觉智能科技（深圳）有限公司 | Inspection method, device, equipment and system for belt conveyor |
| CN113283339A (en) * | 2021-05-25 | 2021-08-20 | 中南大学 | Machine vision-based method for monitoring material shortage abnormity of industrial mineral aggregate conveying belt |
| CN113306989A (en) * | 2021-05-28 | 2021-08-27 | 华能国际电力股份有限公司营口电厂 | Method for timely detecting deviation of belt conveyor of power plant coal conveying system |
| CN113610072A (en) * | 2021-10-11 | 2021-11-05 | 精英数智科技股份有限公司 | Method and system for identifying person crossing belt based on computer vision |
| CN113682763A (en) * | 2021-09-01 | 2021-11-23 | 陕西竹园嘉原矿业有限公司 | Intelligent speed regulation and foreign matter identification system for coal flow transportation |
| CN113723629A (en) * | 2021-08-11 | 2021-11-30 | 国网河北省电力有限公司保定供电分公司 | Machine room operation and maintenance method, electronic equipment and computer readable storage medium |
| CN113790437A (en) * | 2021-08-03 | 2021-12-14 | 大唐三门峡电力有限责任公司 | Artificial intelligent coal conveying system and method |
| CN114021983A (en) * | 2021-11-05 | 2022-02-08 | 宁夏广天夏电子科技有限公司 | Three-dimensional running monitoring method and system |
| CN114140684A (en) * | 2021-11-08 | 2022-03-04 | 深圳江行联加智能科技有限公司 | Method, device and equipment for detecting coal blockage and coal leakage and storage medium |
| CN114275483A (en) * | 2021-12-31 | 2022-04-05 | 无锡物联网创新中心有限公司 | Intelligent online monitoring system of belt conveyor |
| CN114671215A (en) * | 2022-04-02 | 2022-06-28 | 丁国栋 | Application of belt unattended operation based on image recognition and AI algorithm |
| CN114778755A (en) * | 2022-03-18 | 2022-07-22 | 淮北矿业股份有限公司 | Coal quality on-line measuring system based on big data |
| CN115116010A (en) * | 2022-08-29 | 2022-09-27 | 山东千颐科技有限公司 | Belt deviation-preventing visual identification system based on image processing |
| CN115311753A (en) * | 2022-06-21 | 2022-11-08 | 华能新华发电有限责任公司 | Intelligent inspection system for coal conveying gallery |
| CN115755743A (en) * | 2022-10-18 | 2023-03-07 | 华能国际电力股份有限公司上海石洞口第二电厂 | Automatic fuel plough lifting system based on marker binarization filtering |
| CN117726967A (en) * | 2023-11-07 | 2024-03-19 | 中信重工开诚智能装备有限公司 | Intelligent vehicle operation monitoring system and control method for auxiliary shaft of coal mine |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4075799B2 (en) * | 2003-12-26 | 2008-04-16 | Ｊｆｅスチール株式会社 | Method for detecting powder particles |
| CN106408833A (en) * | 2016-11-02 | 2017-02-15 | 北京弘恒科技有限公司 | Perimeter intrusion detection method and system |
| CN110040470A (en) * | 2019-05-21 | 2019-07-23 | 精英数智科技股份有限公司 | A kind of monitoring method of artificial intelligence video identification belt deviation |
| CN110053943A (en) * | 2019-05-21 | 2019-07-26 | 精英数智科技股份有限公司 | A kind of monitoring method of artificial intelligence video identification belt coal piling |
| CN110422589A (en) * | 2019-08-02 | 2019-11-08 | 精英数智科技股份有限公司 | A kind of coal mine elevating conveyor coal piling safety alarming method, device, equipment, medium |
| CN110490995A (en) * | 2019-08-26 | 2019-11-22 | 精英数智科技股份有限公司 | A kind of belt operating status method for monitoring abnormality, system, equipment and storage medium |
| CN111003446A (en) * | 2019-12-10 | 2020-04-14 | 安徽银河物联通信技术有限公司 | Belt deviation detection method |
-
2020
- 2020-04-26 CN CN202010336718.3A patent/CN111432179A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4075799B2 (en) * | 2003-12-26 | 2008-04-16 | Ｊｆｅスチール株式会社 | Method for detecting powder particles |
| CN106408833A (en) * | 2016-11-02 | 2017-02-15 | 北京弘恒科技有限公司 | Perimeter intrusion detection method and system |
| CN110040470A (en) * | 2019-05-21 | 2019-07-23 | 精英数智科技股份有限公司 | A kind of monitoring method of artificial intelligence video identification belt deviation |
| CN110053943A (en) * | 2019-05-21 | 2019-07-26 | 精英数智科技股份有限公司 | A kind of monitoring method of artificial intelligence video identification belt coal piling |
| CN110422589A (en) * | 2019-08-02 | 2019-11-08 | 精英数智科技股份有限公司 | A kind of coal mine elevating conveyor coal piling safety alarming method, device, equipment, medium |
| CN110490995A (en) * | 2019-08-26 | 2019-11-22 | 精英数智科技股份有限公司 | A kind of belt operating status method for monitoring abnormality, system, equipment and storage medium |
| CN111003446A (en) * | 2019-12-10 | 2020-04-14 | 安徽银河物联通信技术有限公司 | Belt deviation detection method |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112381795A (en) * | 2020-11-16 | 2021-02-19 | 山西三友和智慧信息技术股份有限公司 | Tree ring automatic detection method based on deep learning |
| CN112987035B (en) * | 2021-02-07 | 2023-06-13 | 交信北斗科技有限公司 | Beidou edge computing equipment and method for acquiring inspection facilities based on equipment |
| CN112987035A (en) * | 2021-02-07 | 2021-06-18 | 北京中交创新投资发展有限公司 | Beidou edge computing equipment and method for acquiring inspection facilities based on equipment |
| CN113023293A (en) * | 2021-02-08 | 2021-06-25 | 精锐视觉智能科技（深圳）有限公司 | Inspection method, device, equipment and system for belt conveyor |
| CN113283339A (en) * | 2021-05-25 | 2021-08-20 | 中南大学 | Machine vision-based method for monitoring material shortage abnormity of industrial mineral aggregate conveying belt |
| CN113306989A (en) * | 2021-05-28 | 2021-08-27 | 华能国际电力股份有限公司营口电厂 | Method for timely detecting deviation of belt conveyor of power plant coal conveying system |
| CN113790437A (en) * | 2021-08-03 | 2021-12-14 | 大唐三门峡电力有限责任公司 | Artificial intelligent coal conveying system and method |
| CN113723629A (en) * | 2021-08-11 | 2021-11-30 | 国网河北省电力有限公司保定供电分公司 | Machine room operation and maintenance method, electronic equipment and computer readable storage medium |
| CN113682763A (en) * | 2021-09-01 | 2021-11-23 | 陕西竹园嘉原矿业有限公司 | Intelligent speed regulation and foreign matter identification system for coal flow transportation |
| CN113610072A (en) * | 2021-10-11 | 2021-11-05 | 精英数智科技股份有限公司 | Method and system for identifying person crossing belt based on computer vision |
| CN113610072B (en) * | 2021-10-11 | 2022-01-25 | 精英数智科技股份有限公司 | Method and system for identifying person crossing belt based on computer vision |
| CN114021983A (en) * | 2021-11-05 | 2022-02-08 | 宁夏广天夏电子科技有限公司 | Three-dimensional running monitoring method and system |
| CN114140684A (en) * | 2021-11-08 | 2022-03-04 | 深圳江行联加智能科技有限公司 | Method, device and equipment for detecting coal blockage and coal leakage and storage medium |
| CN114275483A (en) * | 2021-12-31 | 2022-04-05 | 无锡物联网创新中心有限公司 | Intelligent online monitoring system of belt conveyor |
| CN114275483B (en) * | 2021-12-31 | 2023-12-19 | 无锡物联网创新中心有限公司 | Intelligent online monitoring system of belt conveyor |
| CN114778755A (en) * | 2022-03-18 | 2022-07-22 | 淮北矿业股份有限公司 | Coal quality on-line measuring system based on big data |
| CN114778755B (en) * | 2022-03-18 | 2023-07-25 | 淮北矿业股份有限公司 | Coal quality on-line measuring system based on big data |
| CN114671215A (en) * | 2022-04-02 | 2022-06-28 | 丁国栋 | Application of belt unattended operation based on image recognition and AI algorithm |
| CN115311753A (en) * | 2022-06-21 | 2022-11-08 | 华能新华发电有限责任公司 | Intelligent inspection system for coal conveying gallery |
| CN115116010A (en) * | 2022-08-29 | 2022-09-27 | 山东千颐科技有限公司 | Belt deviation-preventing visual identification system based on image processing |
| CN115755743A (en) * | 2022-10-18 | 2023-03-07 | 华能国际电力股份有限公司上海石洞口第二电厂 | Automatic fuel plough lifting system based on marker binarization filtering |
| CN117726967A (en) * | 2023-11-07 | 2024-03-19 | 中信重工开诚智能装备有限公司 | Intelligent vehicle operation monitoring system and control method for auxiliary shaft of coal mine |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111432179A (en) | Intelligent coal conveying belt inspection system and method based on computer vision | |
| CN111582016A (en) | Intelligent maintenance-free power grid monitoring method and system based on cloud edge collaborative deep learning | |
| CN104751603B (en) | Crag avalanche monitoring and pre-warning system and method | |
| CN105208343B (en) | Can be used for intelligent monitor system and the method for video monitoring equipment | |
| CN103456024B (en) | A kind of moving target gets over line determination methods | |
| CN102665071A (en) | Intelligent processing and search method for social security video monitoring images | |
| CN105141885A (en) | Method for video monitoring and device | |
| CN102081844A (en) | Traffic video behavior analyzing and alarming server | |
| CN102164270A (en) | Intelligent video monitoring method and system capable of exploring abnormal events | |
| CN106454330B (en) | The fuzzy method for detecting abnormality of vision signal | |
| CN109867186B (en) | Elevator trapping detection method and system based on intelligent video analysis technology | |
| CN112112629A (en) | Safety business management system and method in drilling operation process | |
| CN112282819B (en) | Comprehensive mining working face personnel target safety monitoring method and system based on vision | |
| CN103096121A (en) | Camera moving detecting method and device | |
| CN113012374A (en) | Intelligent park security system and control method thereof | |
| CN115527340A (en) | Intelligent construction site safety monitoring system and method based on unmanned aerial vehicle and surveillance camera | |
| CN109488383B (en) | Intelligent video analysis system and application method thereof | |
| CN113762171A (en) | Method and device for monitoring safety of railway construction site | |
| CN117319609A (en) | Internet of things big data intelligent video monitoring system and method | |
| CN210666820U (en) | Pedestrian abnormal behavior detection system based on DSP edge calculation | |
| CN115641608A (en) | Intelligent identification equipment for personal protection articles for distribution network operation based on edge intelligence | |
| CN115953719A (en) | Multi-target recognition computer image processing system | |
| CN117197713A (en) | Extraction method based on digital video monitoring system | |
| CN201830388U (en) | Video content collecting and processing device | |
| CN110809137A (en) | Campus safety trampling prevention monitoring system and method |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200717 |
|
| RJ01 | Rejection of invention patent application after publication |